You are on page 1of 22

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 1

Blockchain Meets Cloud Computing: A Survey


Keke Gai, Senior Member, IEEE, Jinnan Guo, Liehuang Zhu, Member, IEEE, Shui Yu, Senior Member, IEEE

Abstract—Blockchain technology has been deemed to be an Despite various merits of blockchain technology, there are
ideal choice for strengthening existing computing systems in var- two typical challenges commonly exist in current blockchain-
ied manners. As one of the network-enabled technologies, cloud enabled cloud solutions, deriving from findings of our study.
computing has been broadly adopted in the industry through nu-
merous cloud service models. Fusing blockchain technology with The first type of challenge is that blockchain generally
existing cloud systems has a great potential in both functionality/ encounters technical difficulties when applying it in cloud
performance enhancement and security/ privacy improvement. applications. Most difficulties are caused by technical char-
The question remains on how blockchain technology inserts into acteristics of blockchain, some of which are considered ad-
current deployed cloud solutions and enables the reengineering of vantages. For instance, a pure decentralization setting (e.g.
cloud datacenter. This survey addresses this issue and investigates
recent efforts in the technical fusion of blockchain and clouds. public blockchain) offers a strong autonomous working mode;
Three technical dimensions roughly are covered in this work. however, lack of control in this mode also is considered
First, we concern the service model and review an emerging a weakness in many practical scenes. Centralization-based
cloud-relevant blockchain service model, Blockchain-as-a-Service governance cannot be fully abandoned due to various reasons,
(BaaS); second, security is considered a key technical dimension such as legal issues or governmental duties. Our recent study
in this work and both access control and searchable encryption
schemes are assessed; finally, we examine the performance of [13] also finds that privacy leakage is threatening the consor-
cloud datacenter with supports/ participance of blockchain from tium blockchain-based autonomous trading system, based on a
hardware and software perspectives. Main findings of this survey real-world case, as data stored in blocks are open to the public.
will be theoretical supports for future reference of blockchain- Considering cloud datacenter, even though consortium/private
enabled reengineering of cloud datacenter. blockchain lowers down the impact of the decentralization,
Index Terms—Blockchain, cloud computing, data provenance, tamper-resistance is an obstacle to achieving controllable/
blockchain-as-a-service, blockchain service model scalable cloud systems [14], [15].
The other type of common challenge is formulating/ con-
I. I NTRODUCTION ducting blockchain service models. Blockchain used to be
As an emerging technical term, blockchain is deemed to be known as a synonym of “Bitcoin" when the cryptocurrency
an adoptable alternative for establishing a trustful platform, was first introduced to the public, although blockchain tech-
due to a few of its characteristics, e.g. data traceability and nology was created as a distributed ledger-based storage
tamper-resistance. It has been widely believed that blockchain method, a few years before Bitcoin was born. The success of
can be not only used in financial services (e.g. Bitcoin) but blockchain in cryptocurrency leads a tide of blockchain-based
also implemented in application-oriented scenarios [1], [2]. digital currency or financial services, but the success has been
Smart contract is a major driven factor in blockchain-enabled rarely copied in other industries [14], [16]. Lack of effective
applications as it introduces the ability of the automatic service models is a critical element that restricts blockchain
control [3]. The trustful environment conducted by blockchain implementations, albeit some attempts have been made over
is tightly tied to operations or processes via implementing years [17]–[19]. In the scenario of cloud datacenter, it also
smart contract. Implementing blockchain-enabled solutions is struggles to finding out a seamless way of using blockchain
an expectable technical route for reinforcing cloud datacenter. technology. Our research [15] also argues that a pure decentral-
Based on our investigations, we find that many contempo- ization computing is conflict with many existing management
rary studies are seeking methods that utilize blockchain tech- models in which a centralized administration has been long-
niques to power up existing systems. Reengineering of cloud established. The consequence of our research derives from the
datacenter through blockchain-enabled approach is deemed to investigation on both government and enterprise.
be one of the major trends in achieving trustworthiness and As a matter of fact, conquering two challenges above as
reliability in the intercrossed networking environment [4], [5]. well as other issues in the technical fusion is more complicated
A few representative benefits of blockchain technology, as than it seems to be. Simply applying blockchain technology in
widely accepted, include tamper-resistance [6], [7], transparent cloud-based solutions generally is inflexible owing to multiple
governance [8], decentralization-powered security [9], [10] restrictions, such as the system compatibility, blockchain-
and novel business models [11], [12]. cloud interface, governability demand, or infrastructure de-
ployment. Cognizing connections between blockchain and
K. Gai (first author), and L. Zhu are with the School of Computer Science cloud systems is a crucial fundamental to powering up the
and Technology, Beijing Institute of Technology, Beijing, China, 100081,
{gaikeke@bit.edu.cn, liehuangz@bit.edu.cn}. blockchain-cloud combination. The significance of accom-
J. Guo is with School of Information and Electronics, Beijing Institute of plishing a survey in this field covers at least three sides,
Technology, Beijing, China, 1120162383@bit.edu.cn. aligning with three questions given in the following.
S. Yu is with School of Computer Science, University of Technology
Sydney, Australia, Shui.Yu@uts.edu.au. 1) There is an urgent demand for facilitating blockchain
L. Zhu is the corresponding author (liehuangz@bit.edu.cn). technology in on-ground application scenarios. As a

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 2

widely deployed technology, cloud computing is a


proper objective for carrying out blockchain tech-
niques. However, the question remains on how to make
blockchain and clouds compatible. For example, cloud
computing generally highly relies on its centralization
computing, which is contradictive with blockchain’s
decentralization setting.
2) Even though blockchain provides/ enables new service
models, e.g. adding trustworthy values, which seems
to be consistent with cloud service models, offering
blockchain services is more complicated than regular
cloud services. There are many unsolved problems Fig. 1. Technical dimensions of this survey.
needed to be answered. Unlike infrastructure or software
that fits in On-demand Pay (ODP) manner, blockchain
technology is struggling with data exchanging/ sharing
for developing on-ground blockchain-powered cloud systems
when multiple blockchain networks are merged. Finding
or cloud-powered blockchain systems.
out effective service model that supports both clouds and
blockchain has an extreme demand. Fig. 1 illustrates main technical dimensions addressed by
3) It appears to be easier for cloud computing to act as this survey. As the organizational structure shown in the figure,
a technical support for blockchain; nevertheless, more the organization of this work follows the order below.
explorations also are intensely required in this aspect, In Section II, we present an emerging blockchain-oriented
such as blockchain-purpose infrastructure, optimum of- service model, known as a Blockchain-as-a-Service (BaaS).
floading strategies, multi-participant working mode, and From the perspective of data governance, blockchain-enabled
blockchain-oriented storage mechanism. Understanding data provenance and access control methods are reviewed
contemporary achievements in blockchain-purpose cloud in Section III and IV, respectively. Moreover, we synthesize
offerings has a great requirement. work in blockchain-enabled searchable encryption methods
and blockchain-based applications in data deduplication in
We notice that fusing blockchain and cloud techniques has
Sections V and VI. Next, Section VII describes a few recent
rarely addressed by prior survey studies, even though some
studies that focus on applying smart contract in cloud resource
recent studies surveyed blockchain techniques in different per-
allocations. In addition, Section VIII reviews offloading mech-
spectives, e.g., digital currencies [20]–[22], security [23], [24],
anism in blockchain. Section IX shows recent achievements in
privacy [21], [25], edge-integration [26], IoT-integration [27],
performance enhancement from the hardware dimension and
[28], and smart city-integration [29]. Therefore, this survey
Section X exhibits relevant work in the field of blockchain-
focuses on the technical fusion of blockchain and cloud com-
related storage. Furthermore, discussions and main findings
puting. Recent studies addressing the relevant field have been
are presented in Section XI. Finally, we wrap up the work
synthesized in this work. A few aspects, from the perspective
with a conclusion in Section XII.
cloud offerings/ functionality, covered by this survey include
data provenance, access control, searchable encryption, data
deduplication, auto control and resource allocation, hardware II. B LOCKCHAIN - AS - A -S ERVICE
enhancement, and data storage.
A. Concept of BaaS
The purpose of conducting this survey aims to indicate
recent studies of blockchain that can be utilized for powering BaaS is a type of blockchain service model that borrows
up cloud systems or those novel mechanism that uses cloud- the concept from cloud computing. This service model deems
based methods to reinforce blockchain systems. Key applica- blockchain systems or components to be computing resource
tion specification of blockchain is addressed at each technical that can be used for supporting cloud systems or other appli-
dimensions, such as efficiency (throughput capacity), com- cations [30]. The major intention of using BaaS is allowing
patibility (database, network, etc.), energy cost, blockchain customers to focus on core business rather than struggling with
platforms, and security. technical obstacles of blockchain.
Main contributions of this work are twofold. In work [5], a metaphor, “Cloud over Blockchain", is given
First, this survey provides a comprehensive review that to describe a blockchain service offering in a cloud service
concentrates on reengineering of cloud computing from the model. As widely accepted, the continuous growth of cloud
perspective of blockchain implementations. Comparisons in demands has introduced numerous service models. Besides
the same technical dimension are given in this work, which three fundamental cloud service models (IaaS, PaaS, SaaS),
provides an explicit research/ practical guideline for scholars/ emerging cloud services are transmitting even partial process-
practitioners. Second, this survey offers a brief tutorial about ing components or processes into the transferrable manner
fusing blockchain technology with cloud computing. The orga- for service demanders, e.g. Backend-as-a-Service, Process-as-
nization structure of this work depends on a series of technical a-Service, and Security-as-a-Service. The work [31] further
dimensions, which assists academic/ industrial involvers to discussed the service interoperations in IoT through the im-
build up their knowledge scaffold in the field and is meaningful plementation of BaaS, which argued that trusted environment

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 3

and smart contract executions could be used for enhancing TABLE I


interoperability of services. C OMPARISONS OF BAA S I NDUSTRIAL D EPLOYMENTS
The diversity of the cloud service model implies that either a Platform Multi-chain Blockchain Type Security
system or a governable network can be a service content. Sim- AWS Eth, HF No Permissioned IAM
Azure Eth, HF Yes Permissioned Active Directory
ilarly, blockchain infrastructure or backend can be the service Google Eth, HF No Permissioned GCS
offering in a BaaS. To be specified, BaaS allows customers to Oracle HF No Permissioned, Consortium IF
SAP HF No Permissionless Servie Key
leverage a cloud-based approach to obtain blockchain-related IBM HF No Permissioned IBM SSC
supports. For instance, Alibaba Cloud BaaS provides a variety Notations: Eth.: Ethereum; HF: Hyperledger Fabric; IAM: Identity and
of services via offering customers blockchain systems, such as Access Management; GCS: Google Cloud Spanner; IF: Identity Federa-
tion; SSC: Secured Services Containers
transaction tracking database, smart contract, and consortium
governance. With various BaaS providers, the purpose of BaaS
is distinct. Some common target functions include security, B. BaaS Industrial Deployment
cost saving, system integration, and control optimization.
In this survey, we mainly investigate the method for imple- From the perspective of the service presentation, contempo-
menting blockchain techniques in the position of supporting rary BaaS products were similar to BPaaS (Business Process
cloud services. The basic idea of BaaS is that the blockchain as a Service), both of which emphasized the interconnections
network/application is treated as a service offering, on which between logical business activities and physical deliveries.
users are allowed to configure blockchain settings, such as Emerging trends of blockchain had attracted interests of CSPs.
blockchain network types and smart contract rules. Infras- Many IT companies, e.g., Microsoft, IBM and Amazon, pro-
tructure for establishing blockchain network is offered by the vide BaaS in their mature cloud environments. IBM BaaS
service provider and partial codes of blockchain are available is attempting to provide services for vehicular systems [39];
for open source. Exploring the establishment of novel BaaS Oracle BaaS is driving away at service offerings in logistic and
has been addressed by recent studies, e.g., FSBaaS [32], payment [40]. A comparison of BaaS services will be given
uBaaS [33], and NutBaaS [34]. We notice that unified BaaS in this section.
still is under exploration and most prior attempts are at the Microsoft Azure [41] is a cloud platform that provides the
stage of system design. The challenging part is that technical fast blockchain deployment, which supports Ethereum, Corda
difficulties still exist in communication, consensus, and data and Hyperledger Fabric for the deployment and configuration
synchronization. Technical restrictions cause the lack of real- of a blockchain network. The Azure’s user only needs con-
world implementations in unified BaaS. figure certain parameters rather than figuring out all technical
Besides classic features of cloud computing, two specific details. Besides, Microsoft’s solution can automatically backup
characteristics of BaaS concept are highlighted in the follow- the on-chain data to the off-chain cloud storage. Current ver-
ing for the further explicit explanation. sion of Azure mainly supports the single-node configuration in
1) Cloud Service Providers (CSPs) manage/ govern all Fabric and the deployment of consortium blockchain is under
required blockchain computing resource (e.g. infras- explored.
tructure or operations) and provide customers with an Next, another common cloud platform, Amazon Web Ser-
agile service offering, so that customers can obtain vices (AWS) [42], has provided BaaS in their mature and
customized acquisitions to host their blockchain apps wildly-used cloud environment since 2016. AWS’s BaaS can
or partial blockchain functions (e.g. smart contract). support both Ethereum and Hyperledger such that service
2) BaaS screens the complexity of the blockchain imple- alternatives from both blockchain systems are available to
mentation. An ODP manner offers adaptable and scal- users.
able service scope, from establishment and configuration IBM Blockchain Platform [43] provided a public cloud that
to manipulation and maintenance. Plug-in architecture users could deploy the blockchain on it. Compared with other
generally is available for quick adoptions. BaaS, the main disadvantage of the IBM Blockchain Platform
Moreover, applying blockchain techniques is believed to is that it can only support the solution template implemented
be an approach for eliminating drawbacks or creating new by the Hyperledger Fabric, the widely-used Ethereum is
values in clouds. For example, different CSPs are involved not supported in IBM BaaS solution. However, other BaaS
in the integrated cloud data management system so that data providers failed to establish a life-cycle management of users’
sharing/transferring activities are a frequent phenomenon [35], blockchain. IBM provides data life-cycle management which
[36]. Lack of control on data flows has become a concern when can ensure a reliable outsourced data management for users.
multiple parties participate in data sharing actions. The issue Furthermore, IBM BaaS relied on secure containers. It can also
can be solved by deploying blockchain-based data-tracing support users to configure their Blockchain on private clouds
system. or on-premises environments. These above characteristics offer
In general, CSPs of BaaS are responsible for constructing IBM BaaS a secure and reliable cloud environment. Table I
and maintaining the blockchain platform, including perfor- illustrates a brief comparison among popular BaaS services.
mance enhancement (e.g. source scheduling and API design) Despite many merits of BaaS, most existing BaaS is at-
and risk mitigation (e.g. security protection [37]). BaaS pro- tached to single cloud ecosystem due to the restriction of
vides an opportunity for users to maintain ubiquitous access the blockchain system. Multi-chain technique still is under
to blockchain networks [38]. exploration yet, such that multi-cloud setting needs more

