Designing Dependable Mobile Computing Systems with Mobile Agents

Dependability is an important metric for evaluating system performance of both wireless networks and mobile agents operating on these networks. This book deals with dependability analysis of mobile agent based mobile computing system in wireless networks and investigates ways of improving it. Here dependability is viewed as a combination of availability and reliability while security affects both. The effectiveness of agent paradigm over message passing is also shown in terms of system reliability. In order to avoid the huge computational complexity arising out of dynamic topology, all dependability estimation models are developed using Monte Carlo simulation method. It is observed that agents can effectively operate even when the underlying network is unreliable. Effect of environmental parameters on dependability analysis of the system is shown which are not encountered in traditional reliability analysis. Distributed trust based reputation schemes are proposed to protect the nodes and agents from both active and passive attacks. Further, reinforcement learning techniques are applied to the mobile agents for deciding about a suitable migration strategy in hostile conditions.