Data Compression for Radar Signals: an Svd Based Approach

Multiple platform coherent location systems operate by computing the time difference of arrival (TDOA) and frequency difference of arrival (FDOA) among signals received at geographically separated platforms. The bandwidth of the data link is the major bottleneck in the processing. Previously developed data compression methods [1, 2] can not satisfy the compression ratio and the accuracy requirements because they were designed for the generic signal case and do not fully exploit the characteristics of the radar signal. A new compression scheme presented in this thesis is built from the ground up with the characteristics of the radar signal in mind. It is based on the idea that a radar pulse train can be modelled as one prototype pulse and a parameter vector for each pulse to transform the prototype pulse to each specific pulse. A compression ratio of 10 ∼ 20 : 1 has been achieved with minor, if any, FDOA/TDOA accuracies degradation in most cases. The two major techniques involved here are the fractional delay filter and the singular value decomposition (SVD). This thesis starts with some preliminary technical background used later in this thesis. Then two chapters are dedicated to the fractional delay filter and the SVD method, respectively. In addition to presenting and verifying our compression scheme, a newly developed LMS adaptive FIR fractional delay filter and an alternative to the cross-ambiguity processing based on the parameterization method are developed. Extensive simulation results are presented throughout the thesis. In the last chapter, conclusions and suggestions for future work are given.