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 4

research for deployment. Improvements based on existing BaaS models also had been
made. For example, Chen et al. [49] proposed a Functional
C. BaaS under Exploration BaaS (FBaaS) model such that the BaaS was expanded to the
An efficient and secure blockchain service is a common server-less architecture. The expansion derived from the Big
goal for BaaS products. Obtaining a flexible host service from Data Open Architecture (BDOA) and the model consisted of
clouds becomes an alternative option. Samaniego et al. [30], four layers, including infrastructure, component, service, and
[44] discussed and analyzed the performance difference of business logic layers. Considering the software development
BaaS system running in both cloud and fog environments. The aids, Lu et al. [50] proposed a BaaS-based development assist
results showed that, in the cloud context, a BaaS system could toolkits to support blockchain design pattern services, which
have higher-level computation capability and storage resource covered both data management and smart contract designs.
than that of fog computing, while latency time was longer. In addition, smart contract-oriented service also is a research
Samaniego et al.’s work [44] also evaluated communication direction in the field of BaaS. Zhang et al. [51] attempted a
costs between the BaaS server and IoT devices with varying smart contract-based secure billing approach in ride hailing
amounts of clients and network conditions and demonstrated services. Even though the study did not provide a route of
that a fog-based BaaS system had a better performance than X-as-a-service, it still showed a potential blue map of smart
a cloud-based setting. contract-as-a-service as the rule defined by smart contract
In most current blockchain systems, an assumption was could be embedded into other systems. Smart contract could
made that the demand of the trustful third-party was reduced play a role of enabler for other service offerings. For example,
due to the decentralization setting. Interactions between stake- it [52] has been proved that Mobility-as-a-Service (MaaS)
holders were assumed to be secure no matter whether the could be achieved by the smart contract-enabled trusted trans-
stakeholder was trustful. This assumption could be challenged actions.
when the implementation of BaaS was implemented, according Finally, consensus was a vital part of blockchain for eglitar-
to recent investigations. Singh et al. [38] discussed trust ianism. Marandi et al. [53] discussed a potential mechanism of
management issues in BaaS and pointed out that inserting consensus-as-a-service on a Service Level Agreement (SLA).
service providers into the blockchain system could cause The method combined the agreement with the consensus in
recentralizations. One of the reasons was that the service order to meet the target throughput and degree of fault-
provider could be or had connections with stakeholder(s) so tolerance. In the cloud/fog context, a PoW consensus approach
that the blockchain offering might lack of trust. There was could be an option to achieving optimizations of computing
a great chance that the majority of voters were threatening resources. Kumar et al. [54] investigated the feasibility of
as blockchain system was outsourced to the third party. A using maximization-factorization statistics with a PoW con-
potential solution was signing a service agreement to restrict sensus to reduce time and energy costs, as the proposed
activities of CSP(s). statistical method could realize precise probability in minimal
From the perspective of the service provider, trust concerns time. Zhou et al. [55] paid attention to the Vehicle-to-Grid
also restricted the improvement of BaaS service [45] [38]. (V2G) scenario and utilized consortium blockchain trading
Typically, service providers needed to prove the capability mechanism to enable a low-cost demand-supply matching
of data security by offering transparent operations on the method. The information asymmetry issue was addressed by
distributed ledger. Singh et al. [38] presented four optional this work, which was solved by a contract-based control
solutions: (i) enhancing users’ controllability by using PaaS- method as well as a consensus.
alike setting, (ii) mitigating the recentralization by establishing Summary: In this section, we had discussed BaaS in both
CSP federations, (iii) working on an authenticated trustful industry and research. With the support of BaaS offerings,
environment (e.g. ARM’s trust zone), and (iv) strengthening users can concentrate on the functionality and usability of their
access controls. blockchain-based apps rather than exploring blockchain net-
A few more views on BaaS models have been explored. work establishment. From the perspective of cost-saving, BaaS
Melo et al. [46] assessed reliability and availability of BaaS is efficient due to the easy-configuration and outsourcing-
model, which presented Dynamical Reliability Block Diagrams maintenance. A few tech giants had been developing various
(DRBDs) to construct the “master" and “slave" in hyperledger. BaaS service models as the branch of their cloud services.
The investigation showed that cloud backend offering resulted Our investigations depicted that the industry had ambitious
in a higher-level availability and reliability of BaaS system. attempts on exploring BaaS, while limited research achieve-
Similarly, Lee et al. [47] attempted to using BaaS in identity ments were revealed. A wider implementation of BaaS needed
management services. A Blockchain-based-ID as a Service to take over challenges in trust management, data security, and
(BIDaaS) was proposed in order to eliminate the demand of recentralizations.
screened third party in identity management. This approach
offered a virtual ID’s registration service by publishing related III. B LOCKCHAIN - ENABLED DATA P ROVENANCE IN
transactions with virtual ID and ID’s signature information. C LOUDS
When the user was accessing the BIDaaS service, a mutual
authentication attaching to BaaS ledger was applied for ver- A. Data Provenance Issues
ifying identities. Xu et al. [48] investigated the feasibility of IDC (International Data Corporation) [56] forecasted that
data auditing by applying BaaS. global data sphere would reach 175 Zettabyte by 2025, of

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 5

which half of the cloud data would be stored in public


clouds. Data provenance is an essential of traceable data
usage in assisting the management of such high volume data,
considering both efficiency and reliability. Provenance refers to
a type of metadata that records and describes operation data. In
the scenario of cloud computing, a functional provenance tells
when, where and how data are stored, accessed, modified and
deleted in cloud datacenter, which means CSPs are expected
to provide reliable cloud-data management when a competent
provenance is implemented.
The application of provenance in cloud benefits both CSPs
and users. To address providers’ demands, provenance meta- Fig. 2. High level architecture of blockchain-enabled data provenance for
data could be utilized to debug [66] for discovering potential cloud datacenter.
security vulnerabilities. Titian [67] provided a data provenance
to support Data-Intensive Scalable Computing (DISC) systems
in order to simplify the debug process. Provenance system based provenance architecture, programs named hooks were
can also assist CSPs to identify abnormal process in clouds, responsible for monitoring changes in cloud environment to
e.g. unexpected running applications, which continually con- record these operation events. Besides, provenance data were
suming resources without notice when provenance was not sensitive and vulnerable to data misusing [74]. In ProvChain
applied [68]. The automatous data life-cycle management structure, user’s identities appeared in the hashed form in order
is the major subject offered by provenance. Westerkamp et to protect users’ privacy. Only service provider could map the
al. [57] proposed a blockchain-based approach for tracing hashed value to user’s identity. However, some sensitive data
manufacturing processes that covered both products deliveries were still stored as a plaintext in the blockchain.
and components during manufacturing. The concept of the SmartProvenance [59] system, introduced automatic verifi-
token was used for linking blocks with products or product cation in data provenance process. Differ from verifications in
components. ProvChain that relied on auditors for verification, SmartProve-
On the other hand, from the perspective of users, provenance nance utilized a voting mechanism to design a peer-to-peer de-
can protect users data from the threat of malicious insiders centralized verification scheme; hence, the trusted auditor was
[69], e.g. adversarial data mining. Provenance provides a no longer needed in SmartProvenance. Both provenance col-
platform with a function of recording both administrative and lections and verifications were implemented by smart contract
malicious operations. A long-run of fulfillment of Service to make the whole system fully automatic. SmartProvenance
Level Agreement (SLA) can be monitored under the restriction stored all sensitive data off-chain but remained hash values
of provenance information [70]. Recent study [71] also proved on-chain for privacy-preserving purpose. GridMonitoring [61]
that provenance could be used as a recovery tool in case of also used blockchain technology to generate temper-resistant
mis-detection. provenance data record of power usage data that stored in
Benefits of provenance as discussed above are based on the cloud. Thus, inconsistency in actual power usage between
assumption of metadata that were secure and reliable. How- users and power providers could be tackled with the aid
ever, provenance records still had a chance to be tempered by of smart contract, which provided a transparency and non-
the threat agent, which could disable/ misused the provenance reputable data record.
system [72]. As mentioned in [58], provenance services were Layered nature of cloud indicated that involvement of
subjected to accidentally shut-down and malicious attacks. It federated service providers in service delivery process was
suggested that storage and analysis process be required to inevitable [5]. Resource and task offloading occurred fre-
realize reliable provenance collections. quently between different CSPs. In federated cloud, ProvChain
and SmartProvenance were not appropriate because they were
B. Blockchain-enabled Cloud Data Provenance designed for a single provider situation. Applying [58], [59],
As a temper-resistant distributed ledger, blockchain could [61] in federated cloud environment was challenging due to
ensure the security of provenance data. The basic idea of the lack of interoperability.
blockchain-based data provenance is using blockchain’s char- To address this challenge, Xia’s [60] team launched MeD-
acteristic in traceability to record each activities happened to Share to achieve data provenance and auditing in the federated
data on blocks. Fig. 2 shows the high level architecture of cloud environment. In this work, provenance function was
the blockchain-enabled data provenance for cloud datacenter. designed to support secure data sharing among trust-less
Saquib et al. [73] emphasized that blockchain techniques service providers and prevent malicious attacks that dam-
contribute to data provenance by its immutable, deterministic aged in both financial and reputation aspects. Data owners
and public nature. Smart contract plays a vital role in balancing hosted a complete control of data provenance [75] due to an
data provenance, functionality, and trusted environment, no access control-oriented smart contract as well as a tamper-
matter whether the data are stored on-chain or off-chain. resistant provenance mechanism were designed and imple-
ProvChain [58] was a private chain in order to collect, store mented. When violations or misbehaviors were detected by
and verify provenance data in the cloud. In this blockchain- CSPs at the provenance phase, an automatic access control

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 6

TABLE II
B LOCKCHAIN - BASED P ROVENANCE M ECHANISMS IN C LOUDS

Approaches InO PBAC Validation Privacy Cloud Environment Consensus


Westerkamp et al. [57] N N Miners & participants N/A Supply chain-related PoW
Provchain [58] N Y Auditor Hashed identity OwnCloud PoW
SmartProvenance [59] N Y Voters No plaintext Google Drive PoW
MedShare [60] Y N N N/A Self-deployed PoW
GridMonitoring [61] Y N N N/A Self-deployed PoW
Neisse et al. [62] N Y N No Plaintext Distributed database PoW
Al-Mamun et al. [63] Y N Y N/A HPC Cloud Proof-of-Reproducibility
BlockCloud [64] Y N Y Encrypted Identity Multiple environments PoS
Maw et al. [65] N N Indexer contract N/A Cloud storage Proof of Authority
InO: Interoperability; PBAC: Provenance based access control.

was to be performed to revoke access to malicious or abnormal were stored in volatile memory, which communicated by high-
entities. Patient’s sensitive data were well-protected during speed and persistence protocols to reduce I/O overhead. In ad-
the data sharing process among medical institutes under the dition, a neoteric consensus protocol, Proof-of-Reproducibility
condition given in the work. (PoR), combined the idea of PoW and PoS, to realize trust-
Another work completed by Nessie et al. [62] constructed worthy provenance data validation and reproduction in volatile
a blockchain-based data provenance tracking system to fit in environment. Experiment results indicated that the proposed
the protection requirement of European Union (EU) citizen’s method outperformed traditional database and files provenance
personal data. Alike MedShare [60], smart contract in [62] system in latency overhead.
was utilized to record provenance information and store access Similarly, BlockCloud [64] also focused on the consensus
policy. In this system, three types of smart contract were protocols in blockchain-based cloud provenance system. In
cooperated together to support comprehensive provenance this approach, authors discussed 5 PoW provenance system
records from the perspective of both controller and specific security challenges, which were de-anonymization, 51% at-
group of data. Such that comprehensive metadata records were tack, blockchain fork, consensus delay and selfish mining. To
from two aspects (e.g., controller and data perspective), which tackle the problems above, BlockCloud was designed under
ensured the data sharing manipulations among CSPs. The work PoS to offer reliable data provenance. The provenance process
[76] considered the data stored in blocks and used differential in BlockCloud was similar to ProveChain [58]. Challenges of
privacy techniques to screen original data. The implementation PoS based cloud provenance system also were discussed in
of differential privacy in blocks was examined as an efficient this paper.
approach to against data mining-based attacks. Summary: Data provenance technology provided operation
From the perspective of security, blockchain-based prove- record and management throughout data’s lifecycle. Tradi-
nance method was deemed to be an approach for securing tional approaches in data provenance were centralized, com-
data operations in multiple manufacturing processes. Maw plex and lack of protection and validation. In this section, we
et al. [65] proposed a secure data operation architecture listed existing works that tackle the current challenges with
that addressed both immutability and redundancy dimensions. the support of blockchain. Further, privacy and interoperability
Blockchain system was used to construct a checking mecha- of provenance data were discussed. In future, the blockchain
nism for achieving integrity and a traceable replication mech- solution of cloud data provenance should be focused on
anism for data recovery. This solution could be considered smart contract based system that provides incentive to honest
a storage service combining with blockchain supports. Dis- behavior and punishment for malicious.
tributed network architecture also contributed to strengthening
self-defensive capability of data protection [77]. IV. B LOCKCHAIN - BASED ACCESS C ONTROL IN C LOUDS
Considering data integration, blockchain also was a techni- A. Access Control in Cloud Computing
cal alternative. Chen et al. [78] demonstrated that blockchain Access control was an essential method to provide cloud
technique could protect data integrity that was superior to data security and privacy, which kept cloud data from intrusive
traditional schemes. The approach was a stochastic blockchain by unauthorized users. An unreliable access control method
mechanism, by which the volume of cooperative nodes was also affected other functions, such as authentication, autho-
limited and IoT edge devices were used to work offloading. rization and data auditing. In this section, we discussed about
Some cloud services aimed to offer High Performance Com- challenges of traditional access control methods in clouds.
putation (HPC) instead of high volume of storage under pay- Traditional access control methodologies in clouds mainly
as-you-go pattern. Thus, these cloud data centers were diskless were based on well-established access control policies. Ex-
and shared the remote storage. Above blockchain provenance isting traditional policies were categorized into four aspects:
designs cannot perform well in HPC system due to the high namely, Discretionary Access Control (DAC), Mandatory Ac-
I/O overhead. Al-Mamun et al. [63] proposed an in-memory cess Control (MAC), Role-Based Access Control (RBAC)
blockchain to realize trustworthy and efficient provenance in and Attribute-Based Access Control (ABAC). In DAC, the
HPC system. In this novel architecture, distributed ledgers legitimate user (e.g. service provider) was responsible for

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 7

determining how other users access to objects (e.g. cloud environment. Decentralization could avoid hazards caused
users) [79]. By this method, a flexible access control for cloud by dishonest third party or participants [85]. Nugyen et al.
users was achieved due to the fact that no rigid rule was needed [86] addressed the scenario of medical records sharing and
in DAC. Contradicted to DAC, MAC was implemented by examined the performance of BAC. The evaluation suggested
a predefined trusting policy that could not be dynamically that BAC could offer trustworthy access controls for multiple
changed. System administrator was responsible for access medical providers.
controls instead of objects, such that the approach focused Thanks to the tamper-resistant and transparent nature of
on the confidently rather than integrity [24]. blockchain transactions, some approaches utilized blockchain
In RBAC model, access rights were assigned to subjects transactions to instruct the access control process in cloud
based on their roles and responsibilities in the system rather environment. Zyskind et al. [87] developed a decentralized
than their identities [24]. The nature of RBAC caused the personal data managing system for mobile data’s off-chain
drawback because of the lack of the consideration in other storage. Two different transactions exist in this blockchain
aspects of subjects. ABAC was proposed to further tackle network. Taccess, the first type of transaction, was designed
these issues. It configured the access rule based on the attribute for access control management. The other type of transaction,
analysis of objects and subjects [80]. Major benefit of ABAC Tdata, was responsible for data storage. Data owners were
was relevant with its comprehensive consideration during allowed to modify the access authentications by configuring
authentication. Even though authentication of ABAC was a different policy set in Taccess transaction. In addition, Tdata
time-consuming process, the computation resource it cost was cooperates with check policy protocol to govern the read/
negligible in the cloud environment. write operations. In such way, users fully controlled their data
As discussed above, each access control method has pros by implementing digitally-signed transactions, such that mali-
and cons. A common weakness of traditional access control cious intrusions (from unauthorized users) could be prevented
mechanisms is that they highly rely on a centralization set- in this blockchain-enhancement DAC model. To be specific,
ting, which generally lacks transparency, traceability, tamper- the protocol-based transaction implemented a dynamic and
resistance, and multi-party governance. Considering the spe- fine-grained access control protocol was implemented, which
cific application environment, the trade-off between security consisted of four protocols, including compound key genera-
and efficiency exists and can be hardly solved in nature. tion, permission check, access control and data on/off chain
protocols.
B. Blockchain-based Cloud Access Control Moreover, authors in [87] also discussed extensions of
Differing from traditional access control methods, blockchain in their work. At the first phase, the extension
Blockchain-based Access Control (BAC) has a few benefits could be realized by efficient off-chain data processing, so
deriving from characteristics of blockchain. Our investigation that the security of off-chain data needed to be secured during
suggests two key advantages. First, BAC introduces consensus data processing. To tackle this issue, an analysis method was
into the operation of access control, such that, logically proposed, which divided data into shares by using a secure
speaking, all stakeholders can be involved during the process. multi-party computation model [87]. The second phase of
Establishing a consensus generally needs an agreement- the extension was a measurement of trust in the blockchain
level consent made by participant voters or deciders, network. The new trust evaluation proposed as the sigmoid
which strengthens the security from the perspective of the function of the difference between the number of “good" and
decentralization. Second, traceability supported by blockchain “bad" actions. Through this trust measurement, the evaluation
provides a traceable and immutable governance capability for results showed that it could support the blockchain system
access control. This feature raises the difficulty of adversaries. against sybil attacks.
In this section, we review representative recent work in BAC. Another study, Engima [88], proposed a privacy-preserving
Due to the layered nature of cloud architecture, access decentralized computation system. It was constructed as an
control mainly played two roles in clouds. The first was the optimized secure multi-party computation system. Different
cloud service role who controlled the access of cloud users parties were designed for distinct computation tasks as well
in accessing the cloud data and services. A recent study, as preserving privacy jobs. With the implementation of this
BlockSLaaS [81] proposed a blockchain-assisted approach for design, computations and storage redundancy was eliminated
offering Logging-as-a-Service (LaaS). The proposed mecha- to address the scalability challenge. ProvChain [58] collected
nism addressed cloud forensics, which provided a represen- temper-resistant on-chain provenance data to perform Prove-
tative reference for merging blockchain and access control nance Based Access Control (PBAC), which was defined by
techniques. On the other hand, it played a visual role such that Nguyen [89] in 2012.
Virtual Machine’s (VM’s) access to physical machines needed We also observed that scalability was a challenge when
governance in case of threats from side channel analysis [82]. transaction-based access control was deployed. BBDS [90]
A decentralized access control technology powered by was proposed in order to tackle this issue, which relied
blockchain could avoid the risk of single point failure and on a lightweight block structure to improve the system’s
data misusing caused by third-parties. By applying blockchain efficiency and scalability. This model was applied to protect
technology, data owners could flexibly and completely control sensitive medical data stored in clouds. The unique design of
the access of their own data [83]. Recent study [84] demon- block structure provided high scalability in this private chain
strated that BAC could enable data sharing in an untrusted network. In addition, the identity-based authentication, cryp-

