You are on page 1of 5

Proceedings of the Second International Conference on Artificial Intelligence and Smart Energy (ICAIS-2022)

IEEE Xplore Part Number: CFP22OAB-ART; ISBN: 978-1-6654-0052-7

Design of High-Confidence Embedded Operating


System based on Artificial Intelligence and Smart
Chips
2022 Second International Conference on Artificial Intelligence and Smart Energy (ICAIS) | 978-1-6654-0052-7/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICAIS53314.2022.9742917

Qinmin Ma
School of Artificial Intelligence, Shenzhen Polytechnic, Shenzhen, Guangdong, 518055, China
Maqinminshenzhencn1211@gmail.com

Abstract— Design of the high-confidence embedded operating (2) Trusted execution environ ment refers to a trusted,
system based on artificial intelligence and smart chips is studied isolated and independent execution environ ment on the device
in this paper. The cooperative physical layer security system is independent of the untrusted operating system. It provides a
regarded as a state machine. Relay nodes with untrusted secure and confidential space for private data and sensitive
behavior will affect the physical layer security of the system, and computing in the untrusted environment. Its security is usually
the system tries to prevent the untrusted behavior of relay nodes. guaranteed by hardware related mechanisms.
While implementing public verification, it realizes the protection
of data privacy. The third party can directly verify the data The application of game theory to the research of physical
holding of the data stored in the cloud without verification by the layer security has been a research hotspot in recent years. The
user, and in the process of system expansion and growth, main application of game theory to the selection o f the relay
software can ensure vigorous vitality. For the verification, the nodes is that the behavior type of the node is known. In the
smart chips are combined for the systematic implementations. figure 1, the initial structure has been denoted.
The experimental results have shown the satisfactory results.

Keywords— High-confidence model; embedded operating


system; artificial intelligence; smart chips

I. INT RODUCT ION


Different fro m the traditional cloud co mputing model, the
edge computing model adds a layer of netwo rk edge nodes
between the data source and the cloud computing center. Fig. 1. T he High-confidence Embedded Operating Syst em Structure
These network edge nodes can not only perform t wo -way In the designed model, the high-confidence scenarios will
data transmission with the cloud co mputing center, but also be combined. The consensus protocol of the core blockchain
efficiently perform some co mputing tasks with high real-t ime requires each node in the blockchain network to process each
requirements near the data source. In the early wo rk, the block in the blockchain, so that all nodes in the whole system
automatic conversion of code mod ification mostly adopted the can exchange data freely and safely in distrusted environment.
method of the manual feature engineering, that is, manually
proposed rules to limit and explain the adaptation conditions, However, in the practical applicat ions, in a distributed
modification process and modification results of the general environment, due to the node's own energy constraints or the
modification scheme in the modification example. control of malicious nodes, node behaviors not only have
cooperative behaviors, but also selfish, malicious blocking and
However, these rules are often based on researchers' expert other behaviors may be possible.
knowledge of specific languages, and it takes a lot of time and
energy to summarize and refine. A majo r feature of co mputing These untrustworthy behaviors greatly affect the security
is distribution on demand, and the blockchain itself is a performance of the system. High scalability represents a kind
resource, and there is a demand for on-demand supply. of elasticity. In the proces s of system expansion and growth,
software can ensure vigorous vitality. With few changes or
It is an integral part of the cloud co mputing, so cloud even just the addition of the hardware equip ment, the linear
computing technology and the blockchain technology can be growth of the processing capacity of the whole system can be
integrated with each other, hence, the core ideas are listed. realized, and high throughput and low latency performance
can be achieved, in the next sections, the details of the model
(1) In recent years, the PDP schemes are still a research
hotspot, but most of them are limited to a core single private will be modelled and analyzed.
verification or public verification, wh ich has great limitations.
Although the trustworthiness of the server is well verified, it
ignores whether the user is trustworthy research.

