Research On Location Technology Of Ultra-Dense Networks Based On Randomly Distributed Nodes

In recent years,as an important public security feature,wireless localization has been widely used in the fields of industry,agriculture,medicine and even national defense.With the rapid growth of the number of network devices,the localization of blind nodes in the 5th generation Ultra-Dense Networks(UDN)system is getting more and more attention,and the impact of node density and randomness on localization accuracy is also increasing.Therefore,how to make use of the distribution of nodes under UDN to obtain high-precision and low-cost location algorithms has become a hot research topic.For the localization problem in UDN,this paper studies the localization system under random distribution of nodes based on range-based and range-free localization technologies respectively.The main work includes:(1)Based on range-based localization technology,the two-dimensional Gaussian random distribution and Time of Arrival(TOA)ranging characteristics of reference nodes are used as a priori information for the first time,and Cramér-Rao Lower Bound(CRLB)based on this priori condition is derived.Theoretical analysis shows that the CRLB of reference nodes subjected to random distribution is smaller than the average CRLB of reference nodes subjected to fixed distribution.In addition,three locating algorithms are proposed based on Maximum Likelihood Estimate(MLE),including iterative,closed-form and hybrid-locating algorithms.Iteration can achieve good performance under a large enough reference node,but it may cause divergence of iteration under a poor geometric accuracy factor or a small number of reference nodes,resulting in poor localization performance.The closed-form solution has convergence,but the theoretical variance is greater than that of the iteration method.Therefore,a hybrid-localization algorithm is presented,which combines the iterative method with the closed-form method.Its core is to compromise the high localization accuracy of the iterative method with the convergence advantage of the closed-form method.The simulation results show that the proposed CRLB is 15.85% lower than the traditional CRLB,and the localization accuracy of the closed-form and hybrid-localization algorithms is 20.65% and 24.22% higher than the traditional closed-form solutions,respectively.It further demonstrates that the proposed algorithm has better localization performance and can reach CRLB asymptotically than the traditional range-based localization method.(2)Based on range-free localization technology,the two-dimensional Gaussian random distribution of reference nodes and the connectivity between nodes are used as priori information for the first time,and CRLB based on this priori condition is derived.Moreover,the theoretical variance of Centroid Location(CL)algorithm under any node distribution is derived for the first time.Compared with the existing theoretical variance which only deduces CL under uniform node distribution,the proposed theoretical variance can be used to evaluate the performance of CL under any node distribution.In addition,an iterative-hybrid method based on MLE is proposed to improve the localization accuracy.This algorithm combines the iterative method with the CL algorithm,which effectively utilizes the prior information and connectivity of the spatial distribution of nodes.The simulation results show that the localization accuracy of the iterative-hybrid method based on range-free is 2.56% higher than that of CL algorithm and can reach CRLB asymptotically.