1 of 19

MULTI-AGENT SYSTEMSINFORMATION SECURITY

2 of 19

2

  • A MULTI-AGENT SYSTEM (MAS) is a collection of interconnected software and hardware agents that are capable of interacting with each other and with the environment, possess intelligent capabilities, and can act individually and cooperatively to achieve goals.
  • 1996–2005 – International FIPA Association (Foundation for Intelligent Physical Agents).
  • 2002 – M. Wooldridge (ed.), “An Introduction to Multi-Agent Systems”.

*Vasilev V.I. Axborot xavfsizligining intellektual tizimlari: o'quv qo'llanma. – 3-nashr. – M.: Innovatsion injiniring, 2017. (6-bob. Ko‘p agentli axborot xavfsizligi tizimlari. – B. 159-185).

3 of 19

On New Perspectives of Modern Artificial Intelligence

3

4 of 19

Multi-Agent Technology for Real-Time Resource Management

4

Classic systems

Multi-Agent systems

Simultaneously active programs (coroutines)

  • Hierarchy of large programs
  • Sequential execution of operations
  • Top-down instructions
  • Centralized solutions
  • Data-driven
  • Predictability
  • Stability
  • Tendency to reduce complexity
  • Full control
  • Large networks of small agents
  • Parallel execution of operations
  • Negotiation
  • Distributed solutions
  • Knowledge-based
  • Self-organization
  • Evolution
  • Tendency to increase complexity
  • Creating conditions for development

5 of 19

5

AGENT- An agent is an autonomous entity capable of perceiving the external environment through sensors and acting upon the environment using effectors (actuators).

  • General structure of an agent

  • Smart agent – an entity with its own (local) knowledge base and the capability to perform intelligent (smart) actions.

6 of 19

AGENT LIFE STICK

6

7 of 19

Key Characteristics of Intelligent Agents

7

Autonomy

Social behavior

Reactivity

Beliefs

behavior

Desires

Intentions

BDI Cognitive Model (Belief-Desire-Intention)�Scenarios of Intelligent Agent (IA) behavior:�a) Individual behavior;�b) Coordination of joint activities;�c) Cooperation (aimed at developing the best method to achieve a goal).

8 of 19

MAS Design Stages

8

  1. Building the ontology (domain model).
  2. Defining the structure of problems solved using MAS.
  3. Identifying the composition of agents (MAS architecture).
  4. Defining scenarios of interactions between agents.
  5. Selecting agent communication protocols/languages.
  6. Implementing MAS software, debugging, and testing.

9 of 19

Information Security in Multi-Agent Systems

  • This refers to the processes carried out in multi-agent systems, where numerous agents interact and work together to achieve a common goal, ensuring that information is securely stored, transmitted, and critical data is maintained.
  • The main purpose of information security in multi-agent systems is to protect against external threats (e.g., hackers, viruses) as well as internal threats (e.g., agent malfunctions or data loss).

9

10 of 19

The key elements of information security in multi-agent systems are:

  1. Data Confidentiality:�Information exchanged between agents must not be accessible to unauthorized parties. Encryption methods (e.g., symmetric and asymmetric encryption) are used to ensure confidentiality.
  2. Data Integrity:�Information exchanged between agents must remain unaltered and arrive correctly. Hash functions (e.g., SHA-256, MD5) are used to detect any changes in the data.
  3. Data Availability:�Information exchanged between agents must be continuously accessible. Protection methods against DDoS (Distributed Denial of Service) attacks are applied to maintain availability.
  4. Authorization and Authentication:�Each agent’s access rights to the system and its ability to access controlled information are verified. Public Key Infrastructure (PKI), digital certificates, and passwords are used for this purpose.

10

11 of 19

The main issues of information security in multi-agent systems are:

1. Communication Security:�Communication between agents must be secure. To address this:

Encryption: Ensures that data exchanged between agents remains protected.

Digital Signature: Verifies the author of the data and checks that it has not been altered.

2. Confidentiality Protection:�Data exchanged between agents must remain confidential. To achieve this:

Symmetric and asymmetric encryption algorithms are used.

