An Efficient Checkpointing Scheme for Mobile Computing Systems

In this paper, an efficient checkpointing scheme is proposed for mobile computing systems. It uses communication-induced checkpointing with the exception that basic checkpoints are not taken. A new concept of forced checkpoints, different from the existing one, is incorporated to ensure a reduced reexecution after recovery. The proposed algorithm runs periodically to find the set of the globally consistent checkpoints (GCC) as in the synchronous approach, instead of searching for the set after recovery from a failure as in the asynchronous and communication-induced approaches. Thus, the recovery is as simple as in the synchronous approach.