Study On Signal Intergration For Distributed Coherent RF Detection System

Long time signal integration is one of the key technologies to solve the problem of distributed coherent RF detection system. Detect before track(DBT) method demands more target prior information, the key is how to carry out the migration compensation according to the target prior information, which can improve the SNR(signal to noise ratio). Track before detect(TBD) method searches target trajectory among multiple frames data, then accumulates energy along the target trajectory to improve target SNR. Compared with DBT, the key of TBD is how to obtain the target priori information. There are lots of algorithms to achieve DBT, and the essence of migration compensation methods demands to be further summed up. Less attention has been given to TBD at signal level, it is difficult to track and detect extremely low SNR target, the data models demand further improvement.In view of the above problems, this paper studies the DBT migration compensation methods, and track before detect based on particle filter(PF-TBD) algorithms based on coherent integration. The main work of this paper is as follows:1. From the view of search method, and signal space dimension of radar signal detection and track algorithm, the paper analyzes the development trend from DBT to TBD, followed analyzes the characteristics of the development.2. The paper analyzes the mechanism of range migration and Doppler migration, analyzes the essence of migration compensation methods in range migration and Doppler migration. The nature of long time signal integration methods in existing DBT has been carried on summary, and based on it, a method based on delay compensation and another method based on one-dimensional matching filter redesigned have been proposed.3. The paper establishes the nonlinear Bayesian framework of the track problem, introduces the solving method for particle filter track problem and signal detection method based on particle filter. On this basis, the paper establishes the PF-TBD model.4. For the difficult to modeling in traditional TBD methods for extremely low SNR target, the paper proposes one PF-TBD method based on coherent integration, which to modeling for the target at signal level. In the part of track, sampling target radial distance, radial velocity, radial acceleration and amplitude and phase information, the target state estimation is obtained. In the part of detect, coherent detector has been designed according to the target state estimation.5. For the difficult problem of phase sampling in the PF-TBD algorithm based on coherent integration, the paper proposes one TBD method based on matching filter redesigned, which phase sampling is ignored. In the part of track, sampling target radial distance, radial velocity, radial acceleration and intensity, the target state estimation is obtained. In the part of detect, one coherent detector based on matching filter redesigned has been designed according to the target state estimation.