978-1-6654-0052-7/22/$31.00 ©2022 IEEE 58

Authorized licensed use limited to: Indian Institute of Technology- Goa. Downloaded on September 16,2022 at 11:13:35 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Second International Conference on Artificial Intelligence and Smart Energy (ICAIS-2022)
IEEE Xplore Part Number: CFP22OAB-ART; ISBN: 978-1-6654-0052-7

II. T HE PROPOSED MET HODOLOGY which extends the public block The usefulness of the chain
allo ws interaction to occur when needed, but it can still be
A. The High-confidence Embedded System
traced back to a reliab le base layer. In real life, due to profit-
The scalability limitation caused by the consensus protocol driven, malicious users are real. Therefore, when verify ing the
of the blockchain makes the transaction throughput of the possession of data in the cloud, it is necessary to verify the
blockchain low. Whether the transaction throughput can be credibility of the cloud service provider and the cloud user at
then improved is the key issue whether the blockchain can be the same time. Although the existing studies have proposed
widely used. On the one hand, given a modification examp le, models based on public verificat ion, private verification, and
the method parses the code before and after modification into the hybrid verificat ion, they are all aimed at the cloud service
an abstract syntax tree, and finds the correspondence between providers for credib ility verification without considering the
its nodes. Based on the correspondence between nodes, the credibility o f the users. Hence, for the better understanding,
method uses graphs to represent the given core sample code the figure 3 shows the system model.
modification to then enhance the model's ability to captu re the
structural info rmation in the code modification; on the other
hand, the method is effective through the graph convolutional
network and Transformer architecture co mb ining with each
other to enhance the model's ability to capture the dependency
informat ion between codes, especially long-range dependency
relationship. In the figure 2, the structure is defined.

Fig. 3. T he High-confidence Embedded System Model

B. The Artificial Intelligence Embedded Model


It is the demanding requirements of h igh performance, low
power consumption, and the strong real-time performance that
drive the embedded intelligent computing system to adopt a
heterogeneous multi-processor architecture.
The heterogeneous mult i-processors described here and the
Fig. 2. T he General Structure Model
symmetric mu lti-processors used for scientific co mputing, it is
The CPU will save its state informat ion in a core special completely different. Therefore, how to then design a hybrid
memo ry area called system management RAM. Th is s pecial heterogeneous computing platform in wh ich a general-purpose
memo ry area cannot be addressed and accessed by programs embedded processor and also a dedicated intelligent processor
running in other execution modes. By default, the addressing cooperate with each other, with an excellent performance-to-
request to the system management RAM will be transferred to power rat io, has then become the primary problem facing the
the image memory. embedded intelligent co mputing system. Embedded software
This feature allows the system management RAM to be technology will be then programmed again in the application
used as the secure storage device. It is worth noting that the process. The basic principle is the same as that of the JA VA
system management interrupt handler is loaded into the programming software. The processing power of the processor
system management RAM by the BIOS when the system is is improved through programming again. The develop ment of
started and the handler can access the physical memo ry space embedded software technology will then give JAVA software
without restriction, and can execute privileged instructions. In development provides great help.
the following formulas, the models are defined. On the one hand, the integration of mu ltip le software
ysd  Ps hsd xs  nsd (1) applications can be realized through the application of general
embedded software technology, which increases a s oftware
ysr  Ps hsr xs  nsr (2) application platfo rm. On the other hand, the increase in the
application platforms can provide for the develop ment and
yse  Ps hse xs  nse (3) embedding of JA VA software technology. The development
The consensus of the blockchain requires each node in the of modular software technology provides more support, wh ich
blockchain network to process each block in the blockchain, not only contributes to the development of the smart software
so that all nodes in the entire system can exchange data freely programming technology, but also then enables technological
and securely in a trustless environment. At the same t ime, innovation of multip le technologies. The hybrid embedded
since all nodes on the block mus t verify each transaction, the training system refers to the design of the embedded train ing
throughput of transactions on the blockchain is lo w. Un like system by adopting independent or integrated design ideas
traditional b lockchains, TBchain uses traditional blockchains when it is inconvenient to design the independent or integrated
as the base layer or the "first layer", and builds a three -layer embedded train ing system. The design of a hybrid embedded
blockchain model on top of the secure base layer blockchain, training system is usually to increase the function of the

