Dynamic Prediction of Data Validity for Mobile Databases

Optimistic replication which does not lock resources and can tolerate disconnections appears to be the natural choice for mobile computing environment. The replication approach allows updates to be performed independently at any replica, so the fundamental problem in this scheme is that data at any replica has the probability of invalidation. To resolve this problem, this paper presents a new dynamic prediction of data validity algorithm (DPDVA), which predicts the validity of data replica stored in mobile nodes by local history information. Experimental results show that DPDVA is an effective prediction algorithm to maintain database consistency. The more local information mobile nodes store, the better the prediction effects will be.