If necessary, data exchange between agents can be restricted.

3. Coordination Security:�Coordination among agents must be carried out securely. To ensure this:

Protocol Protection: Communication protocols between agents (e.g., FIPA ACL, SOAP) are verified to operate securely.

Logic Protection: The coordination logic between agents is ensured to function securely.

4. Resource Protection:�Agents must securely manage access to resources. To achieve this:

Resource Access Policy: Specifies which resources each agent can access.

Resource Monitoring: Prevents agents from misusing resources.

11

12 of 19

Methods of Information Security in Multi-Agent Systems:

1. Encryption:

  • Symmetric Encryption: Uses the same key for both encryption and decryption of data (e.g., AES, DES).
  • Asymmetric Encryption: Uses two types of keys (private and public) for encryption and decryption (e.g., RSA algorithm).

2. Digital Signature:�Used to verify the author of the data and ensure it has not been altered (e.g., RSA, DSA).

3. Hash Functions:�Used to check the integrity and authenticity of data, ensuring it has not been modified (e.g., SHA-256, MD5).

4. Firewall and Intrusion Detection Systems (IDS):�Used to protect data exchanged between agents from external threats.

5. Protocol Protection:�Ensures that communication protocols between agents (e.g., FIPA ACL, SOAP) operate securely.

12

13 of 19

Multi-Agent Insurance Detection System (SPIIRAN)

13

14 of 19

Functions Performed by Demon Agents (Agent-Demons, AD) in MAS:

14

  • AD-E (AD-Event): Initial processing of incoming messages.
  • AIA (Agent-Identification / Authentication): Identifies the sources of messages and verifies their authenticity.
  • ACA (Agent-Confidentiality Access): Restricts user access to network resources.
  • AD-P1 and AD-P2 (AD-Patterns): Detect suspicious events or attack indicators.
  • IDA-1 and IDA-2 (Intrusion Detection Agents): Intelligent agents that detect attacks and make decisions on responding to them.

15 of 19

Multi-Agent System for Computer Network Security Analysis

15

16 of 19

Software Tools for Developing Multi-Agent Systems

16

Java programming language; provides a library of software modules and communication protocols; supports FIPA standards.

  • JADE (Java Agent Development Framework) – Telecom, Italy
  • AgentBuilder – IBM, USA
  • JACK Intelligent Agents – [No specific country mentioned, widely used framework]
  • Bee-gent – Toshiba, Japan
  • IDK (INGENIAS Development Kit) – GRASIA, Spain
  • MAS-DK, SPIIRAS – Saint Petersburg, Russia

17 of 19

JADE

17

  • Application Areas: Mobile networks, Internet, industrial applications.
  • Integration: Java EE (JMS, Web, …), CORBA, XML.
  • Tools for Agent and MAS Development:

Runtime environment where agents can "live";

Library of classes to simplify agent development;

Set of graphical tools for managing and monitoring the activities of running agents.

  • Implementation Features:

Agent description language – Java;

Inference engines available via BPI and FSM plugins;

Library of protocols compliant with FIPA standards;

Ontology support;

Semantic Web and web service plugins.

18 of 19

Conclusion:

In multi-agent systems, information security is extremely important because the data exchanged between agents can be critical. By using encryption, digital signatures, and hash functions, the information shared among agents can be secured. To protect against external and internal threats, firewalls, intrusion detection systems (IDS), and other methods are employed.

18

19 of 19

Additional information

19

  1. Ustyujanin, A.E. Multi-Agent Intelligent Systems: Course. Moscow: MIPT, 2007. http://window.edu.ru/resource/434/57434
  2. Shvetsov, A.N. Agent-Based Systems: From Formal Models to Industrial Applications. Vologda State Technical University, 2008. – 101 p.
  3. Gorodetskiy, V.I., Grushinskiy, M.S., Khabalov, A.V. Multi-Agent Systems (Overview). 18.10.2015 // S.P. website. Kurdyumova http://spkardyumov.ru/networks/mnogoagentnye-sistemy-obzor/