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 8

tography algorithms and transaction verification cooperated TABLE III


together to implement access control. However, BBDS was C OMPARISONS OF B LOCKCHAIN - BASED ACCESS M ANAGEMENT
M ECHANISM
not based on any open-sourced blockchain platform, which
made the BBDS immature and lack of extensive experimental SM Technique Sc. Platfrom ID FG
Zyskind et al. [87] Trans DAC N Self-Deployed N Y
verification. Enigma [88] Trans DAC Y Enigma N Y
As we stepped into the blockchain 2.0 era, smart contract BBDS [90] Trans RBAC Y Self-designed N N
ProvChain [58] Trans PBAC Y Ethereum N N
was another widely applied alternative that could be used to MedShare [60] SC PBAC∗ Y Self-designed Y N
strengthen access control. Some works constructed the smart FairAccess [92] SC OrBAC Y Bitcoin (reg-test) N Y
Novo et al. [93] SC DAC Y Ethereum N N
contract based access management system for sensitive health Zhang et al. [82] SC DAC+PBAC∗ N Ethereum Y N
DRAMS [95] SC ABAC N Ethereum N Y
and medical data under telemedicine scenario. Alansari et al. [96] SC ABAC N Ethereum N Y
Rahman et al. [91] designed a blockchain-based therapy Wang et al. [94] SC ABAC & DAC Y Ethereum N Y
management framework. While user’s medical data were Notations: SM: Service Mode; Trans: Transaction; SC: Smart Contract;
Sc.: Scalability; IN: Intrusion Detection; FG: Fine-Grained.
stored in off-chain clouds, access policy of off-chain data ∗ Revoke only.
was embedded into smart contract. Major drawback of this
approach was the dependency of trust third parties, such as
physio-therapy center, caregiver and therapist. MedShare [60]
e.g., cloud federation was not take into account. Thus, these
system used different kinds of smart contract cooperated with
designs can only provide confidentiality of single cloud data
cloud provenance data to achieve access control. While some
rather than whole cloud federation. In 2016, Federation-as-
contracts were responsible for judging potential misbehaviors
a-Service (FaaS) [97] proposed an access control system for
and threats based on provenance data, others execute potential
federated cloud environment. In this work, authors presented
misbehaviors and threats based on these provenance data. In
an attributed based access control mechanism with both Policy
addition, punishment was executed by revoking access rights
Enforcement Point (PEP) and Policy Decision Point (PDP).
of malicious cloud users. However, MedShare could only
PEP collected user’s request while PDP conducted access
revoke the access authentication; other essential operations in
permission.
access control were neglected.
Furthermore, the decentralized cloud environment also was A few more attempts ware made deriving from FaaS
an application scenario of smart contract-based access con- framework. For example, DRAMS [95] was constructed on
trols. FairAccess [92] implemented a smart contract-based the top of the FaaS framework. Smart contract was used to
access control mechanism under Organization Based Access collect, compare and check user’s logs for verification and
Control (OrBAC) model. This system could grant, read, intrusion detection in this system. DRAMS was immune to
delegate and revoke the access rights by different access access control decisions/ responses modifications and policy
tokens from transactions. Smart contract inside the transac- alterations. Challenges in this approach included vulnerable
tion provided fine-grained and context-aware access control off-chain system component and the latency issues. Similar
policies. To proof-of-concept, this system was constructed on to the intention of DRAMS, Alansari et al. proposed [96]
a Raspberry PI device. Similarly, Novo et al. [93] created an attribute based fine-grained access control method. The
a single smart contract system to manage access policies. system was spreaded out though the distributed federation,
In this work, a wide range of edge devices could access so that it preserved user’s privacy by implementing Oblivious
the blockchain network via management hubs. Besides, the Commitment-Based Envelope (OCBE) protocol. Blockchain
management hub provided a real-time access policy update maintained the attribute and policy integrity in this work. We
for all edge devices. In Zhang et al.’s work [82], three smart provide a comparison table for showing differences between
contracts, namely, access control contract, judge contract and various blockchain-based access management mechanism in
register contract, cooperated together to achieve both access Table III.
management and punishment for misbehaviors. Summary: Access control was an essential method to
Combining with the concept of ABAC, Wang et al. [94] prevent user’s data from intruding by unauthorized attackers.
proposed a blockchain framework for attributed-based fine- Traditional access control methods face challenges in signal
grained access control in the distributed storage system. Tra- point failure, unreliable trusted third parity and lack of user’s
ditional attribute-based encryption relied on trusted Public control. By implementing blockchain technology, users could
Key Generator (PKG). Compromising the PKG would lead fully control their data without threat of single point failure.
to system failure and sensitive information leakage. In Wang In addition, smart contract provided automatic access manage-
et al’s [94] design, blockchain was responsible for key man- ment as well as detection and punishment of misbehaviors.
agement rather than trusted PKG. Thanks to the involvement Moreover, these access control methods were all applied for
of blockchain, users fully controlled data through a tamper- secure cloud storage.
resistant and traceable key management, so that the method However, cloud VMs also needed access control mecha-
was free from the leakage and misusing caused by semi-honest nism to avoid side-channel attacks. In our best knowledge,
providers and single point failure. blockchain application in VM access management was short
Despite plentiful access control methods we had mentioned of study. In future, cloud VM security powered by blockchain
above, these researches (exclude MedShare, which only per- would be the next research trend. In Table II we compared
form revoke operation) were designed for single CSP scenario, and discussed different works.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 9

V. B LOCKCHAIN - ENABLED S EARCHABLE E NCRYPTION IN dressing the key leakage issue, the work [108] developed a key
C LOUDS aggregation searchable encryption scheme that against CPA.
Some approaches verified the search result with the support
A. Current Issues in Searchable Encryption
of broadcasted transactions. For instance, Searchchain [109]
Due to the concern of lack of control in personal outsourced was one of the methods using this technical route, which
data, users could choose encrypting data before uploading to deployed on the top of Obvious Key word Search with Authen-
the cloud to avoid exposing plain-texts. By this effort, the tication (OKSA) mechanism to provide a privacy-preserving
honest-but-curious service provider could not exquisite and public key encryption. Limitation of traditional Oblivious
analysis personal sensitive data. With the trade-off between Keyword Search (OKS) was addressed by novel OKSA mecha-
security and availability, this type of protection methods might nism by providing key word search authorization. Searchchain
reduce the service availability. Difficulty in searching the was proposed to reinforce privacy-preserving when CSPs
encrypted outsourced data was one of the common challenges. validated users’ access authentication via a specific keyword.
To address this challenge, searchable encryption was proposed Data retrieval information was recorded on the block and was
to ensure that users could acquire the search result without broadcasted to all nodes for verifications through a consensus
downloading all encrypted data stored in cloud. without acknowledging any keyword information.
Existing approaches of searchable encryption could be di- Bpay et al. [110] proposed a blockchain-based time com-
vided into two categories, which were Searchable Symmet- mitment scheme that used different kinds of transactions. In
ric Encryption (SSE) and searchable Public Key Encryption this model, dishonest party would be punished by bitcoin
(PKE). Song et al. [98] proposed a searchable encryption compensation without any Trusted-Third Party (TTP). In a
scheme on the basis of the symmetric encryption as a two- follow up study, TKSE [111] firstly came up with two-sided
layered form, which was known as a representative approach verification in searchable encryption system. Both malicious
of the first generation search encryption. On the fundamental service provider and malicious data owner could be punished.
of Song et al.’s work, Boneh et al. [99] put forward the public In order to verify the search result, authors constructed a
key based asymmetric searchable encryption. The work [100] merkle tree with ciphertext leafs, then check the results by
proposed conjunctive keyword search method to achieve multi- its root. Similar to Bpay, the payment fairness [116] was also
keyword search. However, all mechanisms discussed above based on the idea of time commitment.
were based on the exact matching, so that the availability In [112], returned keyword value and corresponding search
performance could be lowered down. To further tackle this tokens were recorded on search transactions for further ver-
issue, Li et al. [101] replaced exact matching by the fuzzy ification. Multi-set hashing, an incremental hashing scheme,
keyword search, which returned the result with the highest was used to process data integrity verifications. There were
similarity. two types of participants in this model. First, Clint Peers were
All work summarized above were constructed under the as- data owners where storage peers acted as service providers.
sumption that CSPs followed honest-but-curious model [102]. The other participant, Client Peers, requested verifiable cipher-
Nonetheless, this assumption was weak in a practical scenario text search from storage peers. Besides, this scheme provided
due to various insider threats. Dishonest servers might return an optimized storage overhead and dynamic updates. Unlike
false results under the consideration of energy saving and [110], [111], fairness mechanism in this work was provided
fault covering. It implied that verification process was desired by accountable metadata instead of time-commitment.
in searchable encryption scheme. Although some attempts Although verification mechanisms could provide soundness
[103]–[105] tried to verify integrity of returned values, no search results, miners still had a chance to deliberately skip
punishment could be made to incentive honest behaviors verifying complex transactions in order to focus on highly-
without a trusted party. Besides, verifications, from server side rewarded mining activities. The phenomenon was called the
to against malicious users, were not well-researched. Verifier’s Dilemma [117]. In the work [113], authors explored
smart contract to provide soundness keyword search with-
out complicated verification process on the data owner side.
B. Blockchain-enabled Searchable Encryption In other words, result soundness could be ensured as long
Some recent work focused on solving existing blockchain- as the Ethereum blockchain was safe. By embedding the
based mechanism. Cai et al. [106] accomplished a study on search algorithm in smart contract, correct results could be
strengthening encrypted keyword search through blockchain. ensured only when the contract was successfully executed
The work found the issue that malicious nodes could ruin on blockchain. The time consuming verification process of
search results when using a distributed hash table protocol to the searched data was no longer needed. Besides, author
integrate the encryption with keyword search. The proposed stored encrypted indexes in dictionary in order to reduce
solution could detect and remove malicious nodes as majority the computation overhead. Packing method also was used to
of nodes run a self-determining strategy. In order to against reduce gas cost.
Keyword Guessing Attacks (KGAs), Zhang et al. [107] pro- In addition, fair payment also implemented by smart con-
posed a blockchain-assisted PKE, named as a SEPSE. A few tract in this approach. Introduce fairness into the paying
methods were proposed by this work in order to reduce the process could incentive the honest and punish the malicious.
success chance of KGA, such as screening key encryption, Time commitment was used to ensure fairness in both single
periodically key renewal, and key request monitoring. Ad- user and multi-user setting. For example, Zhang et al. [110],

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 10

TABLE IV
S EARCHABLE E NCRYPTION

Approaches Service Mode Category Verification Fairness Soundness Accountability Model Configuration* Updateability
Searchchain [109] Transaction OKSA Provider Side N N Y Server-User Model N
Bpay [110] Transaction SSE Client Side Y Y N User-Server-User Model N
TKSE [111] Transaction SSE Double Side Y Y N User-Server-User Model N
Cai et al. [112] Transaction SSE Client Side Y Y Y User-Server-User Model Y
Hu et al. [113] Smart Contract SSE Do not Need Y Y N UserA-Server-UserB Model Y
Chen et al. [106] Smart Contract SSE Do not Need Y Y N UserA-Server-UserB Model Y
Wang et al. [94] Smart Contract SSE Do not Need Y Y N UserA-Server-UserB Model N
Zhang et al. [114] Smart Contract PEKS Do not need N Y N UserA-Server-UserB Model N
* Details and explanations of the configuration classification can be obtained from [115].

[111] introduced a fair payment to encourage honest behavior searchable encryption tackled the above challenges of SSE.
in SSE process. Blockchain in PEKS needed more research. We provide a few
Index of user’s file was stored on blockchain while the comparison results of searchable encryption solutions in Table
storage of file was outsourced to public clouds. Fairness [116] IV.
also was provided by smart contract in the form of time
commitment. Both single user setting and multi-user setting
were considered in performing fairness. Subsequent research VI. B LOCKCHAIN IN C LOUD DATA D EDUPLICATION
completed by Chen et al. [106] extended Hu’s work [113] into
electronic health record sharing scenario, which executed with A. Data Deduplication in Clouds
queries from different health agents. Different from Hu’s work,
index for EHRs was generated by complex logic expressions According to IDC’s report [56], cloud will store nearly 88
and stored in blockchain. By this effort, data owner could fully Zettabyte of data by 2025, where 75% of it is in the form of
control the access to their own data. duplicated data [118]. To maximize cloud storage efficiency,
Also addressed medical data, Zhang et al. [114] came majority of CSPs utilized data deduplication technology in
forward with a searchable public keyword encryption scheme their Storage-as-a-Service (StaaS) product, e.g. DropBox and
under personal health information sharing scenario. Private GoogleDrive. This technology could be beneficial both CSPs
blockchain, as deployed in a private cloud, was designed to and users in saving bandwidth, increasing storage efficiency
store the PHI data while the consortium chain stored the and reducing cost of power and infrastructure, as well as
indexes of PHI for encrypted search. The time controlled revo- resulting in a low price of services for customers. Nonetheless,
cation in this work preserved user’s data security from honest- deduplication approaches still encountered critical security
but-curious phyician. After the key word search operation was challenges.
finished, the trapdoor was disabled such that phyician could Data stored on cloud server could be cypher-texts in order
not use the trapdoor to access patient’s future data. to ensure the outsourced data confidentiality. However, CSPs
Combining with other technical methods was alternative generally denied users to encrypt their outsourced by conven-
in this domain as well. A recent work [94] developed a tional encryption method (e.g., AES) [119], which influenced
functional blockchain framework, which cooperated attribute the efficiency of the deduplication. Instead, convergent key
based encryption and SSE to achieve both fine-grained access encryption [120] was utilized to achieve cyphertext dedupli-
control and searchable encrypted keyword. Smart contract cation. Message-lock encryption (MLP) [121] later suggested
carried out fair payment mechanism by depositing user’s that convergent encryption was a kind of form of it. The
search fee in advance. However, fair pay algorithm in [94] authors also proved that the MLP was not semantic secure.
failed in compensating users when CSPs were dishonest e.g. A subsequent research [122] introduced a TTP to deliver
no punishment to CSP. Thus, fair payment mechanism in [94] tags that assisted with duplicate check. Potential challenges of
was not effective as [110], [111], [113] under the circumstance this approach, based on our observations, main derived from
of malicious service provider. its centralization design. This deduplication system would be
Summary: Searchable encryptions enable cloud users to disabled by a TPP single point failure. By intruding the TTP,
search their own outsourced data. Since CSP(s) might be attackers could obtain file tags for further side-channel attacks
honest-but-curious or malicious, encrypted search results could in source-based deduplication system [123].
be incorrect and misleading. With the support of blockchain, Moreover, data integrity also was threatened within dedupli-
the transaction verification and smart contract were used to cation process. After deduplication, only one copy remained
make search result authentic. In addition, fair payment was that could be a primary target for attackers. It meant that
also important. Malicious users could refuse to pay money both service outage and malicious administrator could easily
after they got correct data. Users could still pay for the erase storage content in an irreversible way. Data auditing
invalid search results when the payment system was not was essential in protecting users’ data in cloud, in which
correctly designed. Fair payment was implemented by the the deduplication was implemented. One solution to achieving
time commitment that embedded into either transactions or reliable auditing was to deploy a trust authority that relied on
smart contract. Until now, majority of the blockchain-based the treat of single point failure [124].

