Support-Constrained Mixed-Norm Optimization Techniques for Estimating Multipath Activity in Shallow Water Acoustic Channels

Abstract

In this article, we propose channel estimation techniques for shallow water acoustic channels that exhibit rapidly fluctuating high-amplitude multipath activity across different regions of the channel support. Specifically, we impose support constraints on the channel impulse response that confine tracking the shallow water acoustic channel to within the most active regions within the channel delay spread. The key idea is to focus on channel estimation performance through the banded nature of the multipath arrivals in the delay versus time channel that we denote as subregions of the channel delay spread. Because we are examining sparse shallow water acoustic channels, we adopt the well-known Lasso metric and related mixed-norm optimization techniques, but in contrast to traditional sparse sensing approaches, we use masked constraints to prioritize the estimation of regions of interest within the channel delay spread. This provides a tradeoff between computational time and estimation performance that we explore in detail over experimental field data from the SPACE08 experiment and over simulated channels emulating diverse oceanic conditions.