Underwater channel estimation exploiting multipath feature morphology

Abstract

Real-time accurate channel estimation has been an ongoing challenge because of diverse oceanic events that cause rapid fluctuations of high-energy multipath activity across the delay spread. This work leverages current compressed sensing and sparse optimization techniques with topological signal processing to improve estimation time and localize channel estimation to salient parts of the delay spread. This work improves the estimation time by tracking the channel as a union of overlapping multipath and other scattering events, which are modeled as “feature braids” in the delay-time domain. A channel feature braid may be intuitively visualized as the topologically connected trajectory of a group of channel delay taps, which represent the support of dominant or persistent scattering events, e.g., surface bounce multipath scattering. We present algorithms that harness support-constrained mixed norm optimization techniques to track the evolving support of channel feature braids. We validate our channel feature tracking algorithm independently in experimental field data as well as BELLHOP channel simulations across a diversity of oceanic conditions. This work shows that braiding used in estimation can improve estimation time and track high-energy events that develop within the delay vs time channel representation.