An unsupervised clustering-based scale-lag receiver design approach for the doubly selective underwater acoustic channel

Abstract

This paper details the architecture of a novel scale-lag rake receiver for the doubly selective underwater acoustic (UWA) channel. The shallow UWA channel is known to have a large Doppler spread. The cause of this is primarily due to the large spread of the angle of arrival of the received signal's paths. The Doppler scale value is dependent on the angle of arrival of the received signal. The Doppler spread is significant, and multiple Doppler scales are present in the UWA channel. Therefore, the receivers operating with only a single Doppler scale compensation perform sub-optimally. The optimal receiver should incorporate all possible Doppler scales for better performance. The receiver needs to have multiple sampling rates to incorporate all Doppler scales. It is usually done by having multiple receive channels, one for each sampling rate. It requires tremendous resources as its complexity increases exuberantly. This paper shows that these Doppler scales are clustered and that an optimal receiver is implemented using fewer receiver channels, each corresponding to one such cluster. It reduces the receiver's complexity. The simulation results demonstrate that using the Doppler scale value associated with the path of the highest magnitude per cluster yields the best performance.