You are on page 1of 6

2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).

Taxonomy of attacks on Industrial Control


protocols
Zakarya DRIAS Ahmed SERHROUCHNI Olivier VOGEL
Schneider Electric Telecom ParisTech Schneider Electric
Carros, France Paris, France Carros, France
Zakarya.drias@schneider- Ahmed.serhrouchni@telecom- Olivier.vogel@schneider-
electric.com paristech.com electric.com

Abstract— Industrial control systems (ICS) are highly distributed this usually includes the security functions and features such
information systems used to control and monitor critical infra- as authentication and encryption which expose the protocol
structures such as nuclear plants, power generation and users to a security threats. Another source of vulnerabilities is
distribution plants, Oil and Gas and many other facilities. The the convergence of the protocols to run over IP networks to
main architecture principles of ICS are; real time response, high
meet the evolving needs of the business. The most used
availability and reliability. For these specific purposes, several
protocols has been designed to ensure the control and supervision protocols in the ICS are DNP3 and Modbus. DNP3 is widely
operations. Modbus and DNP3 are the most used protocols in the used in the energy sector where the control system is
ICS world due to their compliance with real time needs. With the monitoring and controlling geographically dispersed
increasing of the connectivity to the internet world for business outstation. Modbus protocol is used for the communication
reasons, ICS adopted Internet based technologies and most of inter- controller or between controller and field devices where
communication protocols are redesigned to work over IP. This the real time operations are required.
openness exposed the ICS components as well as communication Both protocols are specified to be easy to use but without any
protocols to cyber-attacks with a higher risk than attacks on security awareness. This lack of security in DNP3 and Modbus
traditional IT systems. In order to facilitate the risk assessment of
motivated us to conduct an investigation on the different
cyber-attacks on ICS protocols we propose a taxonomy model of
different identified attacks on Modbus and DNP3.the model is attacks on both protocols and build a generic taxonomy model
based on the threat origin, threat type, attack type, attack for attacks on communication protocols in general and
scenario, vulnerability type and the impact of the attack. We industrial control protocols in particular in order to simplify
populate this Taxonomy model with identified attacks on Modbus the risk analysis. Our study takes into account only attacks on
and DNP3 from previous academic and industrial works. IP based protocols, since all the field protocols are converging
to be integrated to the IP stack.
Keywords—ICS, Modbus, DNP3, Attack Taxonomy This paper is structured as the following; in section II, we
start with an overview of industrial control protocols in typical
I. INTRODUCTION control system architecture with a deep focus on Modbus and
DNP3. In section III, we follow by proposing the attack
Industrial control systems is a large term used to describe taxonomy model and several attack on Modbus TCP and
several types of systems such as, DCS (Distributed Control DNP3 identified in academic and industrial works. In section
systems), SCADA (Supervisory Control and data Acquisition), IV, we populate the taxonomy model by identified attacks. We
IAS (Industrial Automation system), IACS (Industrial conclude by discussing our perspectives and future works.
Automation and Control Systems) or even PLC(Programmable
Logic Controller). ICS are typically used in industries such II. OVERVIEW ON INDUSTRIAL CONTROL PROTOCOLS
Power plants, Water and waste water facilities, Oil and Gas
In ICS architectures the communication protocols are used in
refineries and distribution, Nuclear plants …etc. These control
the different architecture layers; corporate network, operation
systems are critical to the operation of critical infrastructures
network (Control Room), and control network respecting the
that are often highly interconnected and mutually dependent
operational constraints of each network. Corporate network
systems. It is important to note that approximately 90 percent
protocols are similar to ones used in the IT systems where in
of the critical infrastructures operate many of the industrial
Operation and control networks, customized and proprietary
processes mentioned above [13] SCADA systems are a set of
protocols are used to guarantee the required performances,
Software and hardware used generally in the control and
reliability and precision in Control and supervision operations.
monitoring of geographically dispersed assets and process (Ex:
Figure 1 is illustrating the different protocols most used in ICS
Gas Distribution) where the centralization of data acquisition
layers. The critical part in automation system architecture is
and control are critical to System Operation. Where the DCS is
the communication between the Control room and Control
generally focused on the automatic control of a process usually
network Devices, as well as inter-communication between
within a closed area (Ex: Gas Refineries).
control devices (PLCs, RTUs) . The most used protocols for
Understanding how ICS work requires a basic understanding
those communications are; DNP3 and Modbus that we detail
of the underlying communication protocols and their semantic.
in the following subsections.
In the Industrial automation and control sector, many
specialized protocols are used today for different purposes, A. Modbus Protocol
control commands, Monitoring, real time data exchange. Modbus is an application layer protocol commonly used in
Those Protocols are designed to ensure efficiency, Reliability control systems. It was published initially by Modicon in
and precision real time operation. Therefore, any other 1979, and the protocol standards are now managed by the
inefficient functions are taken off the protocol. Unfortunately,

