TV Passive Radar Signal Real-time Processing Based On CPU/GPU Synergy Structure

As electronic counter-countermeasures play a more and more important role,competition of Radar research among different countries is carried out fervently.Since the 1930 s,plenty of time and large human and material resource are put into study about traditional Radar in every country,of course,the results are plentiful and mature.On the other hand,every country put large effort into how to prevent and destroy active radar of the enemy.So traditional radar faces some threats,such as anti-radiation missile,stealth aircraft,electromagnetic interference,low-altitude penetrating and so on.The reason that these threats are dangerous for active radar is that active radar broadcasts electromagnetic wave and is easy to be detected.Then,the location and electromagnetic frequency of the radar can be quickly acquired.In addition,active radar works in high frequency band which has poor effect on stealth aircraft and low-altitude penetrating.In front of these threads,passive radar breaks new ground for electronic counter-countermeasures.This is because passive radar doesn't broadcast electromagnetic wave and its signal source is a third non-cooperative signal,including FM radio signal,television(TV)signal,cellular phone base station signal,satellite signal and so on.Among various signal sources,analog TV signal is selected in this paper for research and implement as analog TV signal has high energy and large coverage area.In comprehensive consideration of instantaneity,portability,adjustive capability and cost,Graphic Processing Unit(GPU)is selected as the implement platform.Therefore,GPU is introduced concisely firstly,including development history,hardware architecture,application function and optimization skills.Then,analog TV signal is analyzed in the paper.It is discovered that the ambiguity function of audio signal is like a pin.So audio signal is perfect to be utilized to passive radar.The processing of audio signal includes only clutter cancellation,distance-dopler processing and constant false alarm ratio(CFAR)detecting,and the trace points can be got.Whereas,the ambiguity function of video signal has distance ambiguity sidelobe,so it needs special processing i.e.Mismatched filter.But for reason that the computational task of mismatched filter is too heavy to meet the real-time requirement,the idea of subsection is introduced here.After showing the character of analog TV signal and selecting processing algorithm,the theories of relevant algorithms including ECA clutter cancellation,subsection mismatched filter,distance-dopler processing,fast CFAR detecting and sidelobe cancellation are presented.At last,a GPU implement scheme about analog TV passive radar signal processing is given in the paper.The scheme shows how to divide calculation tasks fine-grainedly and allocate them on GPU threads.To reflect the effect of GPU-accelerated calculation,the running times of large matrix multipliacation and inversion on GPU and Central Processing Unit(CPU)are compared.