Sonar Data Processing Using Multicore Architecture Processor and Embedded Linux

TLDR: This project is aimed at bringing up a bare multicore power architecture processor board with Embedded Linux, benchmarking it for performance and implementing a digital signal processing application exploiting multiple cores and SIMD units in each core.

Abstract: Background: Traditionally, sonar signal processing is performed on dedicated signal processors to meet the real-time processing requirements Recently, the computational capability in embedded general purpose processors has also multiplied with the introduction of multiple cores and vector processing units Embedded Linux is the usage of Linux kernel and various open source components in embedded systems The key advantage of Linux and open source in the embedded system is the ability to reuse components which allows to design and develop complicated products based on existing components and to have full control of the software part of the system This project is aimed at bringing up a bare multicore power architecture processor board with Embedded Linux, benchmarking it for performance and implementing a digital signal processing application exploiting multiple cores and SIMD units in each core Methods: The T4240RDB processor is developed using Yocto Project Yocto is an open embedded project used to build custom-based Linux images Next, the processor is benchmarked for its processing capabilities and then, signal processing application is done on the processor Findings: Yocto Project can be used to develop architecture-specific images as well as images can be modified and additional recipes and packages can be added The T4240RDB can easily do signal processing application Application: The T4240RDB can be integrated with the sonar subsystem and can be used for real-time data processing