978-1-4673-9265-5/15/$31.00 ©2015 IEEE


2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).

Figure 1 Overview of communication protocols in Typical ICS


Architecture
Figure 3 Modbus TCP Message Structure
Modbus foundation. Modbus is one of the most widely used
control systems protocol for monitoring and controlling
purposes [17].Originally, Modbus was running over serial
communication channels. The protocol was extended to run
over TCP communications and there are gateway products that
convert serial Modbus to Modbus TCP and vice versa. Figure 4 Modbus TCP Protocol Stack
Modbus was created for real time purposes in order to receive
and send commands to controllers and field devices. While
designed, security has not been taken into account, which B. Distributed Network Protocol (DNP3)
makes from Modbus a point of failure in ICS architecture. DNP3 is an application layer protocol developed by GE Harris
Modbus has two variants, Modbus serial and Modbus TCP, in in early 1990s. The aim of the protocol is to define how to
our work we focus only on Modbus TCP as it is the most used communicate the control command between SCADA devices
nowadays. The Modbus TCP protocol stack is showed in in dispersed geographical areas. The first priority of DNP3 is
Figure 2; where Modbus message is encapsulated in TCP to ensure the reliability of the communication between the
packet. For a connection between the Modbus client and SCADA master (DNP3 Client) and remote outstations (DNP3
server, the server (Slave) listens on the port number 502. server). While created, DNP3 was a serial protocol but the
Modbus is a Request /Response protocol where there is no openness of Industrial control systems on internet technologies
exception reporting. The protocol message format is detailed and the need of simplifying the access to remote outstations
in the figure 3. Modbus Message consists on Modus led to extend DNP3 to work on IP, by encapsulating DNP 3
Application Protocol (MBAP) header and Modbus Application Data frame in TCP and UDP Packets [17].
PDU. The MPAB contains four fields; transaction identifier, DNP3 reliability is due to the frequent use of CRC checks for
Protocol identifier, Length, Unit Identifier. Transaction any exchange between Master and slaves in addition to the
identifier allows matching the request and responses while the TCP CRC. Unlike Modbus, DNP supports bidirectional
protocol identifier the application protocol encapsulated by the communication flows; therefore, the outstation (Slave) can
MBAP, for example Modbus is designated by 0. The length initiate a communication with the master for alarming
field indicates the length of the required data including Unit purposes. Later in this chapter we detail the different use cases
ID and, functional code and function parameters. The function of DNP3 protocol. A master sends a request message to the
code indicates the transaction type. In Modbus specification salve, with a specific Class and command. Initially the master
[18] we find 127 function codes with the ability to implement polls the point values for monitored data using the command
customized functional codes. Some examples of function Read and class 0, then stores them the master database as static
codes are: FC 1 – Read Coils, FC 6 – Write Single Register, reference data points, then it polls the events data by sending
FC 7 – Read Exception Status, FC 17 – Report Slave ID. Read command on the class 1, 2, 3 which corresponds to the
Reference [18] details the function codes of Modbus. buffers of the slave the polled data are compared with the
static data.

