FPGA Accelerated Phased Array Design Using the Ant Colony Optimization

Abstract

The objective of this paper is to investigate the utilization of field programmable gate arrays (FPGA) in the field of electromagnetics by applying the ant colony optimization (ACO) method in the design of phased array antennas for multiple beam satellite communication systems. The amplitudes of the array elements are optimized to reduce the cochannel interference in a multiple beam satellite communication system. The potential gains in the speed of the calculations are investigated in comparison to conventional simulation techniques of the same application on a regular PC. Two different FPGA platforms and implementation approaches are compared for performance to two software developments implemented using Matlab and C languages. It has been shown that significantly accelerated performance can be achieved for the particular application. This kind of speed improvement can enable handling more complex requirements and constraints for the same application in a very reasonable amount of time, which would otherwise be impossible with conventional computational platforms and techniques. This magnitude of speed improvement is due to the configurable nature of the FPGAs. Unlike central processing units (CPU) in a conventional computer, which have to deal with a preset set of instructions to properly function; FPGAs are completely programmable to carry out a set of functions in the most efficient manner for the particular algorithm at hand. In this study, the FPGA has been configured to function as an efficient “ACO machine.” Both parallelization and pipelining have been utilized to achieve this performance. The details of the implementation on the FPGA platform and the achieved acceleration are discussed in the paper.