The Methods Of Passive Wideband Signal Intrapulse Analysis Based On GPU

Under the conditions of information warfare, the real-time detection and analysis of radar signals is crucial. Passive radar is widely used in the passive detection because of specific advantages in the field. passive radar does not generate RF signals, enemy radar reconnaissance system can not detect the radar Radar reconnaissance system can not detect the radar.For passive detection systems, in order reconnaissance the signal of the scope of the investigation and intercept the total probability,we require it high sensitivity and wide receiver, but with a conflict between broadband and high-sensitivity receiver,we often conduct digital channel processing after A/D. In order to make digital channelized quickly and and analyze radar intrapulse signal accurately, proposes a parallel processing solution based on GPU, improved the bottleneck of high-speed A/D and low-speed signal processing greatly.All contents of this paper achieve in the software environment, including:1. The processing method of digital channelization based on GPU were studied and discussed. The digital channelization technology contains the polyphase filter and FFT algorithm, we can use parallel computing methods. Exerting high parallel computing capability of GPU and testing the performance of the program.2. Optimizing digital channelization method based on GPU. According to the digital channelization technology of achieving the whole process. Since the CUFFT library has the ability to FFT transform multi-batch data, so increases the matrix transpose program after polyphase filter. all data can be performed simultaneously by FFT transform based on GPU, it can Increase the speed of operation.3. LFM signal parameter estimation methods were studied based on GPU. Due to the chirp signal into sine signal after the time autocorrelation, At this point simply apply the appropriate algorithm to estimate the parameters for sinusoidal signals. The main steps of the algorithm is peak search and peak search is the problem of solving a maximum. We can use the GPU to perform high-performance parallel computing and achieve good acceleration effect. 4. Optimizing the LFM signal parameters estimation method based on GPU. According to memory Optimization of GPU, we use features of twiddle factors of FFT algorithm and previously stored into the GPU texture memory. The peak search can take advantage of cache of shared memory in GPU, This can increase the speed of operation and achieve real-time results.