Doppler Tolerant Range Sidelobe Suppression For Meteorological Radar With Pulse Compression

Dispersive pulse transmission and pulse compression upon reception is frequently used in radar to achieve high energy per pulse with low peak power while maintaining large bandwidth for fine range resolution. Pulse compression results in range sidelobes that must be suppressed for many applications. While suppression methods exist, they are very sensitive to unknown doppler shift, a situation frequently encountered in weather radar applications. We present a method for alleviating the doppler sensitivity. The method consists of processing a sequence of pulse echoes in three steps: (1) Pulse to pulse doppler filter in a doppler filter bank (usually an FFT); (2) heterodyning each doppler output from the bank with a signal that removes the doppler phase variation across the uncompressed pulse, i.e., along the range dimension; (3) pulse compression and range sidelobe suppression based upon zero doppier design. The particular criterion of sideloble suppression in this paper is minimization of integrated sidelobe level. INTRODUCTION Radars that are peak power limited to low or moderate levels may use dispersive pulse transmission and pulse compression upon reception to achieve high energy per pulse with low peak power. The dispersive transmission enables high bandwidth to be achieved for fine range resolution. The fine range resolution is obtained in the receiver by pulse compression or matched filtering. Matched filtering results in range sidelobes that can be troublesome in an extended clutter environment because of clutter “flooding” into the range increment being examined for the presence of scatterers and for the measurement of the charateristics of these scatterers. While there exist techniques for suppressing range sidelobes, these techniques are very sensitive to doppler frequency shifts in the echo. This is particularly significant in radar meteorology where the reflectors of interest are extended collections of small scatterers, such as precipitation. For mapping reflectivity, clutter “flooding” from range sidelobes can obscure the proper measurement of scatterer properties in the desired range increment. Sidelobe suppression in pulse compression radars is, in many respects, an established art, but there are some limitations to known techniques having to do with doppler shifted echoes. In previous approaches to range sidelobe suppression the varying phase shift, due to a doppler frequency shift over the duration of the uncompressed pulse, has been neglected. It has been thought to be too small to have a significant effect. We have found, however, that it is quite significant in that even moderate doppler shifts destroy the sidelobe suppression. Analysis and computation show that the quantity controlling the doppler sensitivity is the product of the uncompressed pulse duration and the doppler frequency shift. This product is measured by the doppler phase shift over the uncompressed pulse duration. Calling this the “doppler phase variation” we have doppler phase variation = 2nfdT, radians where fd is the doppler frequency shift in hertz and To is the uncompressed pulse duration in seconds. If the doppler frequency shift could be confined to a narrow band, then time sidelobe suppression could be very effective. Therefore, the technique or techniques to be described here separate sequences of echoes into several doppler “bins”, and time (i.e., range) sidelobe suppression is applied to each of the doppler bins. OUTLINE OF THE TECHNIQUE The technique is outlined in Figure 1. The received signal is downmixed to yield the complex envelope (I+jQ) which is then sampled by an N D converter. The next step is doppler filtering by K doppler channels using a pulse to pulse DFT to separate the groups of returns about the center doppler frequency of each filter. The doppler filter outputs still contain the doppler phase shifts along the range. Each doppler filter output is multiplied by a complex exponential giving the phase shift corresponding to the center frequency of the doppler filter to remove the doppler phase shift across the pulse. This doppler phase shift removal in the range dimension is crucial because it turns out that the sidelobe suppression doppler sensitivity results from the doppler phase shift over the duration of the uncompressed pulse. The resulting waveform is passed to the pulse compression and range sidelobe suppression filter. The compression and suppression filter design corresponds to the type of code transmitted with zero doppler on the return. The K filters operate in parallel to cover frequencies evenly distributed in the desired doppler velocity band [-Vmax,Vmax]. Because of the removal of the doppler phase shift after each doppler filter, the compression and suppression filters have a common design, being designed for zero doppler. We consider this to be an important aspect because it simplifies manufacture. The design of the combined pulse compression and range sidelobe suppression can take several forms; one form that we chose is a cascade of pulse compression followed by sidelobe suppression. The criteria upon which the suppressors may be designed include: 1. Minimizing the peak sidelobe. 2. Minimizing the integrated square of the sidelobes. DOPPLER FILTER RECEIVED COMPLEX ENVELOPE I +jQ ANALOG TO DIGITAL CONVERSION ZERO DOPPLER PULSE BANK COMPRESSION AND SIDELOBE 1 SUPPRESSION FILTERS BXP I-iZnflr?gl .--ITO ENVELOPE DETECTORS AND FURTHER PROCESSING F exp k i 2 n f ~ 1 r q ) Figure 1. Block diagram of doppler tolerant range sidelobe suppression. f,, fi, . . ., fKp1 are the doppler filter center frequencies. r is the integer time index.z, = range sampling period. 206 91-72810/92$03.00 0 IEEE 1992 3. Suppressing completely the sidelobes in a specified interval In this paper, we report only on the results using the second criterion. The integrated square of the sidelobes is called the “integrated sidelobe level” (ISL). The criterion on minimizing ISL is particularly appropriate when the interfering echoes or clutter is range extensive. In such a case, the interfering clutter power, for uniform clutter extent, is proportional to ISL. The suppression filter is designed in the form of a transversal filter (i.e., a finite impulse response filter). The design equations yield the coefficients or weights that are applied to the taps of the transversal filter. An outline of the derivation of the suppression filter coefficients is given in the Appendix. around the principal. RESULTS Figure 2 shows what conventional range sidelobe suppression can achieve when the echo has no doppler shift. Figure 3 shows the sensitively of sidelobe suppression to doppler sliift calulated for an 11 cm radar and a 33 microsecond transmitted pulse. This figure shows that while very great sidelobe suppression is achievable when the doppler frequency is known, the suppression is extremely sensitive to (unknown) doppler shift. Figure 4 compares the suppresssion capabilities of conventional and our doppler tolerant method of range sidelobe suppression. The superiority of the latter is evident. Figure 5 shows the loss in signal to noise ratio to be expected as a function of the number of suppression filter coefficients. For very deep sidelobe suppression, a loss of approximately 3 dB is to be expected.