Research On Checkpoint Storage Management Technology In Mobile Computing Environment

With the feature of mobility and flexibility, mobile computing environment has a more broad application prospects. It also has a narrow bandwidth and the wireless communication is vulnerable to interference. That is why system often causes failure. Therefore the introduction of fault-tolerant technology in the mobile computing environment is a research hot and difficulty, but also a challenging and developmental task at present.Checkpoint is a common fault-tolerance technology in a mobile computing environment.Checkpoint technology mostly uses reliable memory of the mobile support station to save the process system state into the checkpoint file. However this approach has some problems, the mobile nodes are connected to the mobile support station via a radio link, which led to a certain delay when the checkpoint is transmitted from the mobile station to the mobile host.The following will result in unreliable checkpoint transmission in case of network instability.Currently how to store checkpoint is a hot research and attracts more and more researcher’s attention.This paper studies checkpoint storage technology in mobile computing environment.First, in the analysis of traditional checkpoint storage methods and the two-level method of checkpoint storage, we put forward a decentralized checkpoint storage technology.Checkpoints are stored in local disks of the mobile host and the stable storage of neighbor mobile host in decentralized checkpoint storage technology. When the mobile host is subject to a permanent fault, the checkpoint is not required to transmit from a mobile station to a non-failed node to rollback recovery; instead the recovery is carried out in the neighbor nodes directly. This rollback recovery approach is more reliable and reduces the transmission delay of the checkpoint. Secondly, the paper also presents a checkpoint storage method based on load balancing. The system not only can shields failure, but also can makes the system tasks completed timely and correctly. That can ensure the system to achieve load balancing before and after failure. In this scheme we compute the performance index of mobile hosts using AHP. Firstly the heuristic greedy algorithm is used to allocate task statically and balanced.Then the checkpoints of each task on the mobile host are stored on the neighbor mobile host and local disk according to the task load and performance of the mobile hosts. This checkpoint storage technology has some advantages in terms of fault tolerance and load balancing. Finally, through simulation and modeling, we verify the correctness of the proposed algorithm in this paper.