Research And Application Of The Blind Source Signal Separation Method Based On ICA

In the communication reconnaissance,the effective separation of the received signal is one of the key technologies to improve the effect of electronic warfare,it is also the prerequisite for effective implementation of ECM interference.Therefore,the communication reconnaissance signal separation method has a very important role in modern electronic warfare systems.In the military field,the use of blind source separation of passive radar technology used in stealth aircraft,the passive radar just by receiving signals to detect the target,it does not emit any electromagnetic signal,it is difficult to be detected by enemy radar.In the communication field,blind source separation technique for CDMA blind or semi-blind multiuser detection,it can be effectively suppressed near-far effect and MAI.This thesis mainly studies the communication reconnaissance signal separation method,under the condition of unknown source number and complicated electromagnetic.First study the signal source number estimation algorithm,through comparative analysis of the simulation experiment under the condition of high SNR GIC and MDL is the estimation of source number.When the sampling point is greater than about 400 points and SNR greater than about 2 dB,the algorithm can accurately estimate the number of signal sources.Secondly,improve the whitening process for complex ICA,with signal source number estimation results to set the thresholds of received signal covariance matrix eigenvalue sequence,with a noise variance estimation result to the correction signal subspace main characteristic value,to suppress noise.then for the complex communication reconnaissance signal characteristics,focusing on the complex T-CMN algorithm research,and separation of communication signal reconnaissance scheme is given under the unknown source number and the complex electromagnetic environment,through comparative analysis of simulation experiments,whether nonlinear function isG₂ orG₃,when the signal to noise ratio of less than 12dB,the improved whitened T-CMN separation performance is better than the separation performance of the original T-CMN algorithm.In addition,both the low SNR and high SNR,nonlinear functionG₂ corresponding separation performance superior to non-linear function curveG₃corresponds to the curve.Finally,based on the T-CMN communication reconnaissance signal separation scheme this thesis presents a communications surveillance system specific application cases,the case use civil communication frequency"cover"to achieve the purpose of secret communication,and verify the feasibility of the application cases.Finally,communications reconnaissance signal separation scheme in the software radio platform for the project implementation.Considering the limitation of test conditions,designed the communication reconnaissance scheme signal separation algorithm on the SDR platform deployment and code implementation.Developed a PC interface,with the USRP produced FM test signal,communications reconnaissance signal separation scheme based on T-CMN were tested,effectively separate two-way co-frequency FM radio,.verify the feasibility of communication reconnaissance signal separation scheme based on T-CMN.