Research On GA Based Self-localization Algorithm In Wireless Sensor Networks

Wireless Sensor Network(WSN), which is a new network-technology, can be widelyapplied in agriculture, industry, city management, rescue and relief work. Nowadays, it isbecoming a hot topic in wireless communication researches, among which the sensorself-localization is a fundamental and crucial issue for WSN. Accurate self-localizationcapabilityishighlydesirablein wirelesssensornetwork.Firstly, the thesis gives a brief introduction of the localization algorithms in WSN.Secondly, the theory of localization algorithms, which includes the basic principle,classification and the performance evaluation of localization algorithm is introduced. Thisthesis studies thetypical localizationalgorithms infourclasses.Becausethenode localizationis an optimization problem in nature, the four optimization algorithm-based localizationalgorithms, which are Genetic Algorithm based localization(GAL), Simulated Annealingbased localization(SAL), Evolutionary Strategies based localization(LESS) and DifferentialEvolution based localization(RCDE) are focused on. The pseudo-code or procedure of eachlocalization scheme is introduced and analyzed in details. The advantage and defect of eachlocalization algorithm are comprehensively summarized. Then two relevant solutions arepresentedbasedon the analysisofdefects.Thefirstone isadoptingthenearest3anchorswithrespect to each individual unknown node in localization algorithm, which can reduce thealgorithm complexity and improve the localization precision. Meanwhile, the anchorinformation is attached with a weight which is inversely proportional to the hop countbetween unknown node and anchor node, because the distance measurement error isbecoming larger with the increasing of hop count. The weight can reduce the impact ofdistance measurement error. The second one is introducing the Metropolis selection of SAinto GA, which helps improve the diversity of population and avoid GA running into local optimum solution. The improved optimization approach is called GSAfor short and it is usedfor localization in WSN. The simulation results reveal that GSAL(GSA Localization) is aneffective localization algorithm which uses fewer anchors, needs smaller transmission rangeof node, and is insensitive to measurement noise factor.So GSALnot only has a certain faulttolerance,butcanreducedeploymentexpenses.