Pixel Antenna Optimization based on Perturbation Sensitivity Analysis

Pixel antennas provide a straightforward method for antenna design that can meet a given set of specifications. A key step in the design of pixel antennas is the optimization of the hardwire connections between the pixels in the antenna. Due to the very large number of possible pixel connection combinations this optimization needs to be performed by heuristic algorithms. In this work sensitivity analysis is proposed for the efficient optimization of pixel antennas. In the approach each individual pixel of the antenna is perturbed in turn to obtain the sensitivity vector of the objective function. The sensitivity analysis at each step is performed efficiently through the use of algebraic simplification that avoids matrix inversion and leads to a significant saving in computational load. The optimization process then selects the perturbation with the greatest sensitivity. Two examples are provided to demonstrate the performance of the proposed optimization method. For both examples, the proposed approach performs the optimization with significantly fewer iterations and either achieves a better or similar minimum in the objective function compared to Genetic algorithm approaches. Combined with the efficient computation of the sensitivity vector significant savings in computational load are achieved compared to existing approaches.