Figure 2 Example of Modbus Master Slave exchange

Figure 5 DNP3 Master Slave Exchange


2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).
one set up by the asset owner to a behavior set by the attacker.
The slave can be configured to report the content of the buffer Since Stuxnet attack, the security of ICS is becoming the
1, 2, 3 as unsolicited messages. The figure 5, shows the format highest priority of the critical infrastructures owners.
of DNP3 message, and specifies DNP3 request and reply Most works on attacks taxonomy have focused on the
messages in the application layer of TCP/IP stack. classification of attacks for the IT systems such as [2], [3], [4],
The application header of the DNP3 contains an application communication standards and protocols [5], [6]. Other works
layer Byte to indicate the message control information such as has focused on the attack taxonomy for SCADA systems
dealing with large responses that may exceed the buffer size. which categorized the attacks based on their target in the
The functional code field indicates the purpose of both request SCADA component by listing each attack description and the
and reply message. In DNP3 specification [17], we find 23 vulnerability exploited by the attacker [7], [8], and [11].
defined function codes for request classified into six Huitsing et al [9], East and al [10] have proposed taxonomy of
categories; Transfer function, control function, freeze function, cyber-attacks on industrial protocols by enumerating the
application control function, configuration functions and time cyber-attacks on the protocols regardless the implementation
synchronization functions. A reply message can be: of the vendors. Their taxonomies were missing a generic
Confirmation, Response or unsolicited response. The model for attack on communication protocols and were
Application header incorporates two bytes with a specific specified to each protocol.
function for each bit to indicate errors in the master or the In our paper the main objective is to build taxonomy of attack
slave such as Slave restart, Function code not supported and on most common used industrial protocols in order to simplify
unknown request. Following the application header, the Data the risk analysis in the industrial facility using those protocols.
section of DNP3 message is a set of data objects that Other sub objectives are:
encapsulated the requested data by the Master or in the
unsolicited messages. Figure 4 and is an example of DNP3  Construct a model of cyber-attack classification for
exchange between master and slave. industrial communication protocols with a specification of key
Unlike Modbus, DNP3 supports today, two authentication design principles of the protocols and the context of usage in
modes; shared key authentication between the master and order to determine the critical flows. The model should be able
slave and certificate based authentication. In our paper we to generate the taxonomy.
don't focus on the DNP3 secure, as it is not widely used since  The model should be open to be enriched, we mean by
the key management in the SCADA systems is still not enrichment adding new threats, attack target and other Attack
standardized. scenarios.
 The model should contain different levels of details on an
attack. Due to privacy matters in some attacks or
vulnerabilities disclosure, an attack may be incomplete or
missing some details on the exploitability. In this case we
assume all possible attack scenarios presented in the
taxonomy. Any false scenario will be taken off when the real
attack is disclosed.
These four goals led us to propose a model of attack taxonomy
that we call "TAVI" the proposal of this model for ICS
protocols attack taxonomy is an adaptation of the AVT model
proposed by Fleury and al [12] to be more focusing on
communication protocols and to simplify their related risk
analysis. The following is detailing the attack taxonomy
model actors:
 Threat origin: following NIST SP800-82 R2 [13],
Figure 6 DNP3 Message structure industrial control systems in general and protocols in
particular are threaten by numerous threat vectors
including; Attackers , Bot-network Operators, Criminal
groups, foreign intelligence systems, Insiders, Phishers,
Spammers, Spyware/malware Authors, Terrorists and
industrial spies. For more details on the threat vectors refer
to [13].
 Threats: Based on the definition in [14] a threat is
"Any circumstance or event with the potential to adversely
Figure 7 DNP3 Protocol Stack impact organizational operations (including mission,
functions, image, or reputation), organizational assets, or
individuals through an information system via
III. ATTACK TAXONOMY MODEL unauthorized access, destruction, disclosure, modification
st
In April 21 2011, the most sophisticated attack against a of information, and/or denial of service. Also, the potential
control system has been discovered by Symantec known as for a threat-source to successfully exploit particular
"Stuxnet" [1]. Stuxnet exploited several zero day information system vulnerability". Therefore, we
vulnerabilities in Windows to take control of the control- categorize the threats into five major categories; Spoofing,
command software installed in the control room of the Iranian Data Tempering, Disclosing data, Denial of service and
nuclear facility. Its main target is to damage the control system elevation of privileges. An attack on ICS protocol is part of
by reprogramming the PLCs to change their behavior from the one or several threat categories.
2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).
Threats Attack Vulnerability Impact (CVSS)
Threat Origin Threat Type Attack Scenario Attack Target Vulnerability Type
Attacker Spoofing Probe Client (Master) Lack of Availability
Bot-Network Operator Data Tampering Scan Server (Salve) Lack of Intergrity
Criminal group Data Disclosure Flood Communication layer Lack of Confidentiality
Foreign Intellegence systems Elevation of priveleges Authenticate
Insiders Bypass
Phishers Spoof
Spammer Authors Eavesdrop
Industrial spies Misdirect
Read/Copy
Terminate
Execute
Modify
Delete

