Publications [#330767] of David J. Brady
- Kocic, M; Brady, D; Merriam, S, Reduced-complexity RLS estimation for shallow-water channels,
Ieee Sympsium on Autonomous Underwater Vehicle Technology
(last updated on 2019/11/22)
An adjustable complexity, recursive least squares (RLS) estimation algorithm is presented, which is suitable for adaptive equalization and source localization in shallow-water acoustic channels. The algorithm adjusts its computational complexity, measured in FLOPS per update, in a decreasing fashion with the relative signal strength, by ignoring 'insignificant' dimensions of the channel. The algorithm reverts to the well-known fast RLS algorithms when the signal quality is weak, and may be combined with reduced period updating techniques. Examples illustrate computational savings in excess of one order of magnitude, permitting a tripling of the maximum data rate through these complexity - limited communication channels.