Research On Localization Algorithms In Wireless Sensor Networks

With the rapid development of sensor technology,wireless sensor network(WSN)has received broad application prospects in many fields.Considering that most applications in WSNs are based on the location of sensor nodes,it is of great practical significance to study WSN localization algorithms.This dissertation aims at overcoming several key problems in WSN localization,including reducing algorithm complexity and network power consumption,identifying and mitigating non-line-of-sight(NLOS)errors and enhancing positioning system robustness,and proposes some corresponding algorithms in order to further improve the performance of WSN localization algorithm.Firstly,a node selection algorithm based on software-defined networking(SDN)architecture is proposed.Under the condition of limited total power consumption of the network,a group of reference nodes with the most advantageous to the location results are selected for each localized node,so as to reduce the algorithm complexity and energy consumption.The analogous Cramer-Rao lower bound(A-CRLB)of each reference node for a specific localized node is calculated as a measure of the contribution of the reference node to the location results.Based on the characteristics of SDN network structure,a centralized node selection algorithm is proposed in both non-cooperative and cooperative localization scenarios.The node selection problem is constructed into a 0-1 programming problem with total power constraints,and a set of reference nodes is selected to maximize the total contribution.In addition,in the proposed SDN-based node selection algorithm,an addressing method suitable for software-defined wireless sensor network(SD-WSN)is defined,which is used in the southward protocol interface between the control plane and the data plane.Simulation results show that the proposed algorithm can effectively balance the accuracy,complexity and energy consumption of WSN localization algorithm.Secondly,an NLOS identification and mitigation algorithm based on AdaBoost is proposed to eliminate or suppress the influence of NLOS errors caused by the obstacles in WSN localization scene.The NLOS identification algorithm extracts the signal characteristics of propagation paths,and makes use of the machine learning algorithm named AdaBoost to train a strong NLOS classifier from a series of weak learners,which is used to distinguish line-of-sight(LOS)paths from NLOS ones.After identifying NLOS paths,a link condition indicator(LCI)is proposed to measure the degree of NLOS impacts on different links and an N-probabilistic hard weight(N-PHW)algorithm is designed to modify the traditional sum-product algorithms over a wireless network(SPAWN)into the algorithm with the ability to resist the influence of NLOS.Simulation results show that the AdaBoost-based NLOS strong classifier can effectively identify NLOS paths,and the modified SPAWN algorithm can suppress the NLOS errors and improve the positioning accuracy.Then,an incentive algorithm based on economic model for cooperative localization is proposed,whereby a pricing scheme is designed to lead each agent to the optimal state from a profit perspective,thus encouraging the agents in the network,especially those in a better network state,to actively participate in the cooperative localization.The proposed incentive algorithm is divided into two parts: budget decision and price allocation.In budget decision,an optimization problem based on game theory is constructed,in which each agent is regarded as a participant in the game whose utility function is represented as a profit-related function.The utility function is transformed into a concave objective function via Dinkelbach method.A per-pair negotiation(PPN)strategy is proposed to determine the optimal budget input for maximizing the profit of each agent.In price allocation,by introducing Shapley value to measure the contribution of each reference node,the budget of agent is fairly allocated to each reference node,so that the location accuracy can be optimized under the given budget.Simulation results show that the proposed incentive algorithm can make the agents in better network conditions more willing to participate in cooperative localization,thereby improving the overall positioning performance of the network.Finally,a hybrid algorithm based on received signal strength indicator(RSSI)and inertial navigation is proposed to improve the accuracy,reliability and continuity of the localization system by complementing the advantages of different positioning technologies.The RSSI values are used to estimate the location of the node preliminarily by the finger printing algorithm,at the same time,the inertial data provided by the inertial measurement unit(IMU)on the agent(including the acceleration,angular velocity and geomagnetic intensity of the current location)are collected.The attitude vectors such as pitch angle,roll angle and azimuth are calculated.By processing these information and RSSI data through Kalman filter,the dynamic change law of the agent positions is described by the state equations.A hybrid algorithm is proposed to correct the RSSI localization results.Simulation results show that the proposed hybrid algorithm can overcome the problems of using RSSI positioning or inertial navigation alone,and further enhance the robustness of the positioning system.