Abstract:
The random deployment of Wireless Sensor and Actuator Network (WSAN) nodes in areas often inaccessible, results in so-called coverage holes – i.e. areas in the network that are not adequately covered by nodes to suit the requirements of the network. Various coverage protocol algorithms have been designed to reduce or eliminate coverage holes within WSANs by indicating how to move the nodes. The effectiveness of such coverage protocols could be jeopardised by inaccuracy in the initial node location data that is broadcast by the respective nodes. This study examines the effects of location inaccuracies on five sensor deployment and reconfiguration algorithms – They include two algorithms which assume that mobile nodes are deployed (referred to as the VEC and VOR algorithms); two that assume static nodes are deployed (referred to as the CNPSS and OGDC algorithms); and a single algorithm (based on a bidding protocol) that assumes a hybrid scenario in which both static and mobile nodes are deployed. Two variations of this latter algorithm are studied. A location simulation tool was built using the GE Smallworld GIS application and the Magik programming language. The simulation results are based on three above-mentioned deployment scenarios; mobile, hybrid and static. The simulation results suggest the VOR algorithm is reasonably robust if the location inaccuracies are somewhat lower than the sensing distance and also if a high degree of inaccuracy is limited to a relatively small percentage of the nodes. The VEC algorithm is considerably less robust, but prevents nodes from drifting beyond the boundaries in the case of large inaccuracies. The bidding protocol used by the hybrid algorithm appears to be robust only when the static nodes are accurate and there is a low degree of inaccuracy within the mobile nodes. Finally the static algorithms are shown to be the most robust; the CPNSS algorithm appears to be immune to location inaccuracies whilst the OGDC algorithm was shown to reduce the number of active nodes in the network to a better extent than that of the CPNSS algorithm. Copyright
