Optimization of Microphone Array Geometry with Evolutionary Algorithm

Microphone configuration for beamformer design using the Taguchi method

“…To compare the results obtained by the Taguchi method which small amount of experimental time is only available, two commonly-used heuristic methods including genetic algorithm (GA) [32] and particle swarm optimization algorithm (PSO) [35] have also been used as they have been used on designing microphone configurations [36][37][38][39][40][41]. They are commonly used on solving multiobjective problems.…”

“…The numerical results obtained by the Taguchi method are compared with those obtained by the two commonly used heuristic methods including genetic algorithm (GA) [32][33][34] and particle swarm optimization algorithm (PSO) [35] for determining microphone configurations. These two heuristic methods are also commonly used on solving multi-objective problems and also they have been used on designing microphone configurations [36][37][38][39][40][41]. The comparison indicates that the Taguchi method is capable to develop the microphone configurations with similar beamforming performance compared with the two heuristic methods while shorter computational time is required on the Taguchi method.…”

Planar Array Geometry Optimization for Region Sound Acquisition

“…) Since t(X) is highly non-convex with respect to X, an explicit solution of ( 18) is difficult if not impossible to obtain. But various optimization methods can be used to search for a suboptimal or an optimal solution, including the Monte-Carlo simulations method [34], the simulated annealing methods [35][36][37], the particle swarm optimization [29,38], and the genetic algorithms [20][21][22][23][24][26][27][28]. In this work, we choose to use the genetic algorithm, which exploits the historical information of evolution procedure to guide the searching path and has been shown effective in solving non-convex problems in many fields.…”

“…Early such efforts can be found in the mid 1990's [20]. Then, algorithms were developed to optimize the array geometry for low level of sidelobes [20][21][22][23] or high array gain in one-and two-dimensional spaces with pre-specified source incidence angles [22,[24][25][26][27][28][29].…”