A Markov Chain Monte Carlo Approach to Perform Global Optimized Resource Replication over Ad Hoc Networks

Resource replication has been emerging as one of the most widely studied phenomena in a distributed environment. In this paper, we propose a novel distributed approach, named MetroReplica, to achieve the global optimized resource replication. MetroReplica is a Markov Chain Mote Carlo sampling algorithm based on the metropolis-hastings algorithm, which never employ the global information but a replication query message transmitted via a limited hops link path. Such message is generated and transmitted to the mobile node to be finally assigned with the allocating replica within at most TTL hops forwarding following the proposed Markov transition matrix. The proof to verify MetroReplica is illustrated in this paper and simulation results suggest a small TTL value, e.g. TTL = 10, is sufficient for an implementation in reality to perform well.