978-1-6654-0052-7/22/$31.00 ©2022 IEEE 59

Authorized licensed use limited to: Indian Institute of Technology- Goa. Downloaded on September 16,2022 at 11:13:35 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Second International Conference on Artificial Intelligence and Smart Energy (ICAIS-2022)
IEEE Xplore Part Number: CFP22OAB-ART; ISBN: 978-1-6654-0052-7

embedded train ing system under the core condition that the the main control chip to co mplete the control operation of the
avionics system has been designed. This system uses the Wi- system, including data transfer p rocessing, control and general
Fi network and cellular network that have been fully deployed coordination of various modules to cooperate to complete the
in modern society to transmit the image data that embedded communicat ion work. In this article, the STM 32 main control
devices need to calculate to the cloud. The cloud uses the chip is specifically STM32103ZET6.
deployed server and workstation clusters to perform high - STM32103ZET6 has abundant peripheral resources, which
speed calculations and return results while making full use of greatly simplifies the circuit design work, and also has strong
embedded the features of co mpact equip ment are s mall and signal processing capabilities and anti-interference
suitable for deployment at the front end. In the figu re 4, the
capabilities.
systematic model is defined.
For embedded Ethernet data acquisition system hardware
design and control system respectively, and then based on the
SPI serial interface design imp lementation of the Ethernet data
acquisition system and control system integration, build ing
embedded Ethernet data acquisition and also control system
optimization function, co mp lete the design of the embedded
Ethernet data acquisition and control system. The figrue 6
gives the connection model.

Fig. 4. T he Systematci Model Defined

The cluster server manages all GPU workstations in a


cluster. It will establish the communication server program.
When various computing programs on the GPU workstation
are connected, they will be centrally managed by the cluster
server to monitor the computing load of each task in real time
with the data such as model name and version. Relying on the
communicat ion program, the GPU process can then inform the
cluster server's own model version and load pressure and other
informat ion. Afterwards, after receiving the image data sent Fig. 6. T he System Connection Model
by the cluster server, the GPU immediately starts calculation III. THE EXPERIMENT
and returns the calculation result.
In order to improve the operating efficiency of software The current data acquisition boards generally use a bus,
and hardware, each co mputing process supports a data buffer which can be inserted into the host to form an interposing
queue, and subsequent data will be received in advance and working mode. These bus technologies use the powerful data
processing capabilities of the machine and abundant software
placed in the queue, which can reduce the data waiting time
and hardware resources to provide very favorable conditions
for each calculation shown in the figure 5. for data application as the user environment and short training
period for operators when the system is put into operation.
Software develop ment is convenient, and C2420 itself
integrates 8051 core, wireless transceiver module, etc.
Therefore, when the circu it design is carried out, no
complicated peripheral circu it design is required to form a
complete wireless short-range communication data transceiver
system. The circu it design is simp le, so the project cycle is
shortened and the reliab ility of the system is imp roved. The
embedded Ethernet data collection module has been refined
and divided into: data migration module, data cleaning
module, and data filtering module that filters the collected
Fig. 5. Artificial Intelligence Embedded Model
Ethernet data after collecting data, the figure 7 shows the
C. The Embedded Operating System with Smart Chips simulation result.
Relying on the streaming media informat ion and also the
teaching system supported by the operating system, it can
work in parallel with the core computer CPU, wh ich is truly
independent of the CPU, and its functions have been further
strengthened. The communication system selects STM32 as

978-1-6654-0052-7/22/$31.00 ©2022 IEEE 60

Authorized licensed use limited to: Indian Institute of Technology- Goa. Downloaded on September 16,2022 at 11:13:35 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Second International Conference on Artificial Intelligence and Smart Energy (ICAIS-2022)
IEEE Xplore Part Number: CFP22OAB-ART; ISBN: 978-1-6654-0052-7

