A Hardware - Defined Approach to Software - Defined Radios : Improving Performance Without Trading In Flexibility

The thesis presents an implementation of a general DSP framework on the Texas Instruments OMAP-L138 processor. Today's software-defined radios suffer from fundamental drawbacks that inhibit their use in practical settings. These drawbacks include their large sizes, their dependence on a PC for digital signal processing operations, and their inability to process signals in real-time. Furthermore, FPGA-based implementations that achieve higher performances lack the flexibility that software implementations provide. The present implementation endeavors to overcome these issues by utilizing a processor that is low-power, small in size, and that provides a library of assembly-level optimized functions in order to achieve much faster performance with a software implementation. The evaluations show substantial improvements in performance when the DSP framework is implemented with the OMAP-L138 processor compared to that achieved with other software implemented radios. Thesis Supervisor: Prof. Hari Balakrishnan Title: Professor of Electrical Engineering and Computer Science