Table 1 TAVI Attack taxonomy model

Attack Scenario: is the description of actions that an


attacker to sabotage the protocol parties; client, server and 1) Broadcast Message Spoofing (M1):
communication layer. Fluery et al [12] details a list of The attack is about sending faked broadcast messages to
attack scenarios such as, Probe, Scan, Flood, Authenticate, Modbus server in the device, which doesn't answer to the
Bypass, Spoof, Eavesdrop, Misdirect, Read/Copy, Modbus client but executes any action added to the Modbus
Terminate, Execute, Modify, and Delete. Part of these message in order to modify the behavior of the Field device.
attack actions applies to our case of study; DNP3 and Unfortunately the switches cannot completely filter the
Modbus which will be detailed later in this paper. broadcast traffic. This attack scenario is possible only in the
 Attack Target: the attack target is the portion of the case of using Modbus gateway from Ethernet to serial
protocol specification that attacker is targeting, we identify devices and requires a Modbus protocol sniffer and Modbus
three important parts in the protocol description; Client, message generator.
Server and communication layer (Network). In our 2) Baseline response replay (M2): This attack involves
taxonomy model, an attack is targeting the client, the recording Modbus messages between a Modbus client and a
server, as well as the communication layer. Modbus server, and replaying some of the recorded
messages back to the Server. This attack exploits the fact
 Vulnerability type: Vulnerability describes the
weak part of the protocol specification that makes it that non CRC is used in the Modbus TCP specification.
exploitable by a threat using one of the attack scenarios 3) Direct Slave Control (M3): This attack involves locking
previously discussed. A vulnerability type is a lack of out a master (client) and controlling one or more field
security principles; Availability, Integrity and devices. This attack consists on spoofing the identity of the
confidentiality. We categorize all vulnerabilities in those existing Modbus client and taking the control on the device
three high level categories; Lack of availability, lack of embedding Modbus server. Spoofing the identity dos not
integrity and lack of confidentiality. require a sophisticated effort as the client identity is only
related to its IP address.
 Impact: the impact of an attack is about calculating 4) Modbus Network Scanning (M4): This attack involves
the risk level associated to this attack [15]. For risk level sending benign messages to all possible addresses on a
estimation, we use the Common vulnerability Scoring
Modbus network to obtain information about field devices.
system (CVSS) .CVSS is a universal and free scoring
system for disclosed vulnerabilities. Houmb et al [16] 5) Passive Reconnaissance (M5): This attack involves
Specifies the CVSS algorithms and the different metrics passively reading Modbus messages or network traffic. This
used to determine the final score for vulnerability and how attack helps the attacks to profile the process details by
to deduct its related impact. In our case all the previous reconstructing the device application.
taxonomy model entries are inputs of the CVSS. 6) Response Delay (M6): This attack involves delaying
response messages so that the master receives out-of-date
IV. TAXONOMY OF ATTACKS ON MODBUS AND DNP3 information from slave devices. This attack intends to
sabotage the supervision in case of the command in the
In this section we describe the different attacks on Modbus Modbus message is a diagnostic message. Or interrupt the
TCP and DNP3 that we identified in [9], [10] and published on process if the command in the message is to execute a
[17]. Then, we populate the taxonomy model that we proposed function in the Device application, then the client will
with attacks on Modbus TCP and DNP3. Note, that attacks resend the same message and flood the device by the same
that we discuss in the following sections are targeting the Modbus message or deduct that the device is out of the
protocols specification. Table 2 resumes the population of the network which will false alarm the operator to take an
taxonomy model with the different attacks on Modbus TCP inappropriate action.
and DNP3. 7) Man in the middle attack (M7): This attack involves
A. Attacks on Modbus: introducing a computer with the appropriate (serial or
Ethernet) adaptors to an unprotected communication link.
We assume that most of attacks on Modbus TCP rely on the
This man in the middle device is embedding a client and
ability to implement interception, modification and generation
Modbus server; therefore it can read, modify and fabricate
mechanisms. All Modbus TCP implementations today in the
Modbus messages to or from the device. This attack is
ICS market do not implement security features, which confirm
known to be the most dangerous attack on Modbus
our assumption. The following are published attacks and
protocols as it can take the full control of both protocol
Modbus TCP discussed in [9] and [17].
parties; client and server in the same time.
2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).

