A Fast Localization Method Based On Virtual Nodes In Wireless Sensor Network

As a rapid development of network information technology, Wireless sensornetworks (WSNs) have been widely used in different areas such as environmentalmonitoring, intelligent transportation, personal rescue and target tracking.Localization technology is the premise and basis of these applications. Therefore ithas very important significance in both theoretical and practical to study the wirelesssensor network localization technology.This dissertation firstly introduces the wireless sensor networks research status,summarizes the current mainstream localization technologies of WSNs and makes acomprehensive summary of the advantages and disadvantages about differenttechnologies. On this basis, we propose an integrated fast localization algorithm baseon the least squares (LS) localization algorithm and non-metric multidimensionalscaling (NMDS) localization algorithm aim at solving the problem that localizationaccuracy and efficiency are often not take into account. LS and NMDS are two typicallocalization methods in WSNs. Given some beacon nodes, LS implementslocalization by least squares optimization with high efficiency. But it has pooraccuracy when the number of beacon nodes is sparse. In real WSNs, it is oftenimpossible to set too many beacon nodes that leading to LS algorithm infeasible.NMDS reconstructs the relative positions of nodes with good accuracy even thoughthere are few beacon nodes. But it has high time complexity. To find a practicallocalization method that can be implemented in sparse beacon nodes WSNs, wepropose a fast localization method that first builds a few virtual beacon nodes byNMDS and then locates by LS.In order to verify the effectiveness of the method we make experiments indifferent topology network and compare the LS, NMDS and proposed method. Theexperiments results show that proposed method can achieve fast localization of sensornodes with smaller loss in accuracy.