Sirnulation studies of moving-target detection: a new approach with Wigner-ViIle distribution zy ' L-po,nt 3-pulse _L canceller FFT P.K.Kumar K.M .\d. Pra zyxwvutsrqpo b hu weighting ' - and - magnitude zyxw Indexing terms: Moving-turget detection, Wignei- Ville distribution, Fast Fourier transforms - zy Abstract: The authors discuss the application of Wigner-Ville distribution (WVD) for moving- target detection (MTD) in a typical ground-based radar. The proposed technique enhances the target-detection capability using the fact that the sidelobe levels in the Wigner spectrum are very small in magnitude. This in turn facilitates the realisation of the Doppler-filter bank by WVD without resorting to additional weighting to reduce the sidelobes, unlike in the current MTD-I sys'tem (FFT-based). The MTD scheme proposed exhibits good performance capabilities, by providing higher detection probabilities, lower false-alarm rates and an additional 10 dB gain in improvement factor, zyxwvuts for ground plus weather clutter, over MTD-I. This novel approach to MTD serves as a simple and superior alternative to the MTD-I1 and MTD-I11 schemes, which have similar performance capabilities over MTD- I. Computer-simulation results are presented to support the superiority of the proposed approach. 1 Introduction Radar detection of targets with unknown frequency is of major concern to radar engineers. Traditionally, it has been accomplished by means of Doppler-filter bank; each filter being keyed into one out of L doppler shifts, spanning the whole range of unambiguous fre- quencies (0, zyxwvutsrqp f,), with &, being the pulse-repetition fre- quency (PRF): a scheme of such a processor, called a moving-target detector (MTD), is represented in Fig. 1. The MTD type of processing takes place in high-per- formance coherent-radar systems such as airport sur- veillance radar (ASR). It is designed to improve the target detection in various forms of clutter while pro- viding low-output false-alarm rate. Three MTD schemies are at present in use, namely MTD-I [l], MTD-I1 [2] and MTD-I11 [3]. MTD-I1 and MTD-I11 differ mainly from MTD-I in the Doppler-filter-bank realisation with 7-point FIR and %point FIR filters, respectively [2, 31. Although MTD-I1 and MTD-111 have better per- formance capabilities than MTD-I, from the viewpoint zyxwvu 0 IEE, 1997 IEE Proceedings online no. 1997I377 Paper first received 1st May 1996 and in revised form 8th May 1997 The authors are with the Department of Electrical Engineering, Indian Institute of Technolgy, Madras 600 036, India of probability of detection and improvement factor, MTD-I is still considered to be a useful platform for conducting various research studies [4]. This has been due to its widespread usage and simplicity. For these reasons, the paper details the simulation studies of WVD-based MTD, the MTD-WVD, and compares its performance with MTD-I (FFT-based). The results obtained show that MTD-WVD have similar perform- ance capabilities to MTD-I1 and MTD-111 (for exam- ple, they have l0dB more improvement factor than MTD-I), serves as a simpler substitute, especially from the viewpoint of complexity in computing the FIR-fil- ter coefficients which, in MTD-I1 and MTD-111, needs careful treatment to achieve the desired characteristics. Moreover, any change in coherent train lengths needs the filter coefficients to be redesigned. The novel tech- nique proposed using WVD has no such complexities, but for a small increase in computational burden. This can be tackled with the available fast algorithms to compute the same [5-71. r--1 r--1 I WVD I Ti Drocessor I I magnitude ' -I I I - thresholding report zero- filter Fig. 1 Block Schematic of MTD-I and MTD- WVD The application of WVD in radar-signal analysis has been gaining importance [8-10] because of its powerful capability to analyse clutter like signals in the time and frequency domains. In this paper, WVD is considered in the frequency domain, as it presents very low side lobes, say of the order of -32dB, facilitating the reali- sation of the Doppler-filter bank using WVD without resorting to additional weighting to reduce the sidelobes, as shown in Fig. 1. Fig. 1 shows a block schematic of MTD-I with an FFT-based Doppler-filter bank, while the dotted path shows the filter bank using WVD. The terms MTD-I and MTD are used mter- changingly in this paper, as in previous papers [l, 41. The work described in this paper was entirely simula- tion-based, a method which is of great significance and 259 IEE Proc -Ruduv, Sonar Nuvig , Vol zyxwvutsrqponm 144, No zyxwvutsrqpo 5, 0~ tober I997