Research On Service Discovery In Mobile Ad Hoc Networks

Nodes in mobile ad hoc networks are required to communicate with each other without a priori knowledge, so it is a key problem to automatically discover the available services in the network. As the mobile ad hoc networks are characterized by self-organization and dynamic network topology, it is a challenging issue to efficiently discover the services in such environments. In this paper, an efficient service discovery solution is proposed for mobile ad hoc networks, which combines the DHT technology with the location information of nodes.Firstly, a novel data sharing solution is proposed for mobile ad hoc networks, namely LDHT, which combines the basic idea of distributed hash table (DHT) with the geographic routing technologies of mobile ad hoc networks. Compared with traditional DHT schemes, LDHT is built upon the physical topology of the network rather than the logical overlay, so it eliminates the topology mismatching problem that exists in all traditional DHT schemes, and greatly improves the efficiency of data sharing in mobile ad hoc networks.Secondly, a dimension mapping solution is proposed so that the multiple-dimension service description information can be mapped to the two-dimension hash key space of LDHT. In such a way, the service discovery problem is converted into the data sharing problem on LDHT, so the efficiency and reliability of service discovery can been improved.Thirdly, a regional search algorithm is proposed for LDHT to efficiently support the conditional query of service description information. As the users in mobile ad hoc networks usually lack the enough information, they need to search for available services based on limited information. The proposed algorithm can efficiently search out all data items that satisfy the given condition so that it can well support the requirement of service discovery.Finally, the proposed solution is implemented with simulation and the performance of the solution is extensive evaluated. Simulation results shows that the proposed solution is not only more efficiency but also achieves better scalability.