Pulse Extraction for Radar Emitter Location

Two related data compression methods for radar signals are described and analyzed. The methods use the singular value decomposition (SVD) of a data matrix containing one pulse in each row to exploit pulseto-pulse redundancy. By using a rank-one approximation to the data matrix it is possible to achieve compression ratios typically of the order of several 10’s and sometimes over 150:1 for typical radar types while maintaining accurate location; the amount of compression depends on the radar’s parameters. Alternatively, by retaining a single singular vector as an extracted prototype pulse, it is possible to use it for matched filter processing, thus providing an improved method for doing noncoherent time-of-arrival (TOA) processing for a slight cost of a small amount of additional data to be sent.