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 11

B. Blockchain-enabled Cloud Data Deduplication by this work, which included a single Byzantine fault and with
multiple faults.
Existing blockchain-enabled approaches mainly focused on
One of the methods of categorizing cloud resource alloca-
multi-cloud decentralized deduplication scheme. Blockchain
tion problems is on the basis of the “distance" from scheduling
technology was used to govern multi-cloud deduplication oper-
tasks to users [135]. That is to say, problems can be catego-
ations due to the incentive of high deduplication rate and fault
rized by different layers. At the cloud infrastructure layer, task-
tolerance performance. For example, CloudShare [125] intro-
scheduling focus on optimizations in datagram exchanging and
duced blockchain in multi-cloud deduplication management.
cloud federations. In addition, task scheduling generally deals
In this work, user side encryption was against attacks from col-
with the mapping issue from VMs to physical machines, as
luding malicious servers. User data’s integrity and ownership
well as optimum migration problems at the platform layer.
were protected by the temper-resistant blockchain transactions.
Finally, issues at the application layer mostly consider specific
Blockchain assisted multiple CSPs quickly synchronize file
optimization goals, such as satisfying Quality of Service (QoS)
informations for dynamically instructing deduplication system.
or energy-saving requirements.
Li et al. [126], [127] presented a smart contract-based
The challenging part is that most cloud resource scheduling
cloud deduplication system. Business Smart Contract (BSC)
problems are NP-hard problems while multiple elements are
performed periodically Proof-of-Retrievability (PoR) through
considered. A trade-off between the time complexity and
a challenge-and-response protocol to verify file integrity, re-
energy cost is representative example in scheduling problems.
trievability and resist side channel attacks. BSC dealt with file
In exhaustive methods, such as liner programming [136],
pointer management and the publication of Transaction Smart
time cost will be exponentially increased along with the
Contract (TSC), which was generated after the server passed
increment of the amount of variables. To tackle this problem,
PoR challenge and automatically transaction and payment
evolutionary approaches [137] (e.g. genetic algorithm, ant
management are processed. [128] outperformed [126] by its
colony optimization and particle swarm optimization) have
automatically file reconstruction, which was benefited from
been proposed to achieve the high-efficiency scheduling.
distributed storage.
However, as scheduling methods typically were based on a
Summary: Despite enormous significance of blockchain- central control hub, most existing methods failed in performing
based cloud deduplication, there were insufficient work done a real-time scheduling, which appeared inflexible in respond-
in this field until now. The conflict between high-redundancy ing varying requirements from users. Blockchain technology
blockchain data and the aim of deduplication was a critical provides a potential for solving the drawback caused by
obstacle. There was no such solution that blockchain could the control center, by constructing a decentralized resource
fully reconstruct cloud storage system with deduplication scheduling system.
technology. It seemed that blockchain served as a subsystem
for ensuring cloud storage security. Existing work stored file B. Blockchain-enabled Cloud Resource Allocation
tags on-chain while files remained off-chain. Considering this
We observe that blockchain technology generally is
setting, only small storage space was consumed while system
merged into cloud resource allocations by utilizing some of
security and data integrity could be enhanced.
blockchain’s technical features, including trustworthy platform
establishment (transparent and traceable token transactions)
VII. B LOCKCHAIN - ENABLED C LOUD R ESOURCE and smart contract applications.
A LLOCATION Distributed incentive scheme of resource allocation could be
achieved by implementing financial incentive tokens. Liu et al.
A. Cloud Resource Allocation
[126] proposed a novel Vehicular Ad hoc Networks (VANETs)
Typically, cloud resource allocation addresses two goals, scheme, where each Electrical Vehicle (EV) could be regarded
which are maximizing energy utilization and optimizing com- to be a mobile cloudlet as well as a moving power plant. The
putation efficiency [129], [130]. As widely accepted, cloud idling electrical resources of EVs were aggregated into source
datacenter is storing a huge volume of data and the volume is pool for flexible resource reallocation, which was directly
growing along with the enlargement of the service scope and influenced by encrypted coins. Data contributions between
the increment of the network. Heavy cloud workload means moving vehicles were rewarded by data coins followed Proof
the huge amount of electricity consumption at cloud datacen- of Data Contribution (PoDC) consensus. Similarly, electricity
ter. Constructing an energy-aware task scheduling mechanism contributions were rewarded by the “energy coin" to incentive
is a primary goal for cloud datacenter to reduce service resource contribution. Source allocation was performed based
cost, such that blockchain-empowered allocation method is on the amount of tokens possessed in each EV, according
an alternative [131], [132]. For example, Zhang et al. [133] to this blockchain-based system. A high-amount coins owner
attempted an approach of using blockchain to enhance com- means it contributed frequently in collaborative management.
putation capability of mobile edge computing. This study tried To incentive high frequency contribution, owners with high
to solve the joint computation offloading as well as coin coins could get a higher priority in accessing the resource
loaning problem by aiming the minimum value of the total pool with a lower price. This type of resource allocation
computing cost. Xu et al. [134] developed a Byzantine Fault- could be seen as token-based resource allocation. However,
Tolerant (BFT) networking approach in order to cover both the performance of this approach had not fully examined due
data security and system efficiency. Two cases were considered to lack of evaluations.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 12

TABLE V
C OMPARISONS OF B LOCKCHAIN - ENABLED R ESOURCE A LLOCATIONS IN
C LOUD C OMPUTING

Token/SC CE Platform Type


Liu et al. [126] Token MVC N/A I-RSA
Flipcoin [140] Token Cloud server Cloudlet I-RSA
Liu et al. [128] SC N/A Ethereum & Bitcoin E-RSA
Xu et al. [141] SC N/A Ethereum E-RSA
Notations: SC: Smart Contract; CE: Cloud Environment;
MVC: Mobile Vehicle Cloudlet; I-RSA: Incentive based RSA;
E-RSA: Energy Based RSA

Infrastructure service provisioning was one of the core Fig. 3. Offloading type classifications for blockchain.
application scenarios, in which a central broker was respon-
sible for resource allocation in most existing cloud solutions.
Ghosh et al. [138] proposed an approach that freed up the communication devices, such as mobile smart phone, could not
restriction of the single broker, so that a transparent allocation be deployed by PoW blockchain due to its constrained com-
operation was created. Thus, blockchain-enabled resource al- putation ability [26], [142]. Outsourcing this cryptographic
location was deemed to be a suitable option for making a puzzle game can be a solution in this scenario. Second,
transparent and trustworthy cloud federation [139]. Another blockchain system has low throughput and high redundancy so
work [140] achieved high-efficiency computation offloading that storing all data on-chain could result in tremendous cost of
between mobile devices and cloud servers with the assist of local hardware storage space. Applying a scalable outsourced
the “FlopCoin". FlopCoin was designed to create an incentive storage service was an option in addressing this issue.
scheme to encourage users executing the offloadable task. Since the invention of Bitcoin [143], PoW had been a major
Members’ mobility and reputation were also considered in consensus mechanism in blockchain system. In PoW mecha-
resource scheduling. Interactions between resource providers nism, miners scramble the right of packaging block by solving
and users were addressed as pricing strategy, such that both a cryptographic puzzle. Although PoW provides high fault
fixed pricing and auction mechanisms were considered in tolerance and security, it is computational exhausted to each
resource price. node in the blockchain network. Thus, resource contained edge
Due to the explosion of the machine learning technology, devices cannot afford such high computation cost. similarly,
reinforcement learning is widely used in the field of resource executing complex smart contracts was also a huge burden to
allocation. Liu et al. [128] proposed a joint framework, uti- resource limited devices [5]. The work [144] proposed a novel
lizing both deep reinforcement learning and smart contract to approach that used smart contract to optimize task allocation
achieve efficient cloud data collection and secure data sharing. in smart grid. The energy supply was dynamically and au-
In a follow up study, Xu et al. [141] firstly embedded the tomatically managed by implementing a smart contract. The
reinforcement learning algorithm into the smart contract to method used a reinforcement learning approach that could fit
perform optimum request migration in peer-to-peer cloud data in a continuously changing environment. Another work [145]
center network. They tackle the problem by proposing off- utilized smart contract to realize contractual routing protocol
policy temporal-difference learning (Q-leaning), which was in IoT, which also emphasized the auto control feature of
both in high efficiency and close to the real migration problem smart contract. Some researches offloaded computational tasks
where rewards and transfer probability were unknown. between devices and cloud servers to tackle this challenge.
Summary: Due to the economical nature of the blockchain,
some blockchain-based systems allocated resource via incen-
tively issuing tokens. The tokenized incentive based cloud A. User’s Profit-based Offloading
resource allocation could increase the resource sharing rate In some approaches, offloading strategy was designed to
that reduce the energy cost. Some approaches embedded rein- maximize the profit in the PoW cryptographic game from the
forcement learning algorithm into the blockchain system. Thus one perspective e.g., miner and service user perspective. These
resource could be optimally allocated among the peer-to-peer works assumes that the unit price of energy was fixed and pre-
cloud data center network. Blockchain-based cloud resource defined by an authority. Thus, the profit optimization problem
scheduling was short of research. Combination of incentive was converted into a problem that minimized the total cost
and optimum allocation methods could be next research trend. and maximized the computation efficiency.
We have listed a few representative work in Table V in order The work [146] proposed a mobile edge computing based
to exhibit major differences among distinct approaches. computation offloading and content caching joint system to
maximize the total net revenue, which was evaluated under
VIII. C LOUD O FFLOADING FOR B LOCKCHAIN the metric that considered task delay and energy consumption.
Despite benefits of blockchain system in security, trans- It was because the unity energy price was constant and the
parency and fault-tolerance, blockchain-based applications fre- optimization problem addressed minimizations. Two offload-
quently needed a powerful backend service supports. First, ing mechanisms were designed for computation insensitive
most blockchain systems utilized PoW consensus. Portable mining task. In the first solution, the full mining task Ahm,ni

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 13

was offloaded to nearby access point that connected to cloud the proposed method performed better than equal allocation
server. The second strategy divided the whole mining task into in regarded of the total reward.
subtasks, then distributed them to device to device users. The
optimal offloading decision was constructed under Alternating
B. Social Welfare-oriented Offloading
Direction Method of Multipliers (ADMM) [147] algorithm
to solve this distributed optimization problem. Relationship In this section, we focused on auction-based offloading
between users and time efficiency was considered under real- method to optimize the revenue or social welfare from the
world radio wireless communications, which paid attention to CSP perspective. Auction was a pricing method to dynamically
Rayleigh fading, noise, channel attenuation and CPU cycling. adjust the price of the resource based on the demand of
Meanwhile, power consumption of both active component, miners. Thus, cloud surplus resources could be effectively
e.g., CPU, and passive component, e.g., static circuit. used. Jiao et al.’s [152] work applied auction mechanism in
Extension of the work [146] was made and the findings edge servers resource allocation and pricing strategy to max-
were presented in [148]. Similar to [146], two offloading imize the cloud/edge service provider’s social welfare. From
strategies cooperated with ADMM algorithm had been pro- the user perspective, miner experienced two stage valuation
posed to achieve optimal total net revenue. Stochastic ge- in the auction process. In the first round, miner’s could not
ometry methods also were used in offloading and caching know the total resource amount and the number of winners.
metrics. Differ from the work [146], deadline constrain in The bid also was called an ex-ante valuation, accorded to
[148] derived from the probability of orphaning block instead the expected reward. In the next round, miners took auction
of the expectation of total delay. The difference between result into consideration and derived ex-post valuation. The ex-
probabilistic and deterministic constrains was revealed by the post valuation was defined as the multiplication of expected
experiments in this work. It showed that with probabilistic reward and network effect. The network effect in this work
constrains in outperformed deterministic constrains in the net was defined empirically by S-curve function that described the
revenue. Token cost in decision making was considered in trade-offs between number of miners, stability of blockchain
the optimization. This work presented a detailed discussion network and amount of resource allocated to each miner under
of the model performance and the miner’s preference under the constant demand model.
different deterministic backhaul constraints (γmBH ). The limi- From the resource provider’s aspect, optimal algorithm in
tation of this work is that the approach highly depended on [152] was proposed based on received bids to maximize
nearby network nodes (edge devices), such that the real-world the social welfare, which defined the difference of sum of
performance might be impacted by the number of participant ex-post valuations and total cost of CSPs. Winner selection
nodes. process used a greedy mechanism in this work to achieve the
In Liu et al.’s [149] follow-up investigations, blockchain maximum social welfare. After the winner set was determined,
offloading strategy was utilized in the video streaming sys- Vickrey-Clarke-Groves (VCG) [157] mechanism assisted the
tems, which enables individual participations. However, video calculation of payment. Five propositions suggested that the
transcoding is a computation intensive task that needed to be auction algorithm was truthful, individually rational, compu-
offloaded to nearby servers. The aim of the offloading was to tationally efficient and optimal in maximizing social welfare.
maximize the average transcoder’s transcoding profit, which Similarly, ETRA [158] was a three-stage auction model for
was issued by tokens. Both small cell base stations and Device- maximizing social welfare. The auction process was divided
to-Device (D2D) users could be offloaded. Then they used into potential winner matching, cloudlet - AP matching and
an ADMM-based algorithm to solve the non-convex problem resource allocation steps.
in a distributed manner. The blocksize in this blockchain However, the work [152] only considered the constant
network dynamically was adapted to achieve the optimization demand situation. In the extended work [153], both constant-
of proposed task. demand and multi-demand were considered in designing
Furthermore, investigations [150], [151] focused on the miner’s auction scheme. Network effect in [153] was de-
total reward from the perspective of Mobile Terminals (MT) rived by the curve fitting of the real-world experiment data
while MTs offloaded computational intensive PoW tasks to instead of empirical assumption to make the function ac-
the Edge Servers (ES). Fairness between MTs was taken curate. In multi-demand bidding, maximizing social welfare
into the consideration. We mainly analyzed the work [150] was a non-monotonic submodule maximization problem un-
because it is the extended version of [151]. Two distributed der knapsack constraint. An approximate algorithm was con-
optimized algorithms were proposed to fit both single ES and structed to obtain sub-optimal welfare for tackling this NP-
multiple ESs scenarios. The solution vertically divided the hard problem. Two efficient auction schemes named FLRS
non-convex Total Reward Optimization (TRO) problem into and MDB were designed to achieve sub-optimal of social
two subtasks: TRO-sub and TRO-top tasks. TRO-sub was a welfare in multi-demand bidding. Although two achievements
convex optimization problem such that it was optimized by [152], [153] above considered the resource constrain of server
bisection-search or diminish step size in single or multiple and blockchain network, real communications had not been
ES, respectively. To optimize the TRO-top, they used a linear- assessed (e.g., Rayleigh fading, noise, channel attenuation).
search with small step to find the best viable interval, which We also noticed that the work [152], [153], [158] discussed
could maximize the sub-problem in single ES, and randomized above only considered one service provider social welfare
search in multiple ES scenario. Experiment results showed maximization model. In the real market scenario, there were

