Enhancing service discovery performance over home networks

• Main Author: Al-Mejibli, Intisar
• Published: University of Essex 2012
• Subjects: 621.382
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.572894

The purpose of this study was to investigate how dropped messages can be avoided during the service discovery process of the Universal Plug and play (UPnP) protocol in home networks. Another aim involved determining the causes of those messages being dropped during the service discovery process, with many protocols being examined. The thesis proposes an algorithm that can compute the .,/]1tr)imum period of time .. required between a consecutive burst of services' reply messages and that determines the minimum required queue sizes for the routers' nodes to manage the traffic and avoid dropped messages. The proposed solution is applied at the application layer in' end nodes, not router nodes, as this would reduce the load on router nodes, freeing them up for other routing functions. The solution takes into consideration many of the network parameters such as network size, message size and existing traffic. An NS2 simulator was used to examine the produced algorithm on two different network configurations - centralized and decentralized - which was tested with three different types of queue management algorithms: drop tail, Random Early Detection (RED) and Random Exponential Marking (REM). In addition, its functionality was examined in a hybrid network, which included both wired and wireless connections. The results show an improvement in the discovery rate and a reduction in the number of dropped messages, highlighting the capability of the proposed algorithm in dealing with modifications to the network topology and in changing the density of the existing traffic. The principal conclusion was that the UPnP scales up well with its extensive use of messaging through managing the rate of replied services' messages.