Research On Service Discovery Protocol In Pervasive Computing Environments

Pervasive computing environment is a kind of physical environment which is infiltrated by various computing components and communicating techniques. It is composed of handheld, wearable, and embedded computers in addition to regular desktop clients and servers. These are connected by some combinations of wireless ad hoc networks and wireless infrastructure-based networks. Service discovery protocol is the technology of automatically finding services matching one's needs in the network. Pervasive computing environment is heterogenous and dynamic, so service discovery protocol must be high-autonomous, and can detect dramatic changes of useful resources in time and provide needed services and information at any time without human intervention. Therefore efficient and robust service discovery protocols with low overhead cost and good expansibility are urgent demands of service discovery in pervasive computing environments.Researching and comparing these main existing service discovery protocols find that GSD(Group-based Service Discovery Prototcol) is an excellent performance service discovery protocol in pervasive computing environments. GSD combines service advertisement and request-broadcast together, service advertisements are based on the concept of peer-to-peer caching, service requests are intelligently forwarded to other nodes in light of service groups. In this way, flooding of service request packets can be avoided. Despite the fact that the intelligent forwarding technique in GSD is referential experience, GSD still has some problems. In order to avoid large redundance service request forwarding packets because of inaccurate intelligent forwarding in GSD, SIGIFSDP(Service Id Guided Intelligent Forwarding Service Discovery Protocol) is proposed. In SIGIFSDP, when forwarding nodes are selected, not only must the service group be matched, but also must the service information guide direction carefully. In this case, the forwarding nodes are the exact nodes to which the service request packet should be transmitted. False forwardings can be avoided. When no forwarding nodes, instead of broadcasting the service request, SIGIFGSD use FFP(Flexible Forward Probability)to reduce linely the forwarding probability along with the further forwarding of service request packet. This can ensure spreading range of service request as well as redusing packet redundancy. With GloMoSim simulation, it is verified that SIGIFGSD has fewer number of service request packets, shorter response time, and higher efficiency of service discovery.In order to exhaustively utilize the cached information and further reduce the number of the forwarding nodes, FNMESDP (Forward Node Minimization Enhanced Group-based Service Discovery Protocol) is proposed. FNMESDP optimizes its request packet forwarding operation in two aspects: 1) tries to reduce the number of nodes that are considered when making service request packet forwarding decisions by exhaustively utilizing the cached information in node's SIC(Service Information Cache); 2) finds a minimum CFNS (Coverage-holded Forward Node Set) meanwhile preserving the coverage of service discovery sessions, and only nodes in CFNS are selected as valid receivers. In FNMESDP, a heuristical algorithm about CFNS and an algorithm forwarding the service request packet are given. The time complexity about forwarding operation is analysed carefully. And the theoretical analysis has verified the extended-coverage preservability of FNMESDP. With GloMoSim simulation, it is verified that FNMESDP can reduce the packet overhead largely, save the response time, and improve the efficiency of service discovery.In order to improve the efficiency of subsequent service discovery by exhaustively utilizing the service reply information, ASESDP(AIP and SRR Enhanced Service Discovery Protocol) is proposed. In ASESDP, two schemes are used to optimize its performance: AIP (Advertisement Information Piggybacked) and SRR (Shortest Reply Route). In AIP, the reply packet carries not only the reply information but also parts of advertisement information. So with the transmission of service reply packet, the advertisement information is spread along the reply path as well as, and cached in SIC of every node on the path. Therefore the AIP scheme can widen the transmission range of the advertisement information. And in SRR, the shortest one among all reply routes can be chosen to forward the reply packet to the source node, which can save the response time. By this way, while the maximum hop number of service advertisement packet is limited, ASESDP can still decrease packet load, improve the efficiency of service discovery.GloMoSim is small and fast. It has good extendability, and is mainly used in wireless network. But the simulation of service discovery protocol can not be performed in GloMoSim. So it is needed to expend the simulation framework of GloMoSim to carry out the service discovery simulation, which is composed of configuration files expanding, application layer expanding and network layer expanding. In order to avoid tiresome manual work and improve the simulation efficiency, automatization of multi-scheme comparison and batch simulation is implemented in GloMoSim. The detailed description of the structure and framework of simulation which carries out service discovery makes the research much clear and perspicuous.