Block NLMS cancellation algorithm and its real-time implementation for passive radar

The major interference in passive radars is the direct path and multipath clutter, which should be cancelled before matched filter processing generally. The block normalized least-mean-squares (block NLMS or BNLMS) algorithm, in which the weight vector is updated only once every occurrence based on a block gradient estimation, is exploited and implemented by using graphics processing unit (GPU) for real-time adaptive clutter cancellation. Validated by experimental data and theoretical analysis, the BNLMS algorithm can provide better performance in both cancelling capability and computation efficiency, which especially, is great for the wideband signal-based passive coherent location (PCL) radars owing to their strong noise and computational complexity.