[7] He, Jiaji, Haocheng Ma, Yanjiang Liu, and Yiqiang Zhao. "Golden Chip -
Free Trojan Detection Leveraging T rojan T rigger’s Side-Channel
Fingerprinting." ACM Transactions on Embedded Comput ing Systems
(T ECS) 20, no. 1 (2020): 1-18.
[8] Wang, Haoxiang, Zhihui Li, Yang Li, Brij B. Gupta, and Chang Choi.
"Visual saliency guided complex image retrieval." Pattern Recognition
Letters 130 (2020): 64-72.
[9] White, Allan L. "Verifying Autonomous Air Traffic Algorithms in the
Presence of Sensor Error and Flight Perturbations." In 2020 AIAA/IEEE
39th Digital Avionics Systems Conference (DASC), pp. 1-8. IEEE,
2020.
[10] Kadiyala, Sai Praveen, Manaar Alam, Yash Shrivastava, Sikhar
Patranabis, Muhamed Fauzi Bin Abbas, Arnab Kumar Biswas, Debdeep
Mukhopadhyay, and Thambipillai Srikanthan. "LAMBDA: Lightweight
assessment of malware for emBeddeD architectures." ACM
Transactions on Embedded Computing Systems (TECS) 19, no. 4
(2020): 1-31.
[11] Nicole, Olivier, Matthieu Lemerre, Sébastien Bardin, and Xavier Rival.
"Automatically Proving Microkernels Free from Privilege Escalation
from their Executable." arXiv preprint arXiv:2003.08915 (2020).
[12] Ho, Bo-Jhang, Bharathan Balaji, Mehmet Koseoglu, Sandeep Sandha,
Siyou Pei, and Mani Srivastava. "Quick Question: Interrupting Users for
Microtasks with Reinforcement Learning." arXiv preprint
arXiv:2007.09515 (2020).
[13] Hao, Yi, Jie Li, Nannan Wang, and Xinbo Gao. "Modality adversarial
neural network for visible-thermal person re-identification." Pattern
Recognition 107 (2020): 107533.
[14] Kim, Hakbin, and Dong-Wan Choi. "Pool of Experts: Realtime Querying
Fig. 7. T he System Performance Estimation Result Specialized Knowledge in Massive Neural Networks." In Proceedings of
the 2021 International Conference on Management of Data, pp. 2244-
IV. CONCLUSION AND SUMMARY 2252. 2021.
Design of the high-confidence embedded operating system [15] Zhang, Fengxiu, and Eric W. Welch. "More than just managerial self-
efficacy: conceptualizing and predicting top managers' means efficacy
based on artificial intelligence and smart ch ips is studied in about the organization under extreme events." Journal of Managerial
this paper. The embedded Ethernet data have been collected Psychology (2021).
and filtered, wh ich is conducive to its control. Th is paper uses [16] Zhu, Banghua, Jiantao Jiao, and Jacob Steinhardt. "Robust estimation via
fuzzy associative space theory to filter embedded Ethernet generalized quasi-gradients." arXiv preprint arXiv:2005.14073 (2020).
data. The cluster server manages all GPU workstations in the [17] Smys, S., Abul Basar, and Haoxiang Wang. "Hybrid intrusion detection
cluster. It will establish the communication server program, system for internet of Things (IoT)." Journal of ISMAC 2, no. 04 (2020):
and at the same time, the simu lations are conducted to validate 190-199.
the model. The future study will focus on how to imp lement [18] Li, Qian, Yong Qi, Qingyuan Hu, Saiyu Qi, Yun Lin, and Jin Song Dong.
the system with higher complexity. "Adversarial adaptive neighborhood with feature importance-aware
convex interpolation." IEEE Transactions on Information Forensics and
Security 16 (2020): 2447-2460.
REFERENCES
[19] Qiu, Zifeng, Yahan Wang, Feng Gao, Sha Liu, Junkai Zhang, Fuchen
[1] Varela, Carlos A. "Towards Provably Correct Probabilistic Flight Yang, and Yan Hu. "UAV Inspection Technology Based on Lightweight
Systems." In Dynamic Data Driven Applications Systems: Third Edge Computing Framework." In Journal of Physics: Conference Series,
International Conference, DDDAS 2020, Boston, MA, USA, October 2- vol. 2010, no. 1, p. 012155. IOP Publishing, 2021.
4, 2020, Proceedings, vol. 12312, p. 236. Springer Nature, 2020. [20] Zheng, Xiaoming. "Data Collection in Quantitative Research." In
[2] Sun, Chen, and En-Hui Yang. "A Watermarking-Based Framework for Research Methods for Student Radiographers, pp. 79-92. CRC Press,
Protecting Deep Image Classifiers Against Adversarial Attacks." In 2021.
Proceedings of the IEEE/CVF Conference on Computer Vision and [21] Haoxiang, Wang, and S. Smys. "MC-SVM Based Workflow Preparation
Pattern Recognition, pp. 3329-3338. 2021. in Cloud with Named Entity Identification." Journal of Soft Computing
[3] Han, Yanjun, Jiantao Jiao, T sachy Weissman, and Yihong Wu. "Optimal Paradigm (JSCP) 2, no. 02 (2020): 130-139.
rates of entropy estimation over Lipschitz balls." The Annals of [22] Zhong, Boxuan, Rafael Luiz da Silva, Michael Tran, He Huang, and
Statistics 48, no. 6 (2020): 3228-3250. Edgar Lobaton. "Efficient Environmental Context Prediction for Lo wer
[4] Zhang, Kai, Martin Danelljan, Yawei Li, Radu T imofte, Jie Liu, Jie Tang, Limb Prostheses." IEEE Transactions on Systems, Man, and
Gangshan Wu et al. "AIM 2020 challenge on efficient super-resolution: Cybernetics: Systems (2021).
Methods and results." In European Conference on Computer Vision, pp. [23] Qingqing, Li, Jussi T aipalmaa, Jorge Pena Queralta, T uan Nguyen Gia,
5-40. Springer, Cham, 2020. Moncef Gabbouj, Hannu Tenhunen, Jenni Raitoharju, and Tomi
[5] Thomas, Sam L., Jan Van den Herrewegen, Georgios Vasilakis, Zitai Westerlund. "Towards active vision with UAVs in marine search and
Chen, Mihai Ordean, and Flavio D. Garcia. "Cutting Through the rescue: Analyzing human detection at variable altitudes." In 2020 IEEE
Complexity of Reverse Engineering Embedded Devices." IACR International Symposium on Safety, Security, and Rescue Robotics
Transactions on Cryptographic Hardware and Embedded Systems (SSRR), pp. 65-70. IEEE, 2020.
(2021): 360-389. [24] Chiang, Yi-Shyuan, Ruei-Che Chang, Yi-Lin Chuang, Shih-Ya Chou,
[6] Lee, Hyunhoon, Young-Seok Kim, Mijung Kim, and Youngjoo Leeb. Hao-Ping Lee, I-Ju Lin, Jian-Hua Jiang Chen, and Yung-Ju Chang.
"Low-Cost Network Scheduling of 3D-CNN Processing for Embedded "Exploring the design space of user-system communication for smart-
Action Recognition." IEEE Access (2021). home routine assistants." In Proceedings of the 2020 CHI Conference on
Human Factors in Computing Systems, pp. 1-14. 2020.

