V.I. Koshelev

The primary function of Doppler signal processor (DSP) is the coherent accumulation of the signals, reflected from the radar targets having a priory unknown velocities. Furthermore, Doppler filtration addresses the tasks of additional suppression of passive jamming, estimation of the target radial velocity and its higher derivatives as well as, in combination with the frequency-modulated signal, target distance estimation. A choice of DSP-s parameters determines such technical characteristics of radar as range and probability of correct detection, resolution with respect to velocity, distance and acceleration. Specific requirements to DSP in radar stations of varying purpose determine the suitability of particular synthesis criteria. For instance, in the synthesis of radar signal detectors it is desirable to maximize the average over all Doppler channels probability of correct detection. One may perform the optimization to compute the windowing functions for pre-weighting pulse train input to DSP. Broadening of the main lobe caused by weighting leads to excessive (up to 1...2 dB) overlap between neighboring channels, which contradicts the required frequency resolution. Therefore it is desirable to perform additional frequency decimation of DSP-s channels by means of a modified FFT algorithm. A directed graph of such an algorithm is obtained by truncating extra branches corresponding to even or odd channels from classical FFT algorithm. A further modification of FFT algorithm that reflects non-equidistant spacing of probing pulses («vobulation») used to eliminate the «blind» velocities is proposed. Conventional structure of multichannel filter employs the same windowing function for all frequency channels. For more flexible processing, it is recommended to use a different window for each channel (or group of channels) of multichannel filter. In the latter case, the filter does not come to conventional DFT algorithm. The algorithm we present can also optimize such structural characteristics of a DSP as channel count as well as the width and tuning of velocity (Doppler) channels. A sharp change in the radial component of target velocity during the observation time causes the change in Doppler frequency of the signal, which leads to its smearing over neighboring Doppler channels and reduces the possibility of its detection and the accuracy of measurement of its parameters. This may also lead to a false decision about the presence of several targets. This problem is important for the on-board pulse Doppler radars. One possible DSP algorithm in this case consists in forming several short DFT blocks whose length depends on detected acceleration. The solution of described set of optimization problems allows to approach closely DSP efficiency to the limits set by optimal processing algorithms synthesized on the basis of maximum likelihood relation in various regimes of radar station.