Two different approaches for wide beam synthesis: Second order cone programming and swarm intelligence

Abstract

Different beam patterns can be generated by imposing different weights and phases to the elements of a phased array. Wide beam patterns are essential for many applications from various fields such as military and telecommunications. In this work, we propose two different approaches for generating wide beam patterns having high power levels. In the first approach, the problem is modelled as a second order cone program (SOCP). In this model, the antenna weights are the optimization parameters. The global optimal solution to this model can be achieved by any SOCP solver. In the other approach, all the antenna elements are assumed to work at the same power level. By choosing only the element phases as the optimization parameters, a non-convex optimization problem is constructed. Particle swarm optimization (PSO) is employed for finding the best solution. Experiments show that, both methods have similar performances for generating wide beam patterns with high effective radiated powers (ERP). It is observed that, SOCP based method suppressed the beam pattern in the undesired azimuth interval more effectively, since its optimization parameters are chosen from a larger search space. However, along with the phase shifters, it requires inserting transducers into the system hardware.