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 14

TABLE VI
O FFLOADING FOR P OW P UZZLE

Optimization Target & Goal Algorithms & Model User-Provider Relationship∗ Smart Contract
Liu et al. [146] Miner’s revenue ADMM Algorithm MPMU N
Liu et al. [148] Miner’s revenue ADMM Algorithm MPMU N
Liu et al. [149] Transcoder’s revenue ADMM Algorithm MPMU N
Wu et al. [151] MT’s Profit Convex Optimization MPMU N
Jiao et al. [152] Social Welfare Single Side Auction SPMU N
Jiao et al. [153] Social Welfare Single Side Auction SPMU N
Li et al. [154] Social Welfare Double Side Auction MPMU As a Broker
Luong et al. [155] Provider’s Profit Two-Stage Stackelberg Game SPMU N
Xiong et al. [156] Provider’s profit Two-Stage Stackelberg Game SPMU N
Notation: SPMU: Single Provider-Multiple Users; MPMU: Multiple Providers-Multiple Users

multiple service providers to offer PoW offloading. Therefore, TABLE VII


the resource allocation and pricing strategy should be devel- C LOUD M INER H ARDWARE (D E .: D ECENTRALIZATION ; F LEX .:
F LEXIBILITY; E.E.: E NERGY E FFICIENCY; H.P.: H ASHPOWER ; U.R.:
oped based on double auction mechanism [159], which took U SER ’ S R EVENUE ; H: H IGH ; M: M ODERATE ; L: L OW )
competitions between CSPs into the account.
Li et al. [154] applied an iterative double-sided auction in Category Stage De. for PoW Flex. E.E. H.P. U.R.
CPU Stage 1 H M L L L
the multi-CSP situation, which attained both social efficiency GPU Stage 2 L M M M M
maximization and privacy preservation were attained. A smart- FPGA Stage 3 M H H M M
contract designed broker was responsible for trading manage- ASIC Stage 4 L L H H H
ment between competitive CSPs and miners. The broker first
collected demands from miners. In each double auction inte-
gration, it extracted the hidden information while preserving In addition, this work took both uniform and discriminatory
CSPs/users’ privacy, and update bidding, pricing and allocation into the account. Experiment results revealed that profit of
strategy to the optimum utility. The broker executed the social CFPs and resource demand from miners were higher under
welfare maximization algorithm for achieving high-efficiency discriminatory than uniform pricing.
market adoptability. Luong et al. [155] utilized a deep neural network to do
the monotone transformation, allocation and pricing, in order
C. CSP Profit-based Offloading to achieve maximum revenue for service provider; however,
A few aspects need to be considered from the perspective this machine learning-based method only considered single
of CSP, such as energy and cost saving. Qiu et al. [160] resource unit.
emphasized that blockchain could not be directly applied in Summary: In this section, we discussed the task offloading
IoT or other networking environments due to most network and resource allocation to assist mobile users solve PoW
nodes were heavily resource-limited. The work suggested that puzzle. We concluded that three objects were concerned to
combining agent mining with cloud mining was a feasible be optimized during the offloading. Table VI presents a
method to diminish the limitation. Chen et al. [132] pointed comparison result about the offloading for PoW puzzle, based
out that computation offloading and blockchain mining needed on our investigations.
to considered in a unified manner. The complexity of the issue From miner’s perspective, mining and PoW task offloading
was high when both computation and mining were considered, strategy were dynamically optimized to achieve miner’s total
as more participants were involved. The work [132] devel- revenue. From service provider’s point of view, game theory
oped a distributed algorithm (exchanging messages) between was utilized to find the optimal equilibrium point to ensure
network nodes to reduce the computational complexity. maximum provider’s profit. Social welfare, which represented
Moreover, some explorations tried game theory to model the total utility in the system, could be maximized by auction
the interaction between selfish service provider and users. mechanism. Future works in this field includes the evaluation
Xiong et al. [161], [162] used a two-stage Stackelberg game of trade-offs between social welfare optimization and provider
to model the pricing process. Service provider in this game profit maximization, deep learning technologies in pricing and
acted as the leader, where miners were followers in this game. auction under multiple service providers.
As a result, provider’s profit was maximized when reached
the equilibrium. In a follow-up study [156], trading between IX. C LOUD H ARDWARE IN B LOCKCHAIN M INING
cloud/fog providers (CFPs) and rational miners also was Increasing difficulties during the blockchain mining process
modeled by two-stage Stackelberg game. During the sub-game made on-premises deployment of miners expensive and space-
in stage II, miners, e.g. followers, determine their demands to consuming. Due to the development of visualization technol-
maximize the resource utility under the price set by providers. ogy and parallel computing, cloud could be a flexible, pay-
Cloud/fog providers, e.g. leaders, decided their pricing strategy as-you-go manner with high computation performance. In this
to maximize provider’s profit. These two subgame formed the section, we focused on two technical dimensions, latency or
two-stage Stackelberg game. Backward induction was applied energy costs. Differing from traditional blockchain mining,
to achieve nash equilibrium, which represented the joint op- cloud-based mining had an observable advantage. The merit
timization of both CFP’s profit and miner’s resource utility. was caused by the centralization setting of clouds, such that

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 15

both energy saving and efficiency performance could be im- peak was reduced and natural involvement in cooling can be
proved due to the hardware resource optimization. Cloud data maximized in night. Kamp et al. [174] discussed some recent
center’s processing unit’s computation ability fundamentally 2-phase cooling system for blockchain miners.
determined its PoW executing efficiency [163]. Therefore, Field Programmable Gate Array (FPGA) was flexible and
some researches and products aimed to increase the mining could be used to accelerate a specific kind of computation
success rate by optimizing the hardware design and fabrica- problem, such as mining. Xilinx’s FPGA cannot suffer mining
tion [164]. The hardware level computational optimization in task due to the lack of cooling and computation capability.
cloud data center had gone through four stages: CPU, GPU, Some hackers [175] developed an open source customized
FPGA and ASIC [165], [166]. For example, Liu et al. [167] FPGA miner with reduced I/O and RAM component, which
proposed a hardware-assisted mechanism for secret sharing, were redundant for mining process.
which scaled up the number of nodes in BFT protocols. Table
VII exhibits a comparison between distinct hardware aspects.
C. ASIC-based Mining
A. CPU-based Cloud Mining Application specific integrated circuit (ASIC) technology
was developed for producing customized integrated circuit
Bitcoin’s cartographic mining puzzle was implemented by
for specific application scenario. Therefore, developing the
SHA256 [143]. In [168], bitcoin mining process was defined
ASIC miner could be rewarded due to its optimized design for
as a constrained input small output problem (CISO). In this
finding the hashed value. A relatively low resource cost could
CISO problem, miners aimed to find the available nonce,
achieve a high hash rate performance. Butterfly Labs (BFL)
Merkle root and time stamp to make H2 < target. However,
[176] launched a ASIC miner based on their FPGA miner
CPU was computational constrained. Mid-state buffer [169]
in 2012. This miner was fabricated under 65nm process with
hashed the initiating information before the nonce creates a
maximum speed 1,500GH/s. In 2013, Avalon et al. [177] pro-
constant hash value, in order to improve the mining efficiency.
posed a miner with 110-nm TSMC fabrication. This product
Furthermore, nine improvements to speed up the CPU mining
could reach 66Gh/s with 60W power. Thanks to the develop-
were discussed in [168].
ment of transistor fabrication technology, performance of the
However, recent development of GPU and ASIC miner
ASIC miner promoted exponentially followed the Moore’s law.
violated the Nakamoto’s 1 CPU 1 vote provision. To encourage
Avalon’s latest product AvalonMiner 852 was fabricated under
CPU miners, Litecoin et al. [170] used SCRYPT [171], which
16-nm technology. It could achieve 15TH/s with 100W/T
consumed both computational and storage capacity in mining
energy efficiency. By 2019, Bitmain’s [178] product occupied
process. Similarly, Ethereum’s PoW mechanism based on
70% percent miners in the market. They firstly proposed 7nm
Ethash. In Ethash’s mining process, memory bandwidth was
miner. It can achieve 40TH/s with tolerable energy cost.
a major element that effected the mining rate.
Summary: We discussed four different cloud mining hard-
ware generation. CPU-based mining was original approach. It
B. GPU/FPGA-based Cloud Mining provided fully decentralization during PoW consensus. GPU
CPU was designed as the general processing unit. Compo- miner, though energy costly, offered high amount of hashrate.
nents other than arithmetic logic unit such as registers and FPGA miner was flexible and optimizable for any mining
branch prediction units, were redundant in mining. Therefore, protocol. However, FPGA miner’s computing capability was
repetitive mining work for CPU was waste of time and energy. limited. To tackle this problem, a customized, energy efficient,
Although SHA256 computation rounds cannot be paralleled and hashpower sufficient ASIC miner was designed and widely
compute, different nonces could be tested simultaneously with used since 2014. Although ASIC miner could take both speed
each other by parallel computation [165]. In GPU cloud and efficiency into consideration, it is not as flexible as
mining center, two elements should be considered. FPGA miner. For FPGA miners, when consensus mechanism
First, hardware overhead could be minimized. In [169], an changed, FPGA’s logic should be rewritten but the hardware
economically GPU based miner was introduced. This miner could remains. However, ASIC’s logic cannot be changed
used AMD’s 7970 GPU rather than NVidia’s to provide higher after tape-out. Therefore, the lifespan of ASIC miner was
SHA256 computation speed. Besides, to reduce the GPU depend on the pace of consensus change. In addition, cloud
overhead, they used 1x PICe slot instead of 8x and 16x in com- data center might not execute homogeneously mining task.
mercial motherboard. Additionally, low-cost PCIe converter Different mining task cause the ASIC cloud heterogeneity. By
connected 16x AMD GPU connector and 1x slot. Therefore, using the FPGA, cloud data center could keep homology. As
overhead per each GPU was greatly reduced. Ekbote et al. a result, trade-offs between cloud data center’s homology and
[172] designed a NVIDIA GPU mining accelerator. Authors efficiency is well-worth study in the future.
in this work developed the framework by CUDA to perform
general purpose computing for mining task. Experiments in
X. C LOUD S TORAGE AND B LOCKCHAIN
this work revealed the prior performance of GPU compared
to CPU in mining task. Using cloud storage is an alternative method for reinforcing
Second, powering and cooling system should be carefully blockchain systems, from the perspective of mass data storage,
and efficiently designed. Skach [173] designed a thermal time which reduces the restriction caused by the limited storage
cooling data center by phase charging materials. Thermal space of blocks. The crucial issue is to determine which data

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 16

shall be stored in blocks in order to retrieve a great balance be- the scalability issues of the blockchain system. Alike work
tween block size and blockchain functionality. Another aspect [87], off-chain data stored in distributed cloud server powered
is that using blockchain techniques to facilitate secure data by the DHT technology.
sharing, e.g., healthcare [75] and smart city [179]. Zheng et
al. [180] pointed out the potential hazard caused by centralized
B. On-chain Metadata Storage
data storage and demonstrated a blockchain-based data sharing
for securing personal health data. Another study [181] also Works we have discussed above all stored hash pointer of
addressed medical data sharing and further assessed the per- the file on-chain. While this method could protect the data
formance of blockchain in enabling multi-party collaborations. from outsider’s temper, the data were lack-of-control under
Qian et al. [179] completed a study that demonstrated a secure the threat of malicious insider. To tackle this problem, Zhu
data exchanging between different organizations. However, [15] proposed a controllable voting scheme based on on-
most recent work concentrated on examining the potential of chain metadata. This work tackled collude attacks during the
merging cloud storage with the existing blockchain techniques. file modification process. In this work, cloud servers offered
The attempts were encountering various restrictions, such as storage-as-a-service to user’s file. By this effort, blockchain
multi-chain collaboration. storage efficiency was greatly enhanced. Meanwhile, three
kinds of temper-resistant metadata (e.g., voting record, file
changing history, and changed data’s hash values) were stored
A. On-chain Hash Value Storage on blockchain to make the system secure and controllable.
Although blockchain was capable of storing temper-resistant In this approach, on-chain metadata provided transparent and
data on-chain, blockchain’s storage capability was limited. For traceable history of previously valid and invalid changes of the
example, Bitcoin’s block size was only 1MB, which cannot document. Therefore, the file revisions could be conveniently
used to storage large volume of data [36]. Meanwhile, storing checked by users. Thus, the system was controllable compared
high amount of data on-chain could lead the full node size to above works. Trade-off between storage efficiency and
of the blockchain network intolerable high to PC and mobile data security was optimized by selecting suitable metadata
users. Thus, off-chain [182] storage technology was introduced on-chain. Extensive experiments revealed the approach was
to tackle this challenge. CSP could offer storage-as-a-service secure, controllable, privacy-preserved, and unforgeable.
as a scalable off chain solution to users. Guo et al. [183] The work [187] applied blockchain to construct a decen-
proposed a two-way-pegged multi-sidechain approach in order tralized PingER model. Metadata included merkle roots with
to increase throughput through a novel multi-sidechain setting. raw data leaves and file off-chain storage locations were stored
For explicitly presentation, we show a brief comparison about on-chain. Raw files offloaded to distributed monitory agents
cloud storage for blockchain in Table VIII. storage system that constructed by DHT. With the on-chain
Zyskind et al. [87] used off-blockchain key value store metadata, access control and identity verification could be
in a blockchain-based automatic access control system. To performed to ensure data security in off-chain storage system.
protect user privacy, user’s data were first encrypted. Then, Summary: In this section, we discussed off-chain solu-
these cyphertext routed to the off-blockchain storage system, tions to address the blockchain storage limitations. Apply-
which was implement by Distributed Hash Table (DHT) based ing the off-chain storage benefited blockchain in scalability,
distributed cloud. Only hash value of each file saved on the storage efficiency and verification speed. As we can see
public ledger. By this effort, user’s data privacy and blockchain in the previous literature, trade-off between data security
system efficiency were jointly optimized. The further study and blockchain system’s efficiency was a essential problem
[88] used off-chain storage in their secure multi-party compu- when designing the on/off chain system. Storing many types
tation system (Enigma). related metadata could ensure the off-chain data secure and
Similarly, Sun et al. [184] applied off-chain technology to controllable. However, high amount of data on-chain could
store high amount of EHR data. In this work, EHR data ad- reduce the blockchain’s scalability and efficiency. Therefore,
dress was stored on-chain, while the EHR data were encrypted how to select suitable data on-chain to tackle both security and
and saved in data owner’s off-chain database. The work intro- scalability issues could be a rewarded future research direction.
duced attributed based signature to secure data sharing in the
off-chain database. During the sharing process, the data owner XI. M AIN F INDINGS AND D ISCUSSIONS
firstly signed the EHR data’s address with his attributes. Then
the signed address was stored in the blockchain transactions. A. Main Findings
Users verify the data owner’s signature when retrieving the We explain main findings of this study in a few angles,
data. Rifi et al. [185] also stored hash pointer of each file on- which include the similarity, connectivity, and creativity.
chain. In Rifi’s work, off-chain database was the widely used Similarity: The primary similarity between two technolo-
Inter Planetary File System (IPFS). gies is that both technologies highly rely on the decentralized/
Following the exploration above, Shafagh et al. [186] re- distributed networking environment. Even though cloud dat-
alized secure and reliable access control and key manage- acenter delivers services in a centralized computing manner,
ment for the off-chain storage system with the support of distributed/ decentralized settings still exist in clouds, such
blockchain. In their work, blockchain acted as a control layer as distributed multi-tenant, heterogeneous clouds deployment,
over the blockchain. Distributed cloud storage used to tackle and third-party service providers.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 17

TABLE VIII
C LOUD S TORAGE FOR B LOCKCHAIN

Approaches On-chain data Application Scenario


Zyskind et al. [87] Hash Pointer of Off-chain Key Value Automatic Access Control System
Enigma [88] Hash Pointer of Personal Sensitive Data Multi-party Computation System
Sun et al. [184] Hash Pointer of Personal Sensitive Data Secure Medical Data Sharing
Rifi et al. [185] Hash Pointer of Data in IPFS Database IoT Data Access Control
Shafagh et al. [186] Hash Pointer of Personal Sensitive Data Auditable IoT Data Sharing and Storage
Zhu et al. [15] Voting Record, File Revise History, Hash Value of Changed Data Controllable Voting Mechanism
Ali et al. [187] File Off-chain Location, PingER Data Merkle Root PingER Data Access and Storage Management

