Towards fault tolerance for multiagent systems

ABSTRACT

There is a growing need for improving the fault tolerance in multiagent systems (MASs). In this thesis, we investigate it in two parts. The first part presents a monitoring approach which detects whether the design and implementation of a MAS is correct. The second part introduces agent-oriented architectures which aim to maximise multiagent survivability. We then present several heuristics to measure the survivability of a given MAS.

REFERENCES

• 1: Jürgen Dix , Thomas Eiter , Michael Fink , Axel Polleres , Yingqian Zhang, Monitoring agents using declarative planning, Fundamenta Informaticae, v.57 n.2-4, p.345-370, October 2003
• 2: V. S. Subrahmanian, S. Kraus, and Y. Zhang. Distributed algorithms for dynamic survivability of multiagent systems. In Dix and Leite, editors, LNCS 3259. Springer. 1--15, 2004.
• 3: Y. Zhang, E. Manister, S. Kraus, and V. S. Subrahmanian. Approximation results for probabilistic survivability. Proc. of the Second IEEE Symposium on Multi-Agent Security and Survivability. to appear, 2005.
• 4: S. Kraus, V. S. Subrahmanian, and N. C. Tacs. Probabilistically survivable MASs. Proc. of IJCAI-03. 789--795, 2003.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
        I.2.11 Distributed Artificial Intelligence
            Subjects: Multiagent systems

General Terms:
Algorithms, Performance, Reliability

Keywords:
heuristics, monitoring, replication, survivability