Parallel Computing for Interferometric Sonar based Imaging

Sonar systems are used in seabed mapping and charting applications. The real time processing of sonar signals demand very high-speed computations due to the high platform velocity of the sonar systems. Since direct measurement of real phase delay is not a practicable solution in high performance imaging systems, a systematic phase unwrapping process is often employed to compute the phase delay from the wrapped phase values. Real-time computations are difficult to realize mainly due to the incorporation of phase unwrapping process at the front end. In addition, conventional phase unwrapping algorithms tend to rely on iterative processing, which are sequential in nature. This work presents a novel method for the determination of phase delay, which is based on Residue Number Systems (RNS) arithmetic. It has been shown that, by incorporating RNS, a high degree of parallelism could be introduced within the unwrapping process. This provides for the realization of a high-speed architecture to compute the required phase delay. This work also shows how the RNS based technique can be extended to facilitate three-dimensional imaging.

Autorentext

Mohamed Firdhous Senior Lecturer, University of Moratuwa Sri Lanka. Firdhous received his BSc Eng., degree from the University of Moratuwa, MSc from the NTU in Singapore and MBA from the University of Colombo. Firdhous is a Chartered Engineer. Firdhous has several years of experience in IT in Sri Lanka, Singapore and USA.