Replica Allocation Considering Data Update Intervals in Ad Hoc Networks

In ad hoc networks, since network division occurs frequently, it is effective to replicate data items. This paper proposes effective replica allocation methods in ad hoc networks where each data item is updated at irregular intervals. The proposed methods allocate replicas based on probability density functions of the update intervals of data items. Also, they invalidate replicas that have been updated with high probability because accesses to old replicas impose extra computational overhead and roll backs. As a result, the proposed methods not only improve data accessibility but also reduce the number of accessing old replicas.