Protocol Threats Attack Vulnerability Impact


Attacks
Modbus CVS S
attacks Threat Origin Threat Type Attack S cenario Attack Target Vulnerability Type S core
Lack of integrity
M1 Attacker, Insiders, Botnet Operator Spoofing, Data tampering Spoof, Authenticate bypass Server (Slave) 7.8
Lack of confidentiality
Scan, Probe, authenticate
M2 Attacker, Insiders, Criminal group Spoofing, Data tampering Server (Slave) Lack of integrity 6.9
bypass
Probe,
Lack of integrity
M3 Attacker, Insiders, Criminal group, Spoofing,Data disclosure AuthenticateBypass,Delete, Client /server 9.2
Lack of availability
modify, execute, terminate
Industrial spies, Attacker, Criminal Client, Server,
M4 Data disclosure Scan Lack of confidentiality 2.9
group, Botnet operator, Communication layer
Industrial spies, Attacker, Criminal Read/copy, Authenticate client, server,
M5 Data Disclosure Lack of confidentiality 2.9
group, Botnet operator,Insiders Bypass communication layer
Attacker, Criminal group, Botnet Lack of integrity, Lack
M6 Data tampering M isdirect, Eavesdrop, Flood Client 7.8
operator,Insiders, Spammer authors of availability
Probe,Scan, Flood,
Spoofing, Data tampering, Authenticate bypass, Spoof, Lack of integrity,lack of
Attacker, Criminal group, Botnet Cient , server,
M7 Data disclosure,Elevation of Eavesdrop, M isdirect, confidentiality, lack of 10
operator,Insiders, Spammer authors communication link
privileges Read/Copy, Terminate, availability
Execute, M odify,Delete
DNP3
Attacks
Industrial spies, Attacker, Criminal Read/copy, Authenticate server, communication
D1 Data Disclosure Lack of confidentiality 2.9
group, Botnet operator,Insiders Bypass layer
SCAN, Probe, authenticate
D2 Attacker, Insiders, Criminal group Spoofing, Data tampering Server (Slave) Lack of integrity 6.9
bypass
Probe,Scan, Flood,
Spoofing, Data tampering, Authenticate bypass, Spoof, Lack of integrity,lack of
Attacker, Criminal group, Botnet Cient , server,
D3 Data disclosure,Elevation of Eavesdrop, M isdirect, confidentiality, lack of 10
operator,Insiders, Spammer authors communication link
privileges Read/Copy, Terminate, availability
Execute, M odify,Delete
Attacker, Criminal group, Botnet Lack of integrity,lack of
D4 Data tampering M isdirect communication link 4.9
operator,Insiders availability
Authenticate Bypass, Spoof
Attacker, Criminal group, ,Insiders, Data tampering, elevation of Lack of integrity, Lack
D5 Execute,M odify Server 9.2
Spammer Author privileges,Spoofing of availability