We observe that this similarity results in similar technical clouds derive from value creations, which can be categorized
attentions. From the perspective of the service, for example, into two motivations. The first intention is adding values to
both technologies address intelligent controls (e.g., resource the existing system. Blockchain technique is used to solve
allocations) to achieve a higher-level service quality. A smart the weakness of cloud solutions, such as lack of control and
contract plays a similar role to a cloud controller. Blockchain lower-level trust. Merits of blockchain are considered to be a
and cloud computing also share some service models, such as supplement of clouds, such that additional values are added
BaaS and X-as-a-Service, which facilitates two technologies to the existing cloud models. The other creativity route is
into one concept. to create new values. Cloud computing provides blockchain
Due to characteristics of distributed networks, blockchain system with resource supply, e.g., infrastructure and software,
and cloud computing also share similar network-relevant con- so that new service model has been created (BaaS). In practice,
cerns (e.g., security and privacy issues). A few cyber threats in BaaS still is at an early stage and more research is required
cloud computing apply to blockchain network, such as identity for fitting in distinct demands.
leakage and data mining-based attacks, although adversarial
methods maybe distinct. Data recorded in blocks are available
B. Discussions
to all authorized users, which implies a possibility of privacy
leakage from mining block data. The situation is similar to data We summarize a few representative challenges and oppor-
stored in remote cloud servers. Even for an anonymous cloud tunities in this section.
dataset, privacy leakage may take place when a successful Challenges: First, BaaS operations in multi-cloud (cloud
linkage attack is launched. Both outsider and insider attacks federation) need more attempts in various dimensions, such
are threatening two technologies. as architecture, communication, and consensus mechanisms.
In addition, similar to many prior researches in optimiz- As multi-chain technique is insufficiently mature, bridging
ing efficiency of cloud systems, enhancing performance of up different BaaS service providers still is a challenging
blockchain systems also is a popular research topic. Efforts issue. Second, challenges in blockchain-enabled cloud data
have been made for both technologies in architecture improve- provenance are varied. It is still challenge to verify whether
ment, hardware re-design, resource allocation strategy, and data are used by unexpected parties in a network environment.
data provenance. It is also hard to decide whether the data in blocks are actually
Connectivity: Interconnectivity exists between blockchain associated with physical objects, e.g., animals or hardware.
and cloud computing, based on our investigations. First, BaaS Technical development in other domains is needed. Third,
is a service model that derives from cloud computing. Many IT security of blockchain still is encountering a great challenge.
companies are seeking new markets by offering BaaS service From blockchain infrastructure to smart contract, potential
offerings. Blockchain infrastructure and backend support are threats exist in current blockchain systems from top to bottom.
two major parts in most current BaaS models. Second, abun- Even though blockchain can strengthen security protection,
dant computing resources in clouds are the supplementation there are many unsolved issues, such as complex network
of blockchain system for enhancing security, strengthening nodes setting, multi-chain environment, hardware attacks, and
efficiency, and improving service quality. The supplementary privacy leakage. Finally, many performance specifications of
offerings not only cover software but also address hardware- blockchain system have not reached other mature active sys-
relevant supports, such as blockchain-oriented equipment (e.g., tems, e.g., throughput capacity, energy cost, and data storage.
blockchain-purpose chips). Consensus mechanism plays a vital Research Opportunities: This work focuses on the inte-
role in building up a trustworthy environment for cloud appli- gration of blockchain and cloud computing; however, other
cations. Finally, with a nature of auto control, smart contract associated technologies shall not be ignored for future updates
has a great potential in a wide scope of cloud applications, e.g., and research. We observe that future integration may have a
resource allocation and intelligent manufacturing. We observe higher-level coverage that combines multiple network-related
that smart contract is a crucial interconnection point by which technologies, based on our investigation on prior studies in
blockchain and cloud resource can be merged. other manners of integrations, e.g., software-defined network
Creativity: Emerging service models have been introduced [188], IoT [189], and cloud radio access network [190].
to public in recent years. From the perspective of value chain, In addition, perfecting BaaS service model will become an
we find that primary creativity of combining blockchain and attractive research topic in both academia and the industry.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 18

Besides infrastructure and backend, more service offerings [5] K. Gai, K. Choo, and L. Zhu, “Blockchain-enabled reengineering of
(e.g, AI-related services or access control) are needed to meet cloud datacenters,” IEEE Cloud Comp., vol. 5, no. 6, pp. 21–25, 2018.
[6] B. Chen, Z. Tan, and W. Fang, “Blockchain-based implementation for
different demands. Moreover, customized data provenance and financial product management,” in The 28th Int’l Telecommunication
object tracing services will turn into a great domain in practice. Networks and Applications Conf. IEEE, 2018, pp. 1–3.
Solving the issue of attaching physical objects to network [7] A. Hari and T. Lakshman, “The Internet blockchain: A distributed,
tamper-resistant transaction framework for the Internet,” in 15th ACM
nodes or data may lead an industrial transform. Furthermore, Workshop on Hot Topics in Networks. ACM, 2016, pp. 204–210.
research on security and privacy issues in blockchain is an [8] F. Hardwick, R. Akram, and K. Markantonakis, “Fair and transparent
unavoidable topic. Future work needs to pay attention to both blockchain based tendering framework-a step towards open gover-
nance,” in 17th IEEE Int’l Conf. TrustCom. New York, USA: IEEE,
blockchain-related attacks (e.g., attacks at smart contract) and 2018, pp. 1342–1347.
emerging threats to blockchain-cloud systems. Finally, high [9] N. Fabiano, “Internet of Things and blockchain: legal issues and
performance is still a keyword in the domain of blockchain- privacy. the challenge for a privacy standard,” in IEEE Int’l Conf. on
IoT. IEEE, 2017, pp. 727–734.
cloud in the foreseeable future. Both software and hardware [10] W. Meng, E. Tischhauser, Q. Wang, Y. Wang, and J. Han, “When
need to be improved in order to deal with complex or heavy- intrusion detection meets blockchain technology: a review,” IEEE
workload environments. Access, vol. 6, pp. 10 179–10 188, 2018.
[11] K. Biswas and V. Muthukkumarasamy, “Securing smart cities using
blockchain technology,” in 18th Int’l Conf. on HPCC. Exeter, UK:
XII. C ONCLUSIONS IEEE, 2016, pp. 1392–1393.
[12] R. Cole, M. Stevenson, and J. Aitken, “Blockchain technology: impli-
This work addresses a few technical dimensions for reengi- cations for operations and supply chain management,” Supply Chain
neering of cloud computing by using blockchain technology. Management: An Int’l J., vol. 24, no. 4, pp. 469–483, 2019.
Three technical dimensions are involved in the work, namely, [13] K. Gai, Y. Wu, L. Zhu, M. Qiu, and M. Shen, “Privacy-preserving
energy trading using consortium blockchain in smart grid,” IEEE Trans.
service, security, and performance. To be specific, this sur- on Industrial Informatics, vol. 15, no. 6, pp. 3548–3558, 2019.
vey explains contemporary explorations in blockchain-enabled [14] I. Eyal, A. Gencer, E. Sirer, and R. V. Renesse, “Bitcoin-ng: A scalable
reengineering of cloud datacenter through following aspects, blockchain protocol,” in 13th {USENIX} Sym. on Networked Systems
Design and Implementation, 2016, pp. 45–59.
including BaaS service model, blockchain-enabled cloud [15] L. Zhu, Y. Wu, K. Gai, and K.-K. R. Choo, “Controllable and trust-
access control, blockchain-enabled cloud data provenance, worthy blockchain-based cloud data management,” Future Generation
blockchain-based cloud searchable encryptions, blockchain- Computer Systems, vol. 91, pp. 527 – 535, 2019.
[16] E. Heilman, F. Baldimtsi, and S. Goldberg, “Blindly signed contracts:
based cloud data deduplications, smart contract-based cloud Anonymous on-blockchain and off-blockchain bitcoin transactions,” in
applications, blockchain-powered offloading, blockchain hard- Int’l Conf. on Fin. Crypt. & Data Sec. Springer, 2016, pp. 43–60.
ware development, and blockchain-related cloud storage. Main [17] N. Herbaut and N. Negru, “A model for collaborative blockchain-based
video delivery relying on advanced network services chains,” IEEE
findings of this work provide future work with a theoretical Communications Magazine, vol. 55, no. 9, pp. 70–76, 2017.
reference in the field of blockchain-enabled reengineering of [18] D. Qin, C. Wang, and Y. Jiang, “Rpchain: a blockchain-based academic
cloud datacenter. social networking service for credible reputation building,” in Int’l
Conf. on Blockchain. Springer, 2018, pp. 183–198.
[19] Y. Xu, G. Wang, J. Yang, J. Ren et al., “Towards secure network
ACKNOWLEDGMENT computing services for lightweight clients using blockchain,” Wireless
Communi. and Mobile Comp., vol. 2018, 2018.
This work is partially supported by Ministry of Edu- [20] F. Tschorsch and B. Scheuermann, “Bitcoin and beyond: A technical
cation - China Mobile Research Fund Project (Grant No. survey on decentralized digital currencies,” IEEE Communications
MCM20180401), National Natural Science Foundation of Surveys & Tutorials, vol. 18, no. 3, pp. 2084–2123, 2016.
[21] M. Khalilov and A. Levi, “A survey on anonymity and privacy in
China (Grant No. 61972034, 61872041, U1836212), Natu- bitcoin-like digital cash systems,” IEEE Communications Surveys &
ral Science Foundation of Beijing Municipality (Grant No. Tutorials, vol. 20, no. 3, pp. 2543–2585, 2018.
20D20116), Pre-study Foundation of Weapons and Equipment [22] M. Conti, E. Kumar, C. Lal, and S. Ruj, “A survey on security and
privacy issues of bitcoin,” IEEE Communications Surveys & Tutorials,
(No.31511020401), Natural Science Foundation of Shandong vol. 20, no. 4, pp. 3416–3452, 2018.
Province (Grant No. ZR2019ZD10), Guangxi Key Laboratory [23] X. Li, P. Jiang, T. Chen, X. Luo, and Q. Wen, “A survey on the security
of Cryptography and Information Security (No. GCIS201803), of blockchain systems,” FGCS, p. 1, 2017.
[24] T. Salman, M. Zolanvari, A. Erbad, R. Jain, and M. Samaka, “Security
Henan Key Laboratory of Network Cryptography Technology services using blockchains: A state of the art survey,” IEEE Commu-
(Grant No. LNCT2019-A08), Australia ARC DP180102828 nications Surveys & Tutorials, vol. 21, no. 1, pp. 858–880, 2018.
and ARC DP200101374, Beijing Institute of Technology Re- [25] Q. Feng, D. He, S. Zeadally, M. Khan, and N. Kumar, “A survey
on privacy protection in blockchain system,” Journal of Network and
search Fund Program for Young Scholars (Dr. Keke Gai). Computer Applications, vol. 126, pp. 45–58, 2019.
[26] R. Yang, F. Yu, P. Si, Z. Yang, and Y. Zhang, “Integrated blockchain
R EFERENCES and edge computing systems: A survey, some research issues and
challenges,” IEEE Communications Surveys & Tutorials, p. 1, 2019.
[1] D. Li and W. Viriyasitavat, “Application of blockchain in collaborative [27] M. Ali, M. Vecchio, M. Pincheira, K. Dolui, F. Antonelli, and
Internet-of-Things services,” IEEE TCSS, vol. PP, no. 99, 2019. M. Rehmani, “Applications of blockchains in the internet of things:
[2] T. Aste, P. Tasca, and T. D. Matteo, “Blockchain technologies: The A comprehensive survey,” IEEE Communications Surveys & Tutorials,
foreseeable impact on society and industry,” Computer, vol. 50, no. 9, vol. 21, no. 2, pp. 1676–1717, 2018.
pp. 18–28, 2017. [28] H. Dai, Z. Zheng, and Y. Zhang, “Blockchain for internet of things: A
[3] A. Dorri, M. Steger, S. Kanhere, and R. Jurdak, “Blockchain: A survey,” IEEE IOT J, vol. 6, no. 5, pp. 8076–8094, 2019.
distributed solution to automotive security and privacy,” IEEE Com- [29] J. Xie, H. Tang, T. Huang, F. Yu, R. Xie, J. Liu, and Y. Liu, “A survey
munications Magazine, vol. 55, no. 12, pp. 119–125, 2017. of blockchain technology applied to smart cities: Research issues and
[4] H. Zhou, X. Ouyang, Z. Ren, J. Su, C. de Laat, and Z. Zhao, “A challenges,” IEEE Communications Surveys & Tutorials, vol. 21, no. 3,
blockchain based witness model for trustworthy cloud service level pp. 2794–2830, 2019.
agreement enforcement,” in IEEE Conf. on Computer Communications. [30] M. Samaniego and R. Deters, “Blockchain as a service for IoT: cloud
IEEE, 2019, pp. 1567–1575. versus fog,” in IEEE Int’l Conf. on IoT. IEEE, 2016, pp. 433–436.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 19

