Integrated Replication-Checkpoint Fault Tolerance Approach for Mobile Agents Systems

Abstract

Mobile agents are executing software entities that are capable of migrating from server to server in heterogeneous networks on behalf of network users. Agent systems are subject to failures that result from bad communication, breakdown of agent server, security attacks and congestion in networks. If any of such event happen, mobile agents suffer loss or damage totally or partially while execution is being carried out. For this reason, reliability must be addressed by the mobile agent technology paradigm in order to challenge such failures. Fault tolerance prevents a partial or complete loss of the agent. This ensures that the agent arrives at its destination correctly. Improving the survivability of mobile agents in presence of failures in unreliable network is an important issue in order to guarantee continuous execution of mobile agents. Many mobile agent-based fault tolerant approaches have been proposed so far. In spite of these efforts, the field suffers from setbacks in the form of persistent challenges. This thesis presents a novel fault tolerance approach, Integrated ReplicationCheckpoint Fault Tolerance approach (IRCFT), to detect agent failures as well as to recover services in mobile agent systems. Our approach is a development of that in FTIRC approach, where the agent put its computation results on the home server after completing its task on first three servers in its itinerary and PFTM approach , the idea of this approach is saving the viii checkpointing and logging information on each server and applying a witness agent to monitor if the actual agent is alive or terminated. IRCFT approach uses cooperating agents to achieve fault tolerance. It makes use of replicated agent. We have also added checkpointing, so as to limit the rollback of agent in case of failure of both the worker agent and all the updating replicas. IRCFT approach handles server and agent failure, it is capable of detecting and recovering most failure scenarios in mobile agent systems. In order to evaluate the performance of IRCFT approach we have implemented it on the Aglet mobile agent systems and evaluated it in terms of parameters such as total round trip time and agent survivability. To support our work, we compare our results with that proposed in FTIRC approach and PFTM approach; where our results were superior. The work shows that there is a distinguishable improvement in the performance and that it is a promising technique in achieving the reliability in mobile agent systems.