Authenticate Bypass,Spoof
Attacker, Criminal group, ,Insiders, Data tampering, elevation of Lack of integrity, Lack
D6 Execute,M odify,delete Server 9.2
Spammer Author privileges,Spoofing of availability

Attacker, Criminal group, ,Insiders, Data tampering, elevation of Lack of integrity, Lack
D7 Execute, M odify Server 9.2
Spammer Author privileges,Spoofing of availability

Probe,Scan, Flood,
Spoofing, Data tampering, Authenticate bypass, Spoof, Lack of integrity,lack of
Attacker, Criminal group, Botnet Cient , server,
D8 Data disclosure,Elevation of Eavesdrop, M isdirect, confidentiality, lack of 10
operator,Insiders, Spammer authors communication link
privileges Read/Copy, Terminate, availability
Execute, M odify,Delete

Table 2 Taxonomy of Attacks on Modbus /TCP and DNP3

4) Transport Sequence Modification (D4): The Sequence


B. Attacks on DNP3:
field is used to ensure in-order delivery of fragmented
We take into account the same assumptions on the ability to
messages. The sequence number increments with each
intercept, modify and generate DNP3 messages.
1) Passive network reconnaissance (D1): An attacker able to fragment sent, so predicting the next value is trivial. An
access the control room or the outstation captures and attacker who inserts fabricated messages into a sequence of
analyzes DNP3 messages. This attack is similar to the fragments can inject any data and/or cause processing
Passive network reconnaissance on Modbus TCP. The goal errors.
of this attack is to get information on the topology of 5) Outstation Write Attack (D5): This attack sends a DNP3
control network, devices functions and device control message with Function Code 2, which writes data objects to
application. an outstation. The attack can corrupt information stored in
2) Baseline Response Replay (D2): An attacker with the outstation’s memory, causing an error or overflow.
knowledge of normal DNP3 traffic patterns simulates 6) Clear Objects Attack (D6) This attack sends a DNP3
responses to the master while sending fabricated messages message with Function Code 9 or 10 to freeze and clear
to outstation devices data objects. The attack can clear critical data or cause an
3) Man in the middle (D3): An attacker installs a “man-in- outstation device to malfunction or crash. Note that the
the-middle” device between the master and outstations that attack involving Function Code 10 is problematic because a
can read modify and fabricate DNP3 messages and/or message with this function code does not require an
network traffic. acknowledgement.
2015 International Conference on Protocol Engineering (ICPE) and International Conference on New Technologies of Distributed Systems (NTDS).
7) Outstation Data Reset (D7): This attack sends a DNP3 [7] Octavio Nieto-Taladriz Garcia Security in Embedded Systems
Challenges and Oportunities, International Conference on Emerging
message with Function Code 15. The attack causes an Security Information, Systems and Technologies, Securware 2007
outstation device to reinitialize data objects to values [8] Srivaths Ravi, Anand Raghunathan, Paul Kocher and Sunil Hattangady
inconsistent with the state of the system. Security in embedded systems: Design challenges, ACM Transactions
on Embedded Computing Systems (TECS), vol.3, no.3, pages 461–
8) Configuration Capture Attack (D8): This attack sends a 491,2004.
message with the fifth bit in the second byte of the IIN set, [9] Huitsing, P., Chandia, R., Papa, M., & Shenoi, S. (2008). Attack
taxonomies for the Modbus protocols. International Journal of Critical
which indicates that the configuration file of the targeted Infrastructure Protection, 1, 37-44
outstation is corrupted. The attack causes the master to [10] East, S., Butts, J., Papa, M., & Shenoi, S. (2009). A Taxonomy of
transmit a new configuration file, which is intercepted by Attacks on the DNP3 Protocol. In Critical Infrastructure Protection III
(pp. 67-81). Springer Berlin Heidelberg.
the attacker. A separate attack is then executed to modify
