Fast block LMS algorithm for interference cancellation in DVB-T based passive bistatic radar

Abstract

In this paper, we propose to use the FBLMS (Fast Block Least Mean Square) algorithm for Direct Path and Zero-Doppler Multipath Interferences (DPI and ZDMI) cancellation in DVB-T (Digital Video Broadcasting Terrestrial) Passive Bistatic Radars (PBRs). Diverse adaptive techniques are considered in passive radars' literature but most of them are computationally intensive when the DVB-T signal is used as an illuminator of opportunity, i.e., very high bandwidth. Using Fast Fourier Transform (FFT) for coefficient updating, the FBLMS algorithm shows a faster convergence rate, a shorter processing time and better CAF (Cross Ambiguity Function) results

Joint multipath signals and noise reduction in passive radar

“…They first estimate the amplitudes of the multipath signals by minimising a mean-square error criterion which is essentially derived under a Gaussian additive noise assumption and then subtract the multipath signals from the surveillance channel. These methods can be grouped into two categories according to the way the amplitudes are estimated, that is, the least square (LS) [17][18][19][20][21][22] and adaptive filtering methods [23][24][25][26][27]. The extended cancellation algorithm (ECA) proposed in [18] represents the most typical LS cancellation method.…”

“…However, the standard LMS method is sensitive to the learning rate and it suffers from low convergence rate and high convergence error, which result in coarse estimation of the multipath signal amplitudes [23,24]. There are rare passive radars using the standard LMS to suppress the multipath signals, many instead exploiting the improved variants, such as the normalised LMS (normalised least mean square) [25,26] and fast block LMS (fast block least mean square) [27]. The RLS method incorporates the concept of forgetting coefficient [24]; thus, it gets better convergence performance, leading to better estimation accuracy of the multipath signal amplitudes.…”

Weight Matrix-Based Least Mean Square Algorithm for Target Detection in Passive Radars

“…Bathes version of ECA is experimentally inspected for clutter attenuation of slow moving targets in [5].In contrast to pure block algorithms, the recursive least square (LSR) and least mean square (LMS) algorithms are iterative methods applicable of cancelling non-stationary clutters [2].The FBLMS scheme is a fast Fourier block version of LMS which dominates LMS in terms of cost. It is shown in [6]that FBLMS provides faster convergence, shorter processing time and more qualified cross ambiguity function (CAF) for DVB-T passive radars. After cleaning the received signal from disturbances, the costly CCF analysis is performed in an extensive range -Doppler field to detect targets.…”