Wireless sensor networks, which integrates technologies of embedded computing,sensing, micro-electro-mechanism, distributed information processing and wirelesscommunication, is a new pattern of information gathering and processing. As animportant support technology for wireless sensor networks, the localization of sensornodes has become one of the main aspects in the field of wireless sensor networks. Thispaper carries on a thorough study for the localization of sensor nodes.First, the research background and status of wireless sensor networks aresummarized, the system structure and application fields are analysed, several kind ofclassical algorithm of node localization for wireless sensor networks are researchedsystematically, classification method and evaluation index of localization algorithm aresummarized.Secondly, combine the simple and easy to realize of the Centroid algorithm and thehigh localization precision of Taylor series expansion-based localization algorithm, anew localization algorithm based on an improved weighted Centroid algorithm and theTaylor series expansion-based localization algorithm is proposed. In view of the defectof large position error of the Centroid algorithm, a weight based on the value of RSSI issubjoined to anchor nodes to improve the position accuracy. In view of the shortcomingof low position rate when the density of anchor nodes is low, locate the normal nodeswhose position is unknown by the coordinate of the normal nodes whose position isknown to improve position rate. A positive definite diagonal matrix is designed toreduce the effect to the nodes localization resulted from the range measurement errorsincreases as the distance between two nodes increases. Use the localization results ofWeighted Centroid algorithm as the initial value of Taylor series expansion-basedlocalization algorithm to overcome the defect that Taylor series expansion-basedlocalization algorithm is sensitive to the initial value and improve the position accuracy.The simulation results from MATLAB show that the position accuracy of the newalgorithm is slightly higher than the position accuracy of Centroid algorithm and theposition rate of the new algorithm is much higher than the position rate Centroidalgorithm when the density of anchor nodes is low, the position accuracy of the newalgorithm is much higher than the position rate Centroid algorithm when the density ofanchor nodes is high. Finally, an Adaptive Particle Swarm Optimization algorithm based on ParticleSwarm Optimization algorithm is proposed. It follows the standards of the adaptabilityof organism and the variant with small probability of organism in the process ofevolution. Two improvements targeted to the Particle Swarm Optimization algorithmare proposed. First, in each iteration, particle in the swarm began to adjust its inertiaweight which is used in the next iteration adaptively based on the relationship ofadaptive value between the particle and the swarm to accelerate convergence rate andimprove position rate. Second, a mutation operation is operated on global optimallocation with certain probability is proposed to solve the problem that the ParticleSwarm Optimization Algorithm has a slow convergence speed and could trap the localminimum. Characteristics between Particle Swarm Optimization Algorithm andAdaptive Particle Swarm Optimization algorithm were compared based on the test bythe benchmark functions. The Adaptive Particle Swarm Optimization algorithm isapplied to the node localization of wireless sensor networks. The simulation resultsshow that this algorithm has rapid convergence rate, low-energy-waste, high positionaccuracy and good stability. |