Research Of Replica Allocation Algorithm In Mobile Ad Hoc Networks

Data replication is an important way to improve the system performance and reliability, which is broadly used in database system, files system, operation system, and distributed system. In mobile ad hoc networks (MANET), wireless communication bandwidth is limit, link failure and node failure often happen and even network is partitioned. Data replication in such a dynamic resource-limited environment is a significant challenge. In this thesis, in order to improve the system performance, four replica allocation algorithms for MANET environment are proposed.Firstly, nodes move freely and the distribution of access requests changes dynamically in MANETs. A dynamic adaptive replica allocation algorithm (DARAM) that can adapt to the nodes motion is proposed to minimize the communication cost of object access. When changes occur in the access requests of the object or the network topology, each replica node collects access requests from its neighbors and makes decisions locally to expand replica to neighbors or switch the replica with neighbors or to contract the replica. The algorithm dynamically changes the replica allocation scheme towards a local optimal one. Simulation results show that DARAM algorithm efficiently reduces the communication cost of object access in MANETs.Secondly, the DARAM Algorithm based on the Mobility Prediction (DARAM-MP) is proposed to reduce the oscillation of replica allocation by reason of nodes move quickly. The mobility prediction method based on link expiration time is utilized to choose stable neighbors and to expand the replica to relatively stable nodes. Our simulations show that DARAM-MP algorithm efficiently reduces the oscillation of replica allocation in MANETs.Thirdly, the battery capacity of nodes is limited in MANETs. The replica allocation should consider the energy cost that influences the system performance evidently. A power-aware dynamic adaptive replica allocation algorithm (PDARAM) that used a power control method is proposed to minimize the energy cost of network. Each replica node collects access requests with energy information from its neighbors and makes decisions locally to allocate replica to nodes. Simulation results show that PDARAM algorithm efficiently reduces the energy cost of network in MANETs.Finally, in order to balance the consumed energy of nodes, a dynamic adaptive replica allocation algorithm based on the maximum network lifetime (MDARAM) is proposed. Each replica node utilizes a weighted method based on remnant energy to compute condition and makes decisions locally to allocate replica to nodes that have more remnant energy. The results of simulation show that MDARAM algorithms efficiently prolong the lifetime of network in MANETs.