Abstract
Personal and Body Area Networks (PAN/BAN) are spontaneously formed through the participation of a number of devices, using wireless technologies. They are considered to be components of larger Personal Networks (PN), which include clusters formed at a user's points of interest. In such a dynamic environment, a Service Discovery (SD) process is required for the detection of participating devices and their offered services and capabilities. In the course of this research, SD requirements in wireless personal networks were identified and the suitability of some of the existing protocols was investigated. Contributions of the research are listed below. 1) New PN Service Discovery and Management Platform (PNSDP & PNSMP): This is a generic platform that considers the characteristics of PNs, and is inter-operable with Bluetooth Service Discovery Protocol (SDP), Simple Service Discovery Protocol (SSDP) of Universal Plug and Play (UPnP) and Service Location Protocol (SLP). Furthermore, a new solution for enabling IP Multimedia Subsystem (IMS) and Service Session Management in PNs was proposed. 2) Optimised structure for PNSDP: An analytical evaluation was performed to aid in the proposal of the optimised structure for PNSDP. The results suggest that the most appropriate solutions are registry-based discovery within the clusters, peer-to-peer discovery in the infrastructure network, and the use of the binary format for SD messaging. 3) Interoperability between SD legacy protocols within PNSDP: The results of an experimental evaluation of the performance of the Bluetooth SDP suggest that it is better to use the search operation instead of the browse. In ad hoc networks, the SD time and the success rate (SR) of the SSDP were evaluated (through computing simulations) employing different levels and types of background traffics, showing the non-sensitivity of the SD time to congestion. A new (single- and multi-PAN) mobility model was proposed and the SLP was evaluated in terms of the SR, SD time and cost. One of the important results in using registry-based SLP was the increasing of the SR by ~60% at an energy cost increase of ~25% in the multi-PAN case. 4) Enhanced SLP (ESLP): Three new techniques were proposed and evaluated: sleep mode for the service nodes (ESLPi), multi-tier discovery for the convergence algorithm (ESLP2), and scope-based clustering (ESLP3). The achieved results indicate a reduction in energy expenditure of up to 3-5%, as well as a lower SD time in ESLP1 and ESLP3 and a higher SR in ESLP3.