Cluster-based coordinated checkpointing protocol in wireless ad-hoc networks

The intrinsic characteristics of mobile communication system render them to be more prone to faults. Hence it is imperative to equip such devices with some fault-tolerance mechanism so as to run a reasonable application on them. Fault tolerance is intrinsically essential to wireless ad-hoc networks where the mobile devises are vulnerable to physical damage, theft etc. or can be exposed to radiations. Rollback recovery is one of the extensively adopted methods used to induce fault-tolerance in the distributed systems. Checkpointing is low cost fault tolerance techniques that can achieve fault tolerance transparently even against unanticipated faults. Mobile ad-hoc networks can be categorized as infrastructure-based and infrastructure-less (ad-hoc) networks. In infrastructure based networks, mobile hosts are supported by mobile support stations (MSS). In infrastructure-less ad-hoc networks, mobile hosts are not supported by any mobile support stations (MSS), which makes it very difficult for checkpointing techniques to implement. In this paper the cluster based coordinated checkpointing protocol is proposed that has minimum checkpointing overheads and does not require extra synchronization message and blocking.