978-1-6654-0052-7/22/$31.00 ©2022 IEEE 61

Authorized licensed use limited to: Indian Institute of Technology- Goa. Downloaded on September 16,2022 at 11:13:35 UTC from IEEE Xplore. Restrictions apply.
Proceedings of the Second International Conference on Artificial Intelligence and Smart Energy (ICAIS-2022)
IEEE Xplore Part Number: CFP22OAB-ART; ISBN: 978-1-6654-0052-7

[25] Garg, Mamta, Ajatshatru Arora, and Savita Gupta. "An Efficient Human chest x-rays using Confidence-Aware anomaly detection." IEEE
Identification Through Iris Recognition System." Journal of Signal transactions on medical imaging 40, no. 3 (2020): 879-890.
Processing Systems (2021): 1-8. [27] Zeufack, Vannel, Donghyun Kim, Daehee Seo, and Ahyoung Lee. "An
[26] Zhang, Jianpeng, Yutong Xie, Guansong Pang, Zhibin Liao, Johan unsupervised anomaly detect ion framework for detecting anomalies in
Verjans, Wenxing Li, Zongji Sun et al. "Viral pneumonia screening on real time through network system’s log files analysis." High-Confidence
Computing 1, no. 2 (2021): 100030.

978-1-6654-0052-7/22/$31.00 ©2022 IEEE 62

Authorized licensed use limited to: Indian Institute of Technology- Goa. Downloaded on September 16,2022 at 11:13:35 UTC from IEEE Xplore. Restrictions apply.

You might also like