Power-aware optimal checkpoint intervals for mobile consumer devices

It is highly desired to employ a checkpoint and rollback scheme on a mobile consumer device for fault tolerance, because spatial fault-tolerant schemes such as hardware replications cannot be used. We can reduce the loss of computation of a process in the presence of failures by periodically saving the process´s state on stable storage as a checkpoint and rolling back to the latest checkpoint when a failure has occurred. However, a mobile consumer device is not considered to have sufficiently large and stable storage to store its checkpoint data. Therefore, a remote checkpoint technique is preferred in wireless environments in which the checkpoint data of a mobile device is kept in a remote checkpoint server instead of the mobile device. Dense checkpoints incur frequent wireless transmissions whereas coarse checkpoints increase the loss of computation. Many checkpoint research papers have tried to minimize the expected execution time. However, an effective solution which minimizes the energy expenditure should be also provided, because battery power is one of the most critical resources of a mobile device. In this paper, we propose the energy-aware optimal checkpoint interval in wireless remote checkpoint environments. We develop a stochastic model with Markov chain and then derive the optimal value. For the practical evaluation, we conducted not only analytical estimation and simulation but also experimental measurements by implementing on a real test-bed.¹.