A comparison between GA, PSO, and IWO for shaped beam reflector antennas

This paper presents a comparison between three evolutionary algorithms (EAs) for pattern synthesis of offset reflector antenna fed by a planar array of horn antennas. To perform the optimization process, an elliptical-shaped beam in the U-V plane (U = sin theta cos phi and V = sin theta sin phi) is considered as the desired far-field radiation pattern. To attain the appropriate excitation value for array elements, three conditions are considered: (1) variable amplitude (with uniform phase distribution), (2) variable phase (with uniform amplitude distribution), and (3) variable amplitude and phase excitation. Obtaining the appropriate excitation value based on the mathematical methods is always complicated and time-consuming. Therefore, genetic algorithm (GA) and particle swarm optimization (PSO) as two well-known EAs have been used widely for different applications and shown the promise to solve complicated problems. This paper compares these two EAs with invasive weed optimization (IWO) which is robust and has simple and powerful process with few tuning parameters. We found that for pattern synthesis of multi-feed reflector antenna in different conditions, IWO can provide accurate and comparable results with GA and PSO methods at approximately same iteration number. The convergence diagrams as well as the optimized radiation patterns for different conditions are presented and compared for GA, PSO and IWO.