Adaptive Weighted Multitaper Method for Spectral Estimation Under Data Loss

Abstract

Modern radar systems frequently encounter data loss during signal acquisition due to hardware failures, interference, or intermittent sampling, severely degrading spectral estimation accuracy. This work introduces an adaptive weighted multitaper method integrated with a circular memory architecture to address spectral leakage under adverse conditions with substantial missing data. Our approach employs a novel weighting scheme were systematically optimized through grid search across 200 parameter combinations using 500 Monte Carlo trials. Testing across SNR levels 0-30 dB with 30% missing data reveals substantial improvements Comparative analysis against the Bartlett, Welch, and Thomson Multitaper methods. For typical radar parameters (N=1024, K=8, M=128), our method requires approximately 8,192 operations versus 40,960 for Thomson MTM, representing a  kind of 80% computational reduction while maintaining superior performance. The method proves effective for radar applications requiring robust spectral estimation despite incomplete data acquisition, maintaining computational efficiency suitable for real-time processing