[31] W. Viriyasitavat, D. Li, Z. Bi, and A. Sapsomboon, “New blockchain- of the 17th IEEE/ACM international symposium on cluster, cloud and
based architecture for service interoperations in Internet of Things,” grid computing. IEEE Press, 2017, pp. 468–477.
IEEE TCSS, vol. 6, no. 4, pp. 739–748, 2019. [59] A. Ramachandran and M. Kantarcioglu, “Smartprovenance: a dis-
[32] Y. Chen, J. Gu, S. Chen, S. Huang, and X. Wang, “A full-spectrum tributed, blockchain based dataprovenance system,” in Proceedings of
blockchain-as-a-service for business collaboration,” in IEEE Int’l Conf. the Eighth ACM Conf. on Data and Application Security and Privacy.
on Web Services. Milan, Italy: IEEE, 2019, pp. 219–223. ACM, 2018, pp. 35–42.
[33] Q. Lu, X. Xu, Y. Liu, I. Weber, L. Zhu, and W. Zhang, “uBaaS: A [60] Q. Xia, E. Sifah, K. Asamoah, J. Gao, X. Du, and M. Guizani, “Med-
unified blockchain as a service platform,” Future Generation Computer share: Trust-less medical data sharing among cloud service providers
Systems, vol. 101, pp. 564–575, 2019. via blockchain,” IEEE Access, vol. 5, pp. 14 757–14 767, 2017.
[34] W. Zheng, Z. Zheng, X. Chen, K. Dai, P. Li, and R. Chen, “Nutbaas: A [61] J. Gao, K. Asamoah, E. Sifah, A. Smahi, Q. Xia, H. Xia, X. Zhang,
blockchain-as-a-service platform,” IEEE Access, vol. 7, pp. 134 422– and G. Dong, “Gridmonitoring: Secured sovereign blockchain based
134 433, 2019. monitoring on smart grid,” IEEE Access, vol. 6, pp. 9917–9925, 2018.
[35] A. Azaria, A. Ekblaw, T. Vieira, and A. Lippman, “Medrec: Using [62] R. Neisse, G. Steri, and I. Nai-Fovino, “A blockchain-based approach
blockchain for medical data access and permission management,” in for data accountability and provenance tracking,” in The 12th Int’l Conf.
The 2nd Int’l Conf. on Open and Big Data. IEEE, 2016, pp. 25–30. on Avail., Reli. and Sec. ACM, 2017, p. 14.
[36] C. Esposito, A. D. Santis, G. Tortora, H. Chang, and K. Choo, [63] A. Al-Mamun, T. Li, M. Sadoghi, and D. Zhao, “In-memory
“Blockchain: A panacea for healthcare cloud-based data security and blockchain: Toward efficient and trustworthy data provenance for hpc
privacy?” IEEE Cloud Computing, vol. 5, no. 1, pp. 31–37, 2018. systems,” in IEEE Int’l Conf. Big Data. IEEE, 2018, pp. 3808–3813.
[37] X. Liu, A. Liu, T. Wang, K. Ota, M. Dong et al., “Adaptive data [64] D. K. D. Tosh, S. Shetty, X. Liang, C. Kamhoua, and L. Njilla, “Con-
and verified message disjoint security routing for gathering big data in sensus protocols for blockchain-based data provenance: Challenges and
energy harvesting networks,” JPDC, vol. 135, pp. 140–155, 2020. opportunities,” in IEEE 8th Ann’l Ubiquitous Computing, Electronics
[38] J. Singh and J. Michels, “Blockchain as a service (baas): Providers and and Mobile Communication Conf. IEEE, 2017, pp. 469–474.
trust,” in IEEE European Sym. on Security and Privacy Workshops. [65] A. Maw, S. Adepu, and A. Mathur, “ICS-BlockOpS: Blockchain for
London, United Kingdom: IEEE, 2018, pp. 67–74. operational data security in industrial control system,” Pervasive and
[39] IBM, “IBM developer: Blockchain,” Mobile Computing, vol. 59, p. 101048, 2019.
https://developer.ibm.com/technologies/blockchain/, 2019. [66] W. Oliveira, D. Oliveira, and V. Braganholo, “Provenance analytics for
[40] Oracle, “Oracle blockchain blog,” https://blogs.oracle.com/blockchain/ workflow-based computational experiments: A survey,” ACM Comput-
blockchain-use-cases, 2019. ing Surveys, vol. 51, no. 3, p. 53, 2018.
[41] Microsoft, “Microsoft azure,” https://azure.microsoft.com. [67] M. Interlandi et al., “Titian: Data provenance support in spark,”
[42] Amazon, “Blockchain on AWS,” https://amazonaws-china.com Proceedings of the VLDB Endowment, vol. 9, no. 3, pp. 216–227, 2015.
/cn/blockchain. [68] P. Buneman and W. Tan, “Data provenance: What next?” ACM SIG-
[43] IBM, “IBM blockchain,” https://www.ibm.com/cloud- MOD Record, vol. 47, no. 3, pp. 5–16, 2019.
computing/cn/zh/newplatform/blockchain/offerings. [69] F. Zafar et al., “Trustworthy data: A survey, taxonomy and future trends
[44] M. Samaniego and R. Deters, “Using blockchain to push software- of secure provenance schemes,” JNCA, vol. 94, pp. 50–68, 2017.
defined IoT components onto edge hosts,” in Int’l Conf. on BDAWT.
[70] Z. Xiao and Y. Xiao, “Security and privacy in cloud computing,” IEEE
ACM, 2016, p. 58.
Communications Surveys & Tutorials, vol. 15, no. 2, pp. 843–859,
[45] A. Kosba, A. Miller, E. Shi, Z. Wen, and C. Papamanthou, “Hawk:
2013.
The blockchain model of cryptography and privacy-preserving smart
[71] P. Ivie and D. Thain, “Reproducibility in scientific computing,” ACM
contracts,” in IEEE Sym. on S&P. IEEE, 2016, pp. 839–858.
Computing Surveys, vol. 51, no. 3, p. 63, 2018.
[46] C. Melo, J. Dantas, D. Oliveira, I. Fé et al., “Dependability evaluation
[72] S. Zawoad, R. Hasan, and K. Islam, “SECProv: trustworthy and
of a blockchain-as-a-service environment,” in IEEE Sym. on Comp. and
efficient provenance management in the cloud,” in IEEE Conf. on
Communi. IEEE, 2018, pp. 00 909–00 914.
Computer Communications. IEEE, 2018, pp. 1241–1249.
[47] J. Lee, “BIDaaS: Blockchain based ID as a service,” IEEE Access,
vol. 6, pp. 2274–2278, 2017. [73] S. Ali, J. Wang, M. Bhuiyan, and H. Jiang, “Secure data provenance
[48] Y. Xu, J. Ren, Y. Zhang, C. Zhang, B. Shen, and Y. Zhang, “Blockchain in cloud-centric internet of things via blockchain smart contracts,” in
empowered arbitrable data auditing scheme for network storage as a SmartWorld. Guangzhou, China: IEEE, 2018, pp. 991–998.
service,” IEEE TSC, vol. PP, no. 99, p. 1, 2019. [74] Muniswamy-Reddy and M. Seltzer, “Provenance as first class cloud
[49] H. Chen and L. Zhang, “FBaaS: Functional blockchain as a service,” data,” ACM SIGOPS Operating Systems Review, vol. 43, no. 4, pp.
in Int’l Conf. on Blockchain. Springer, 2018, pp. 243–250. 11–16, 2010.
[50] Q. Lu, X. Xu, Y. Liu, and W. Zhang, “Design pattern as a service [75] T. McGhin, K. Choo, C. Liu, and D. He, “Blockchain in healthcare
for blockchain applications,” in IEEE Int’l Conf. on Data Mining applications: Research challenges and opportunities,” JNCA, vol. PP,
Workshops. IEEE, 2018, pp. 128–135. no. 99, p. 1, 2019.
[51] H. Zhang, E. Deng, H. Zhu, and Z. Cao, “Smart contract for secure [76] K. Gai, Y. Wu, L. Zhu, Z. Zhang, and M. Qiu, “Differential privacy-
billing in ride-hailing service via blockchain,” Peer-to-Peer Networking based blockchain for industrial Internet of Things,” IEEE Transactions
and Applications, vol. 12, no. 5, pp. 1346–1357, 2019. on Industrial Informatics, vol. PP, no. 99, p. 1, 2019.
[52] A. Karinsalo and K. Halunen, “Smart contracts for a mobility-as-a- [77] G. Liang et al., “Distributed blockchain-based data protection frame-
service ecosystem,” in Int’l Conf. on QRS-C. Lisbon, Portugal: IEEE, work for modern power systems against cyber attacks,” IEEE TSG,
2018, pp. 135–138. vol. 10, no. 3, pp. 3162–3173, 2018.
[53] P. Marandi, C. Gkantsidis, F. Junqueira, and D. Narayanan, “Filo: [78] Y. Chen, L. Wang, and S. Wang, “Stochastic blockchain for IoT data
consolidated consensus as a cloud service,” in {USENIX} Annual integrity,” IEEE TNSE, vol. PP, no. 99, p. 1, 2018.
Technical Conference, Denver, CO, USA, 2016, pp. 237–249. [79] J. Lopez and J. Rubio, “Access control for cyber-physical systems
[54] G. Kumar, R. Saha, M. Rai, R. Thomas, and T. Kim, “Proof-of- interconnected to the cloud,” Computer Networks, vol. 134, pp. 46–
work consensus approach in blockchain technology for cloud and fog 54, 2018.
computing using maximization-factorization statistics,” IEEE IOT J, [80] M. Qiu et al., “Proactive user-centric secure data scheme using
vol. PP, no. 99, 2019. attribute-based semantic access controls for mobile clouds in financial
[55] Z. Zhou, B. Wang, M. Dong, and K. Ota, “Secure and efficient industry,” Future Generation Comp. Syst., vol. 80, pp. 421–429, 2018.
vehicle-to-grid energy trading in cyber physical systems: Integration [81] S. Rane and A. Dixit, “BlockSLaaS: Blockchain assisted secure
of blockchain and edge computing,” IEEE Trans. on Systems, Man, Logging-as-a-Service for cloud forensics,” in Int’l Conf. on Security
and Cybernetics: Systems, vol. PP, no. 99, p. 1, 2019. & Privacy. Springer, 2019, pp. 77–88.
[56] D. Reinsel, J. Gantz, and J. Rydning, “The digitization of the world: [82] Y. Zhang, S. Kasahara, Y. Shen, X. Jiang, and J. Wan, “Smart contract-
from edge to core,” 2018. based access control for the internet of things,” IEEE IOT J, vol. 6,
[57] M. Westerkamp, F. Victor, and A. Kupper, “Tracing manufacturing no. 2, pp. 1594–1605, 2018.
processes using blockchain-based token compositions,” Digital Com- [83] S. Rouhani and R. Deters, “Blockchain based access control systems:
munications and Networks, vol. PP, no. 99, p. 1, 2019. State of the art and challenges,” arXiv:1908.08503, 2019.
[58] X. Liang, S. Shetty, D. Tosh, C. Kamhoua, K. Kwiat, and L. Njilla, [84] I. Sukhodolskiy and S. Zapechnikov, “A blockchain-based access
“Provchain: A blockchain-based data provenance architecture in cloud control system for cloud storage,” in IEEE EIConRus. Moscow,
environment with enhanced privacy and availability,” in Proceedings Russia: IEEE, 2018, pp. 1575–1578.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 20

[85] S. Wang, X. Wang, and Y. Zhang, “A secure cloud storage framework [110] Y. Zhang, R. Deng, X. Liu, and D. Zheng, “Outsourcing service fair
with access control based on blockchain,” IEEE Access, vol. 7, pp. payment based on blockchain and its applications in cloud computing,”
112 713–112 725, 2019. IEEE Trans. on Services Computing, 2018.
[86] D. Nguyen, P. Pathirana, M. Ding, and A. Seneviratne, “Blockchain for [111] Y. Zhang, R. H. Deng, J. Shu, K. Yang, and D. Zheng, “Tkse: trust-
secure EHRs sharing of mobile cloud based e-health systems,” IEEE worthy keyword search over encrypted data with two-side verifiability
access, vol. 7, pp. 66 792–66 806, 2019. via blockchain,” IEEE Access, vol. 6, pp. 31 077–31 087, 2018.
[87] G. Zyskind and O. Nathan, “Decentralizing privacy: Using blockchain [112] C. Cai, X. Yuan, and C. Wang, “Towards trustworthy and private
to protect personal data,” in IEEE Security and Privacy Workshops. keyword search in encrypted decentralized storage,” in IEEE Int’l Conf.
IEEE, 2015, pp. 180–184. on Communications. IEEE, 2017, pp. 1–7.
[88] G. Zyskind, O. Nathan, and A. Pentland, “Enigma: Decentral- [113] S. Hu, C. Cai, Q. Wang, C. Wang, X. Luo, and K. Ren, “Searching an
ized computation platform with guaranteed privacy,” arXiv preprint encrypted cloud meets blockchain: A decentralized, reliable and fair
arXiv:1506.03471, 2015. realization,” in IEEE Conf. on Computer Communications. IEEE,
[89] D. Nguyen, J. Park, and R. Sandhu, “Dependency path patterns as the 2018, pp. 792–800.
foundation of access control in provenance-aware systems.” in TaPP, [114] A. Zhang and X. Lin, “Towards secure and privacy-preserving data
2012. sharing in e-health systems via consortium blockchain,” Journal of
[90] Q. Xia, E. Sifah, A. Smahi, S. Amofa, and X. Zhang, “Bbds: medical systems, vol. 42, no. 8, p. 140, 2018.
Blockchain-based data sharing for electronic medical records in cloud [115] F. Han, J. Qin, and J. Hu, “Secure searches in the cloud: A survey,”
environments,” Information, vol. 8, no. 2, p. 44, 2017. Future Generation Computer Systems, vol. 62, pp. 66–75, 2016.
[91] M. Rahman, M. Hossain, G. Loukas, E. Hassanain, S. Rahman, M. Al- [116] A. Choudhuri et al., “Fairness in an unfair world: Fair multiparty
hamid, and M. Guizani, “Blockchain-based mobile edge computing computation from public bulletin boards,” in SIGSAC Conf. on CCS.
framework for secure therapy applications,” IEEE Access, vol. 6, pp. Dallas, USA: ACM, 2017, pp. 719–728.
72 469–72 478, 2018. [117] L. Luu, J. Teutsch, R. Kulkarni, and P. Saxena, “Demystifying incen-
[92] A. Ouaddah, A. A. Elkalam, and A. A. Ouahman, “Fairaccess: a new tives in the consensus computer,” in 22nd ACM SIGSAC Conf. on CCS.
blockchain-based access control framework for the internet of things,” ACM, 2015, pp. 706–719.
Secur. & Communi. Netw., vol. 9, no. 18, pp. 5943–5964, 2016. [118] IDC, “Idc report,” https://www.emc.com/collateral/analyst-reports/idc-
[93] O. Novo, “Blockchain meets iot: An architecture for scalable access digital-universe-are-you-ready.pdf.
management in iot,” IEEE IOT J., vol. 5, no. 2, pp. 1184–1195, 2018. [119] Y. Shin, D. Koog, and J. Hur, “A survey of secure data deduplication
[94] S. Wang, Y. Zhang, and Y. Zhang, “A blockchain-based framework for schemes for cloud storage systems,” ACM Computing Surveys, vol. 49,
data sharing with fine-grained access control in decentralized storage no. 4, p. 74, 2017.
systems,” IEEE Access, vol. 6, pp. 38 437–38 450, 2018. [120] J. Douceur, A. Adya, W. Bolosky, P. Simon, and M. Theimer, “Reclaim-
[95] M. S. Ferdous, A. Margheri, F. Paci, M. Yang, and V. Sassone, ing space from duplicate files in a serverless distributed file system,”
“Decentralised runtime monitoring for access control systems in cloud in The 22nd Int’l Conf. DCS. IEEE, 2002, pp. 617–624.
federations,” in IEEE 37th Int’l Conf. on Distributed Computing [121] M. Bellare, S. Keelveedhi, and T. Ristenpart, “Message-locked encryp-
Systems. IEEE, 2017, pp. 2632–2633. tion and secure deduplication,” in Ann’l Int’l Conf. on the Theor. and
App. of Crypt. Techni. Springer, 2013, pp. 296–312.
[96] S. Alansari, F. Paci, and V. Sassone, “A distributed access control
[122] S. Keelveedhi, M. Bellare, and T. Ristenpart, “Dupless: server-aided
system for cloud federations,” in IEEE 37th Int’l Conf. on Distributed
encryption for deduplicated storage,” in Presented as part of the 22nd
Computing Systems. IEEE, 2017, pp. 2131–2136.
{USENIX} Security Sym., 2013, pp. 179–194.
[97] F. Schiavo, V. Sassone, L. Nicoletti, A. Reiter, and B. Suzic, “Faas:
[123] D. Harnik, B. Pinkas, and A. Shulman-Peleg, “Side channels in cloud
Federation-as-a-service: The sunfish cloud federation solution,” in
services: Deduplication in cloud storage,” IEEE Security & Privacy,
FaaS: Federation-as-a-Service: The SUNFISH Cloud Federation So-
vol. 8, no. 6, pp. 40–47, 2010.
lution, 2016.
[124] J. Yuan and S. Yu, “Secure and constant cost public cloud storage
[98] D. Song, D. Wagner, and A. Perrig, “Practical techniques for searches
auditing with deduplication,” in IEEE Conf. on Communications and
on encrypted data,” in Proceeding 2000 IEEE Sym. on Security and
Network Security. IEEE, 2013, pp. 145–153.
Privacy. S&P 2000. IEEE, 2000, pp. 44–55.
[125] Y. Li, L. Zhu, M. Shen, F. Gao, B. Zheng, X. Du, S. Liu, and
[99] D. Boneh, G. D. Crescenzo, R. Ostrovsky, and G. Persiano, “Public S. Yin, “Cloudshare: Towards a cost-efficient and privacy-preserving
key encryption with keyword search,” in Ann’l Int’l Conf. on the Theor. alliance cloud using permissioned blockchains,” in Int’l Conf. on
and App. of Crypt. Techni. Springer, 2004, pp. 506–522. Mobile Networks and Management. Springer, 2017, pp. 339–352.
[100] P. Golle, J. Staddon, and B. Waters, “Secure conjunctive keyword [126] H. Liu, Y. Zhang, and T. Yang, “Blockchain-enabled security in electric
search over encrypted data,” in Int’l Conf. on Applied Cryptography vehicles cloud and edge computing,” IEEE Network, vol. 32, no. 3, pp.
and Network Security. Springer, 2004, pp. 31–45. 78–83, 2018.
[101] J. Li, Q. Wang, C. Wang, N. Cao, K. Ren, and W. Lou, “Fuzzy keyword [127] J. Li, J. Wu, L. Chen, and J. Li, “Deduplication with blockchain for
search over encrypted data in cloud computing,” in 2010 Proceedings secure cloud storage,” in CCF Conf. on Big Data. Springer, 2018,
IEEE INFOCOM. IEEE, 2010, pp. 1–5. pp. 558–570.
[102] G. Poh, J. Chin, W. Yau, K. Choo, and S. M. Mohamad, “Search- [128] C. Liu, Q. Lin, and S. Wen, “Blockchain-enabled data collection and
able symmetric encryption: designs and challenges,” ACM Computing sharing for industrial iot with deep reinforcement learning,” IEEE
Surveys, vol. 50, no. 3, p. 40, 2017. Trans. on Industrial Informatics, 2018.
[103] X. Chen, J. Li, X. Huang, J. Ma, and W. Lou, “New publicly verifiable [129] K. Gai et al., “Dynamic energy-aware cloudlet-based mobile cloud
databases with efficient updates,” IEEE Trans. on Dependable and computing model for green computing,” JNCA, vol. 59, pp. 46–54,
Secure Computing, vol. 12, no. 5, pp. 546–556, 2015. 2016.
[104] Z. Fu, J. Shu, X. Sun, and N. Linge, “Smart cloud search services: [130] K. Gai, M. Qiu, and H. Zhao, “Energy-aware task assignment for
verifiable keyword-based semantic search over encrypted cloud data,” mobile cyber-enabled applications in heterogeneous cloud computing,”
IEEE Trans. on ConsR Elec., vol. 60, no. 4, pp. 762–770, 2014. JPDC, vol. 111, pp. 126–135, 2018.
[105] R. Cheng, J. Yan, C. Guan, F. Zhang, and K. Ren, “Verifiable searchable [131] X. Qiu, L. Liu, W. Chen, Z. Hong, and Z. Zheng, “Online deep
symmetric encryption from indistinguishability obfuscation,” in The reinforcement learning for computation offloading in blockchain-
10th ACM Sym. on ICCS. Singapore: ACM, 2015, pp. 621–626. empowered mobile edge computing,” IEEE Transactions on Vehicular
[106] L. Chen et al., “Blockchain based searchable encryption for electronic Technology, vol. 68, no. 8, pp. 8050–8062, 2019.
health record sharing,” FGCS, vol. 95, pp. 420–429, 2019. [132] W. Chen, Z. Zhang, Z. Hong, C. Chen, J. Wu et al., “Cooperative and
[107] Y. Zhang, C. Xu, J. Ni, H. Li, and X. Shen, “Blockchain-assisted distributed computation offloading for blockchain-empowered indus-
public-key encryption with keyword search against keyword guessing trial Internet of Things,” IEEE IOT J, vol. 6, no. 5, pp. 8433–8446,
attacks for cloud storage,” IEEE TCC, vol. PP, no. 99, p. 1, 2019. 2019.
[108] J. Niu, X. Li, J. Gao, and Y. Han, “Blockchain-based anti-key-leakage [133] Z. Zhang, Z. Hong, W. Chen, Z. Zheng, and X. Chen, “Joint compu-
key aggregation searchable encryption for IoT,” IEEE Internet of tation offloading and coin loaning for blockchain-empowered mobile-
Things Journal, vol. PP, no. 99, p. 1, 2019. edge computing,” IEEE IOT J, vol. 6, no. 6, pp. 9934–9950, 2019.
[109] P. Jiang, F. Guo, K. Liang, J. Lai, and Q. Wen, “Searchain: Blockchain- [134] J. Xu, K. Ota, M. Dong, A. Liu, and Q. Li, “SIoTFog: Byzantine-
based private keyword search in decentralized storage,” Future Gener- resilient iot fog networking,” Frontiers of Information Technology &
ation Computer Systems, 2017. Electronic Engineering, vol. 19, no. 12, pp. 1546–1557, 2018.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 21

