Research On High-accuracy Cooperative Localization Technologies For Mobile Ad Hoc Networks

When establishing range-based high-accuracy localization systems in mobile ad hoc networks(MANETs),less anchor nodes is beneficial to decrease the energy consumption of localization systems.Cooperative localization scheme is applicable in MANETs with sparsely distributed anchor nodes,while the cooperations between agent nodes lead to the accumulation and propagation of positioning error.Existing researches indicate that probabilistic localization algorithms can mitigate the mentioned problems in cooperative localization scheme effectively.However,when constructing high-accuracy cooperative localization systems by combining cooperative localization scheme and probabilistic localization algorithms,new problems appear.One key problem is selecting a reasonable anchor node distribution in the initial phase and reselecting anchor nodes in the operational process of the localization systems.Another key problem is decreasing the computation complexity of the localization systems,while ensuring the performance on positioning accuracy.For the problem of selecting initial anchor nodes,the anchor node distribution,which is quite uniform and can represent the aggregation of nodes,can enhance the performance of high-accuracy cooperative localization systems.This paper proposes an initial anchor node selection scheme,in which nodes are clustered and anchor nodes are selected in each cluster,and the anchor nodes distribute approximate to the optimal distribution and quite uniform in each cluster.In the operational process of localization systems,to mitigate the influence of node mobility and partly ensure the energy balance between nodes,this paper proposes an anchor node reselection scheme,in which nodes are dynamically re-clustered into suitable clusters while moving,and an anchor node reselection process is executed in each cluster when needed.Experimental results indicate that the proposed anchor node distribution can achieve a better performance on positioning accuracy and positioning speed than uniform distribution and random distribution when applying NBP.The anchor node reselection scheme can maintain a well-clustered network structure and achieve a better performance on positioning accuracy and energy consumption balance than no reselecting anchor nodes.To establish cooperative localization systems with low-complexity and high-accuracy,this paper proposes a NBP-based low-complexity probabilistic localization algorithm,in which a simplified particle generation rule is proposed and considering a practical ranging error model,a more reasonable weight updating method is proposed.Also,the useful information of last iteration is utilized to accelerate the whole localization process,namely the computation complexity is reduced.Moreover,a more strict convergence condition is applied,considering the limitation on the reference nodes choosing of agent nodes,the communication traffic in the network is alleviated,while the positioning accuracy is enhanced since the accumulation and propagation of positioning error is significantly restrained.Experimental results demonstrate the efficiency of the proposed algorithm in increasing the performance on positioning accuracy and computation complexity,and reveal the outperformance of the proposed algorithm comparing with NBP.