Secure Cooperation in Autonomous Mobile Ad-Hoc Networks Under Noise and Imperfect Monitoring: A Game-Theoretic Approach

Abstract

In autonomous mobile ad-hoc networks, one major challenge is to stimulate cooperation among selfish nodes, especially when some nodes may be malicious. In this paper, we address cooperation stimulation in realistic yet challenging contexts where the environment is noisy and the underlying monitoring is imperfect. We have first explored the underlying reasons why stimulating cooperation under such scenarios is difficult. Instead of trying to enforce all nodes to act fully cooperatively, our goal is to stimulate cooperation in a hostile environment as much as possible through playing conditional altruism. To formally address the problem, we have modeled the interactions among nodes as secure routing and packet forwarding games under noise and imperfect observation, and devised a set of reputation-based attack-resistant cooperation strategies without requiring any tamper-proof hardware or central banking service. The performance of the devised strategies has also been evaluated analytically. The limitations of the game-theoretic approaches and the practicability of the devised strategies have also been investigated through theoretical analysis and extensive simulation studies. The results have demonstrated that although sometimes a gap may exist between the ideal game model and the reality, game-theoretic analysis can still provide thoughtful insights and useful guidelines when designing cooperation strategies.