[135] Z. Zhan, X. Liu, Y. Gong, J. Zhang, H. Chung, and Y. Li, “Cloud [160] C. Qiu, H. Yao, C. Jiang, S. Guo, and F. Xu, “Cloud computing assisted
computing resource scheduling and a survey of its evolutionary ap- blockchain-enabled internet of things,” IEEE TCC, vol. PP, no. 99, p. 1,
proaches,” ACM Computing Surveys, vol. 47, no. 4, p. 63, 2015. 2019.
[136] S. Kumar and P. Balasubramanie, “Dynamic scheduling for cloud [161] Z. Xiong, Y. Zhang, D. Niyato, P. Wang, and Z. Han, “When mobile
reliability using transportation problem,” Journal of Computer Science, blockchain meets edge computing,” IEEE Communications Magazine,
vol. 8, no. 10, 2012. vol. 56, no. 8, pp. 33–39, 2018.
[137] Y. Li, Z. Zhan, Y. Gong, W. Chen, J. Zhang, and Y. Li, “Differential [162] Z. Xiong, S. Feng, D. Niyato, P. Wang, and Z. Han, “Optimal pricing-
evolution with an evolution path: A DEEP evolutionary algorithm,” based edge computing resource management in mobile blockchain,” in
IEEE Trans. on Cybernetics, vol. 45, no. 9, pp. 1798–1810, 2015. IEEE Int’l Conf. on Communi. IEEE, 2018, pp. 1–6.
[138] B. Ghosh, S. Addya, A. Satpathy, S. Ghosh, and S. Chakraborty, “To- [163] P. Fairley, “Blockchain world-feeding the blockchain beast if bitcoin
wards a democratic federation for infrastructure service provisioning,” ever does go mainstream, the electricity needed to sustain it will be
in Int’l Conf. on SCC. Milan, Italy: IEEE, 2019, pp. 162–166. enormous,” IEEE Spectrum, vol. 54, no. 10, pp. 36–59, 2017.
[139] M. Yang, A. Margheri, R. Hu, and V. Sassone, “Differentially private [164] I. Magaki, M. Khazraee, L. Gutierrez, and B. M. Taylor, “Asic clouds:
data sharing in a cloud federation with blockchain,” IEEE Cloud specializing the datacenter,” in 2016 ACM/IEEE 43rd Ann’l Int’l Symp.
Computing, vol. 5, no. 6, pp. 69–79, 2018. on Comp. Archi. IEEE, 2016, pp. 178–190.
[140] D. Chatzopoulos, M. Ahmadi, S. Kosta, and P. Hui, “Flopcoin: A [165] N. Amit, M. Wei, and C. Tu, “Extreme datacenter specialization
cryptocurrency for computation offloading,” IEEE Trans. on Mobile for planet-scale computing: Asic clouds,” ACM SIGOPS Operating
Computing, vol. 17, no. 5, pp. 1062–1075, 2018. Systems Review, vol. 51, no. 1, pp. 96–108, 2018.
[141] C. Xu, K. Wang, and M. Guo, “Intelligent resource management in [166] M. Taylor, “The evolution of bitcoin hardware,” Computer, vol. 50,
blockchain-based cloud datacenters,” IEEE Cloud Computing, vol. 4, no. 9, pp. 58–66, 2017.
no. 6, pp. 50–59, 2017. [167] J. Liu, W. Li, G. Karame, and N. Asokan, “Scalable byzantine con-
[142] Y. Zhang, R. Deng, X. Liu, and D. Zheng, “Blockchain based efficient sensus via hardware-assisted secret sharing,” IEEE Trans. on Comp.,
and robust fair payment for outsourcing services in cloud computing,” vol. 68, no. 1, pp. 139–151, 2018.
Information Sciences, vol. 462, pp. 262–277, 2018. [168] N. Courtois, M. Grajek, and R. Naik, “Optimizing sha256 in bit-
[143] S. Nakamoto et al., “Bitcoin: A peer-to-peer electronic cash system,” coin mining,” in Int’l Conf. on Cryptography and Security Systems.
2008. Springer, 2014, pp. 131–144.
[144] K. Gai, Y. Wu, L. Zhu, L. Xu, and Y. Zhang, “Permissioned blockchain [169] M. Taylor, “Bitcoin and the age of bespoke silicon,” in 2013 Int’l Conf.
and edge computing empowered privacy-preserving smart grid net- on CASES. IEEE, 2013, pp. 1–10.
works,” IEEE IOT J, vol. 6, no. 5, pp. 7992–8004, 2019. [170] C. Lee, “Litecoin,” 2011.
[145] G. Ramezan and C. Leung, “A blockchain-based contractual routing [171] C. Percival, “Stronger key derivation via sequential memory-hard
protocol for the internet of things using smart contracts,” Wireless functions.”
Communications and Mobile Computing, vol. 2018, 2018. [172] B. Ekbote, V. Hire, P. Mahajan, and J. Sisodia, “Blockchain based
remittances and mining using cuda,” in 2017 Int’l Conf. On Smart
[146] M. Liu, F. Yu, Y. Teng, V. Leung, and M. Song, “Joint computation
Technologies For Smart Nation. IEEE, 2017, pp. 908–911.
offloading and content caching for wireless blockchain networks,” in
[173] M. Skach, M. Arora, C. Hsu, Q. Li, D. Tullsen, L. Tang, and
IEEE INFOCOM WKSHPS. IEEE, 2018, pp. 517–522.
J. Mars, “Thermal time shifting: Leveraging phase change materials to
[147] W. Shi, Q. Ling, K. Yuan, G. Wu, and W. Yin, “On the linear
reduce cooling costs in warehouse-scale computers,” in ACM Sigarch
convergence of the admm in decentralized consensus optimization,”
Computer Architecture News, vol. 43, no. 3. ACM, 2015, pp. 439–449.
IEEE Trans. on Signal Processing, vol. 62, no. 7, pp. 1750–1761, 2014.
[174] A. Kampl, “Bitcoin 2-phase immersion cooling and the implications
[148] M. Liu, F. Yu, Y. Teng, V. Leung, and M. Song, “Computation for high performance computing,” Electronics Cooling, vol. 20, no. 1,
offloading and content caching in wireless blockchain networks with 2014.
mobile edge computing,” IEEE Trans. on Vehicular Technology, vol. 67,
[175] S. Ziegenbalg, “Btcminer - open source bitcoin miner,” https://open
no. 11, pp. 11 008–11 021, 2018.
cores.org/projects/btcminer.
[149] M. Liu, F. Yu, Y. Teng, V. Leung, and M. Song, “Distributed resource [176] butterflylab, “Butterflylab,” https://butterflylabs.com/category/bitcoin/.
allocation in blockchain-based video streaming systems with mobile [177] Avalon, “Avalon,” https://canaan.io/.
edge computing,” IEEE Trans. on Wireless Communications, vol. 18,
[178] Bitmain, “Bitmain,” https://www.bitmain.com/.
no. 1, pp. 695–708, 2019.
[179] Y. Qian, Z. Liu, J. Yang, and Q. Wang, “A method of exchanging data
[150] Y. Wu, X. Chen, J. Shi, K. Ni, L. Qian, L. Huang, and K. Zhang, in smart city by blockchain,” in 20th Int’l Conf. on HPCC. Exeter,
“Optimal computational power allocation in multi-access mobile edge United Kingdom: IEEE, 2018, pp. 1344–1349.
computing for blockchain,” Sensors, vol. 18, no. 10, p. 3472, 2018.
[180] X. Zheng et al., “Blockchain-based personal health data sharing system
[151] Y. Wu, J. Shi, X. Chen, K. Ni, L. Qian, and K. Zhang, “Optimal using cloud storage,” in 20th Int’l Conf. Healthcom. Ostrava, Czech
multi-access computation offloading for mobile blockchain,” in IEEE Republic: IEEE, 2018, pp. 1–6.
Int’l Conf. on Communi. Sys. IEEE, 2019, pp. 198–203. [181] X. Liang et al., “Integrating blockchain for data sharing and collab-
[152] Y. Jiao, P. Wang, D. Niyato, and Z. Xiong, “Social welfare maxi- oration in mobile healthcare applications,” in 28th Anu’l Int’l Symp.
mization auction in edge computing resource allocation for mobile PIMRC. Montreal, QC, Canada: IEEE, 2017, pp. 1–5.
blockchain,” in IEEE Int’l Conf. on Communi. IEEE, 2018, pp. 1–6. [182] R. Khalil and A. Gervais, “Revive: Rebalancing off-blockchain pay-
[153] Y. Jiao, P. Wang, D. Niyato, and K. Suankaewmanee, “Auction mecha- ment networks,” in ACM SIGSAC CCS. ACM, 2017, pp. 439–453.
nisms in cloud/fog computing resource allocation for public blockchain [183] J. Guo, K. Gai, L. Zhu, and Z. Zhang, “An approach of secure two-way-
networks,” IEEE TPDS, vol. PP, no. 99, p. 1, 2019. pegged multi-sidechain,” in ICA3PP. Melbourne, Australia: Springer,
[154] Z. Li, Z. Yang, and S. Xie, “Computing resource trading for edge- 2019, pp. 551–564.
cloud-assisted internet of things,” IEEE Trans. on Industrial Info., 2019. [184] Y. Sun, R. Zhang, X. Wang, K. Gao, and L. Liu, “A decentralizing
[155] N. Luong, Z. Xiong, P. Wang, and D. Niyato, “Optimal auction for attribute-based signature for healthcare blockchain,” in 27th Int’l Conf.
edge computing resource management in mobile blockchain networks: on Computer Communi. and Networks. IEEE, 2018, pp. 1–9.
A deep learning approach,” in IEEE ICC. IEEE, 2018, pp. 1–6. [185] N. Rifi, E. Rachkidi, N. Agoulmine, and N. Taher, “Towards using
[156] Z. Xiong, S. Feng, W. Wang, D. Niyato, P. Wang, and blockchain technology for iot data access protection,” in IEEE 17th
Z. Han, “Cloud/fog computing resource management and pricing for Int’l Conf. on Ubiquitous Wireless Broadband. IEEE, 2017, pp. 1–5.
blockchain networks,” IEEE IOT J, 2018. [186] H. Shafagh, L. Burkhalter, A. Hithnawi, and S. Duquennoy, “Towards
[157] V. Krishna, Auction theory. Academic press, 2009. blockchain-based auditable storage and sharing of iot data,” in Cloud
[158] C. Xia, H. Chen, X. Liu, J. Wu, and L. Chen, “ETRA: efficient Computing Security Workshop. ACM, 2017, pp. 45–50.
three-stage resource allocation auction for mobile blockchain in edge [187] S. Ali, G. Wang, B. White, and R. Cottrell, “A blockchain-based
computing,” in 24th Int’l Conf. Par. & Dist. Syst. IEEE, 2018, pp. decentralized data storage and access framework for pinger,” in 17th
701–705. Int’l Conf. TrustCom. New York, USA: IEEE, 2018, pp. 1303–1308.
[159] J. Kang, R. Yu, X. Huang, S. Maharjan, Y. Zhang, and E. Hossain, [188] R. Chaudhary, A. Jindal, G. Aujla, S. Aggarwal, N. Kumar, and
“Enabling localized peer-to-peer electricity trading among plug-in K. Choo, “BEST: Blockchain-based secure energy trading in SDN-
hybrid electric vehicles using consortium blockchains,” IEEE Trans. enabled intelligent transportation system,” Computers & Security,
on Industrial Informatics, vol. 13, no. 6, pp. 3154–3164, 2017. vol. 85, pp. 288–299, 2019.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/COMST.2020.2989392, IEEE
Communications Surveys & Tutorials
IEEE COMMUNICATIONS SURVEYS & TUTORIALS 22

[189] P. Sharma, S. Singh, Y. Jeong, and J. Park, “Distblocknet: A distributed Shui Yu is a Professor of School of Computer
blockchains-based secure SDN architecture for IoT networks,” IEEE Science, University of Technology Sydney, Aus-
Communications Magazine, vol. 55, no. 9, pp. 78–85, 2017. tralia. Dr Yu’s research interest includes Big Data,
[190] X. Ling, J. Wang, T. Bouchoucha, B. Levy, and Z. Ding, “Blockchain Security and Privacy, Networking, and Mathematical
radio access network (B-RAN): Towards decentralized secure radio Modelling. He has published three monographs and
access paradigm,” IEEE Access, vol. 7, pp. 9714–9723, 2019. edited two books, more than 350 technical papers,
including top journals and top conferences, such
as IEEE TPDS, TC, TIFS, TMC, TKDE, TETC,
ToN, and INFOCOM. Dr Yu initiated the research
field of networking for big data in 2013. His h-
index is 43. He is currently serving a number of
prestigious editorial boards, including IEEE Communications Surveys and
Tutorials (Area Editor), IEEE Communications Magazine, and IEEE Internet
Keke Gai [SM20’-M17’-S13’] received the B.Eng. of Things Journal. He is a Senior Member of IEEE, a member of AAAS and
degree majored in automation, from Nanjing Uni- ACM, and a Distinguished Lecturer of IEEE Communication Society.
versity of Science and Technology, Nanjing, China,
in 2004, the M.E.T. (Master’s of Educational Tech-
nology) degree in educational technology from the
University of British Columbia, Vancouver, BC,
Canada, in 2010, the MBA degree in business Ad-
ministration, in 2009, M.S. degree in information
technology, in 2014, from the Lawrence Technolog-
ical University, Southfield, MI, USA, and the Ph.D.
degree in computer science from Pace University,
New York, NY, USA.
He is currently an Associate Professor at the School of Computer Science
and Technology, Beijing Institute of Technology, Beijing, China. He has
published 3 books and more than 110 peer-reviewed journal/conference
papers, including 8 ESI Highly Cited Papers. He has been granted 5 IEEE Best
Paper awards (TrustCom 18’, HPCC 18’, etc.) and 2 IEEE Best Student Paper
awards (HPCC 16’, etc.) in recent 5 years. His research interests include cyber
security, blockchain, edge computing, cloud computing, and reinforcement
learning. He currently is a Senior Member of IEEE.

Jinnan Guo is currently pursuing his bachelor’s


degree in Electronic Science and Technology at
School of Information and Electronics, Beijing In-
stitute of Technology. His research interests include
blockchain, cloud computing, and network security.

Liehuang Zhu received his Ph.D. degree in com-


puter science from Beijing Institute of Technology,
Beijing, China, in 2004, the M.E. (Master of Engi-
neering) degree and B.E. (Bachelor of Engineering)
degree from Wuhan University, Wuhan, China, in
2001 and 1998, respectively.
He is currently a professor at School of Computer
Science & Technology, Beijing Institute of Technol-
ogy, Beijing, China. He has published more than
200 peer-reviewed journal or conference papers, in-
cluding 10+ IEEE/ACM Transactions papers (IEEE
TIFS, IEEE TII, IEEE TVT, IEEE TSG, Information Sciences, IEEE Network,
Computer & Security, etc.). He has been granted a number of IEEE Best Paper
Awards, including IWQoS 17’, TrustCom 18’. His research interests include
security protocol analysis and design, wireless sensor networks, and cloud
computing.

1553-877X (c) 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: University of Canberra. Downloaded on April 30,2020 at 13:55:54 UTC from IEEE Xplore. Restrictions apply.

You might also like