Exploiting a Supervised Index for High-accuracy Parameter Estimation in Low SNR

TL;DR

This paper introduces a novel supervised learning approach that uses a validity index to iteratively refine signal subspace estimates, significantly improving parameter estimation accuracy in low SNR conditions.

Contribution

It proposes a closed-loop, supervised method that leverages a validity index to guide the refinement of the signal subspace for better parameter estimation.

Findings

Enhanced estimation accuracy in low SNR scenarios

Introduction of a feedback-based supervised refinement loop

Potential for further improvements in high SNR conditions

Abstract

Performance of parameter estimation is one of the most important issues in array signal processing. The root mean square error, probability of success, resolution probabilities, and computational complexity are frequently used indexes for evaluating the performance of the parameter estimation methods. However, a common characteristic of these indexes is that they are unsupervised indexes, and are passively used for evaluating the estimation results. In other words, these indexes cannot participate in the design of estimation methods. It seems that exploiting a validity supervised index for the parameter estimation that can guide the design of the methods will be an interesting and meaningful work. In this study, we exploit an index to build a supervised learning model of the parameter estimation. With the developed model we refine the signal subspace so as to enhance the performance of the parameter estimation method. The main characteristic of the proposed model is a circularly applied feedback of the estimated parameter for refining the estimated subspace. It is a closed loop and supervised method not reported before. This research opens a specific way for improving the performance of the parameter estimation by a supervised index. However, the proposed method is still unsatisfying in some scopes of signal-to-noise ratio (SNR). We believe that exploiting a validity index for the parameter estimation in array signal processing is still a general and interesting problem.