[11] Zhu, B., Joseph, A., & Sastry, S. (2011, October). A taxonomy of cyber-
and upload the file to the targeted outstation. attacks on SCADA systems. In Internet of things (iThings/CPSCom),
2011 international conference on and 4th international conference on
cyber, physical and social computing (pp. 380-388). IEEE.
In table 2 we apply the attack taxonomy model on the [12] Fleury, T., Khurana, H., & Welch, V. (2008). Towards a taxonomy of
described attacks on Modbus TCP and DNP3 with applying a attacks against energy control systems. In Critical Infrastructure
CVSS scoring to each attack based on the different entries of Protection II (pp. 71-85). Springer US.
[13] STOUFFER, Keith, FALCO, Joe, et SCARFONE, Karen. Guide to
the taxonomy model as an impact analysis. The result of the industrial control systems (ICS) security. NIST special publi-cation,
impact analysis indicates which attack is may have a 2011, p. 800-82
catastrophic impact on the infrastructure deploying both [14] Gutierrez, C. M., & Jeffrey, W. (2006). Minimum security requirements
for federal information and information sys-tems. Federal Information
protocols. In our analysis we conclude that both Man in the Processing Standards Publica-tion, 17.
middle attack and spoofing attack disclose the highly risky [15] Houmb, Siv Hilde, and Virginia NL Franqueira. "Estimating ToE risk
vulnerabilities in the design of the protocols; lack of level using CVSS." Availability, Reliability and Security, 2009.
ARES'09. International Conference on. IEEE, 2009.
authentication and miss of integrity. [16] KNAPP, Eric D. et LANGILL, Joel Thomas. Industrial Network
Security: Securing critical infrastructure networks for smart grid,
SCADA, and other Industrial Control Systems. Syngress, 2014.
V. CONCLUSION AND FUTURE WORKS [17] DigitalBond, Modbus TCP.
http://www.digitalbond.com/scadapedia/protocols/modbus-2
In this paper we conducted an analysis of attacks on most used
[18] MODBUS, I. D. A. Modbus application protocol specification v1.
industrial control protocols, DNP3 and Modbus. We discussed 1a. North Grafton, Massachusetts (www. modbus. org/specs. php), 2004.
in details the semantic of both protocols and defined a
taxonomy model for attacks on both protocols which we
populated with different attack identified in prior researches on
attacks on ICS. This taxonomy simplifies the risk analysis
related to cyber-attacks on ICS in general and control
protocols in particular. In a future work we will exploit the
results of this taxonomy to conduct a risk assessment on ICS
in prior design stage and to define the security counter
measure by including security requirement and services in the
ICS Design steps. These security services are mutual
authentication between two communicators and cryptographic
mechanisms to protect the integrity of data in transit and data
in rest in both Modbus TCP and DNP3. Redesigning the
protocols to include security services of integrity and
authentication is the subject of our next work. The ultimate
goal is to standardize the services that we will include to the
protocols.
REFERENCES

[1] Nicolas Falliere, Liam O Murchu, and Eric Chien, W32. Stuxnet Dossier
, Symantec Security Response, Version 1.4 February 2011,
[2] Simon Hansman, Ray Hunt, A taxonomy of network and computer
attacks, Computers & Security,DTD5, 2004.
[3] John D. Howard, An Analysis Of Security Incidents On The Internet
1989 - 1995, dissertation, Carnegie Mellon University, April 1997.
[4] Kevin S. Killourhy, Roy A. Maxion and Kymie M. C. Tan,A Defense-
Centric Taxonomy Based on Attack Manifestations, Proceedings of
International Conference on Dependable Sys-tems & Networks:
Florence, Italy, 28 June - 01 July 2004.
[5] Daniel Lough, A Taxonomy of Computer Attacks with Applications to
Wireless Networks, Ph.D Thesis, Virginia Polytechnic Institute and State
University, 2001
[6] MO Chun Man and Victor K. Wei, A taxonomy for attacks on mobile
agent; EUROCON’2001, Trends in Communications, International
Conference on. Volume 2, 4-7 July 2001 Page(s):385 - 388 vol.2

You might also like