Performance analysis of different DSP algorithms on advanced microcontroller and FPGA

Abstract

Digital signal processing (DSP) algorithms are mostly implemented on special purpose digital signal processors. Recently, advanced hybrid microcontrollers and field programmable gate arrays (FPGAs) are also evolving as the suitable platforms for realizing DSP algorithms. This paper aims towards the analysis of performance of popular DSP algorithms on advanced microcontroller and FPGAs. A time complexity analysis for different filter structures like finite impulse response (FIR), infinite impulse response (IIR) and adaptive filter using least mean square (LMS) algorithm etc. has been studied for AVR ATMega64L microcontroller with respect to MC56F8013 digital signal controller. The same algorithms have also been realized on FPGAs and performances have also been analyzed. A comparison has been done on the hardware utilization of different FPGAs (Xilinx Spartan3e, Virtex2, and Virtex4) for different algorithms. Based on the performance analysis, finally, the architecture of a reconfigurable processor has been proposed with a view to exploit the advantages of all these platforms.