A Novel Approach to Balance the Load in Distributed Mobile Environment

A recent topic in mobility management research field is mobility prediction. Mobility prediction can be defined as the prediction of a mobile user’s next movement where the mobile user is traveling between the cells. The predicted movement can then be used to increase the efficiency of Personal Communication Systems. In mobile web environment, it is often desirable to effectively mine multilevel and location-aware associated service patterns. However, the computational complexity of the underlying problems that locations and services are of hierarchical structures and if the source logs are very large, the overhead in integrating the data source log will be too high, essentially render sequential algorithms useless. Under this circumstance the more efficient approach will be mining all the logs in distributed way. That is, the distributed mining algorithm such as two dimensional multilevel methods should be executed on each server concurrently and the records of service request by the same mobile user will be interchanged via networks that can be efficiently used to discover the mobile user’s associated service request patterns. But this approach is based on dynamic load balancing in the sense that load varies during the execution of the processes. Due to the limitations of static load balancing, a peer-to-peer communication framework and a novel receiver-initiated dynamic load balancing algorithm is adopted. The effectiveness of the distributed method has been evaluated based on the dataset from the mobile web environment.