Research On Mixed Communication Signals Blind Reconnaissance Technology Based On Blind Source Separation

In the modern electronic information war,with the rapid development of wireless communication technology,a variety of interference constantly being introduced,resulting in the increasing of complex electromagnetic environment,and on the actual communication environment,reconnaissance party do not know any prior knowledge just like the modulation type of mixed signals intercepted,transmission characteristics of the channel or information sources of interference.Meanwhile,as the need of communicate reconnaissance mission;we first need to isolate a non-partner's signal from the intercepted mixed signal,and then detect and identify the communication signal.Existing separation algorithms have significant limitations,so that cannot meet the needs of communication reconnaissance missions.The emergence of blind source separation technology effectively solve the technical problem of "blind" communication signal reconnaissance.This paper mainly research on improvement of blind source separation algorithm and its application in the field of communication blind reconnaissance.Firstly,the basic principle of blind source separation and modulation recognition technology is researched in this paper.For 2ASK,4ASK,16 QAM,2FSK,4FSK,2PSK,4PSK,DSB,AM,FM,LSB and USB of 12 modulation signals,the Blind recognition of communication signals based on the JADE blind source separation and decision tree recognition is proposed.Firstly,each signal is isolated from a mass of each mixed signal by using JADE blind source separation algorithm,and then the Hilbert transform is used to extract the signal instantaneous characteristic information,and nine characteristic parameter set is structured,finally the classification method based on decision tree is used to identify the modulation types of communication signals.Simulation experiment proves that the use of the algorithm perfectly satisfy the communication signal blind reconnaissance missions.Secondly in view of the communication signals are sparse insufficiently case,a novel algorithm of the mixing matrix estimation in the underdetermined blind signal separation is proposed.The algorithm is based on the linear clustering characteristic of angle in single source intervals,the single source intervals are searched by constructing a angle searching function,at the same time,the number of source signals can be estimated.Then,the points on the different clustering interval are selected as the initial clustering center,and the mixing matrix is estimated precisely with fuzzy C-means clustering algorithm.Simulation results show that the method can accurately estimate the number of source signals,the higher estimation precision and the better noise immunity compared with conventional clustering algorithms.Finally,the application of blind source separation algorithm in DS-CDMA communication blind reconnaissance system is researched,the blind estimation algorithm of users' information and spread code based on FastICA are proposed.Estimation performance of three detected algorithms is compared by MATLAB software,the simulation shows that the estimation performance of FastICA was better than the MMSE and DEC algorithms,and the fewer the number of users,the smaller the error estimation rate,the higher the estimation performance.Because of FastICA blind source separation algorithm can accurately estimate the original user information without knowing any prior information about DS-CDMA users' information and spread code,it is very suitable for communication blind reconnaissance system.