Robust Communication in unknown environments
The objective of the proposed project is to develop engineering tools that are useful to the navy as it operates in an uncertain, partially, or completely unknown ocean environment. In particular, we are concerned with the design and implementation of algorithms for a passive sonar system employed for the detection and classification of an underwater sound source, such as a submarine. A passive sonar detects the identity of the unknown signal source based only the analysis of the distorted version of the signal corrupted by the unknown communication channel, which in this case is the ocean environment. The distortions in the received signal are mainly introduced by signal reflections and delays introduced by the channel. The primary challenge of a sonar system is to remove the distortions without having a priori knowledge of the sound signal and the communication channel. The signal received at the sonar array can be modeled as the convolution of the original sound signal with the channel. Blind deconvolution is a signal processing technique to clean the signal of the unknown channel effects. This project will focus on determining the utility of a newly proposed scheme for the blind deconvolution of array-recorded sounds from a remote source. The scope of the project is not only limited to the sonar system as the blind deconvolution arises in many other related applications such as multipath channel communications in rapidly varying unknown urban environment, encrypted, and robust military communications in unknown terrain, deblurring of images taken through a magnetic-resonance imaging machine, etc. The actual hardware unit developed in this project will then be tested in an underwater environment, where we aim to use an array of passive sensors to record sounds from an unknown sound source. Since the recorded signal is the convolution of the unknown sound source with the unknown underwater channel, the task of our designed unit will be to efficiently deconvolve the sound signal from the water channel, and hence remove the signal distortions. The implementation will involve sampling and digitally processing the samples of the received signal using a fast and scalable implementation of a blind deconvolution algorithm that will decouple the sound signal from the unknown water channel. This system will be an important component of a fully passive sonar system and will be a step forward in building a local capability of developing a sonar system for underwater submarine detection. The long term goal of the project will be to develop local technical resource and capability towards building a sonar system that is useful for the purpose of detection, classification, localization, tracking, and identification of remote unknown sound sources, ocean surveillance, multipath-channel communications and myriad of other applications.