Study On QGA-based Feature Selection Of Target Recognition By UWB Communication Signal

Ultra Wide-band(UWB)wireless communication technology has the ability to transfer the data with ultra wide bandwidth in a very high speed over a short distance.Owing to its way of modulation and multiple access technology,UWB has higher data transmission speed,stronger resolution and penetration,better anti-interference ability and security compared to other wireless communication technology.Based on the above reasons,UWB technology has been widely used in the area of detecting and identifying the foliage-covered target in the jungle.How to extract and select the features of target obscured by foliage from UWB signals and recognize the types of the targets is still a hot issue in current research.This paper conducted a thorough research on the feature selection part of target recognition techniques based on UWB signal,and carried out the following research work based on the data obtained from the UWB target information collection system which set up by our laboratory:Firstly,three improved quantum genetic algorithm have been studied according to the improving directions of recent years.They are improved quantum genetic algorithms based on the adaptive phase rotation,crossover and mutation operator,immune clone respectively.Also,these improved algorithms are applied to solve backpack problems and extremum problems.Simulation results shows that the improved quantum genetic algorithm based on the adaptive phase rotation is remarkable in improving the convergence rate and is more suitable for target feature selection than the other two algorithms.Secondly,an adaptive phase rotation quantum genetic algorithm based target feature selection method is proposed in the paper in order to get the optimal target feature subset.Quantum computing has a high degree of parallelism,exponential storage capacity and index acceleration for classical heuristic algorithm,and is superior to conventional algorithm in computational complexity and convergence speed.Therefore,quantum computing is introduced for feature selection to improve the processing speed and optimal performance of algorithm without affecting the high classification accuracy.First,encoding the target feature,then generating the initial population.The recognition rate of feature subset is served as fitness function.The optimal feature subset among the current population is served as update guide for all the individuals.Next,quantum rotation gate is used to update the population dynamically.In the end,the optimal feature subset is obtained.In addition,the comparison between the proposed and the traditional genetic algorithm(GA)-based method will be discussed.According to the obtained results,the method presented in this paper has faster convergence speed and better global search ability.Finally,the research works of the whole dissertation is summarized,and several valuable research directions of feature selection for the target detection and recognition based on UWB signal are discussed.