Research On Checkpoint Handoff Scheme In The Mobile Computing System

The mobile computing network has broad application foreground due to its characteristics including dynamic mobility, celerity of setting, autonomy, flexibility of topology and equivalence. Compared to the wired distributed computing system, the mobile computing system has certain new characteristics such as low wireless bandwidth, limited memory and stable storage in the mobile host, limited battery life and dynamic mobility of the mobile host. The fault probability of the computing process is increased greatly along with the enlargeing scale of the computing system. But the traditional fault tolerant schemes for the distributed computing system are not appropriate for the mobile computing system.Researches on the fault tolerant schemes for mobile computing systems have received tremendous interests by national and international experts in recent years. Mobile support stations are usually used as the reliable storages in the fault tolerant schemes of mobile computing system as there is limited and unreliable storage in the mobile host. Due to the mobility of the mobile host, the related recovery information may scatter on different mobile support stations when the mobile host moves from one cell to anther. Hence an efficient recovery information management strategy is required to trace and retrieve the recovery information scattered on different mobile support stations.There are three existing checkpoint handoff schemes named as Eager, Lazy and Trickle strategy. These schemes mentioned above can not do well in the efficiency of failure-free execution of mobile computing and the fast recovery after a mobile host's fault.This thesis introduces a new virtual efficient checkpoint handoff scheme, in which the recovery information of a mobile host is partitioned into two parts, including the high-priority and the low-priority parts, and the partition principle is figured out. To utilize this new virtual efficient handoff scheme, the rollback recovery and other sections of the common checkpoint recovery scheme in mobile computing system are adjusted. And then the recoverable consistency of this new virtual efficient handoff scheme is discussed. The fault model of the checkpoint rollback recovery in the mobile computing system is introduced in the performance analysis section. Based on this model, the performances are compared among the new proposed and other checkpoint handoff schemes. The performance simulation of this new management shows the system gets the efficient failure-free execution of mobile computing and the fast recovery after the mobile host's faults. The new handoff scheme makes the mobile computing system achieve more reliable and efficient performance.