Discover Your Island University

Graduate Projects


Project ID: 326
Author: Bo Dong
Project Title: A Novel SemiDefinite-Based Localization System: Design and Evaluation
Semester: 3 2009
Committe Chair: Dr Ahmed Mahdy
Committee Member 1: Dr John Fernandez
Committee Member 2: Dr Mario Garcia
Project Description: This research project focuses on localization in Wireless Sensor Networks (WSN). State-of-the-art localization research and approaches have been thoroughly reviewed to illustrate the challenges imposed by inaccurate distance measurements and noisy backgrounds. Location awareness and distance estimation cannot be underestimated in WSN especially for indoor applications. Compared to existing localization approaches, the SemiDefinite Programming (SDP) approach delivers accurate distance measurements even in hostile backgrounds. In this project, three different SDP models, namely Full SemiDefinite Programming (FSDP), Edge-based SemiDefinite Programming (ESDP), and ESDP with Noise Factor are investigated and their performances have been evaluated by computer simulations using different metrics including radio range, anchor number, noise factor, and global error rate. In addition and based on SDP, three novel localization schemes, namely Flexible-Anchor, Flexible-Radio-Range, and Magnified-Range, are proposed. The new schemes allow indoor systems to survive even in hostile environments. The Flexible-Anchor scheme addresses the case when an anchor malfunctions due to an external force, which is typical for extreme environments. Flexible-Anchor ensures the existence of alternate nodes that can guarantee an undisrupted operation in the event of such node failure. The Flexible-Radio-Range scheme is designed for the situation when some transmitter nodes fail to cover an intended area. In that case, connectivity of the system may be reduced dramatically. This can lead to poor accuracy of position estimation or even disable the whole system. Using Flexible-Radio-Range, the connectivity of the system is maintained by controlling nodesítransmission range. Compared to the other two schemes, Magnified-Range is mainly for normal situations. It is based on the fact that anchors are special nodes that suffer less from energy constraints. Therefore, it is possible to magnify their radio range without impacting the energy consumption of the network. Magnified-Range differentiates between the radio range of anchors and regular sensor nodes. Undoubtedly, Magnified-Range still works well under extreme situations. Performance evaluation and simulation results have shown that the three schemes offer robust localization even in a hostile situation. In fact, the proposed schemes improve localization accuracy by at least 20%. Future research will focus on three main challenges, namely 3-D simulations, hybrid localization systems, and real-life demonstrations.
Project URL:   326.pdf