Abstract
An experience oriented-convergence improved gravitational search algorithm (ECGSA) based on two new modifications, searching through the best experiments and using of a dynamic gravitational damping coefficient (α), is introduced in this paper. ECGSA saves its best fitness function evaluations and uses those as the agents’ positions in searching process. In this way, the optimal found trajectories are retained and the search starts from these trajectories, which allow the algorithm to avoid the local optimums. Also, the agents can move faster in search space to obtain better exploration during the first stage of the searching process and they can converge rapidly to the optimal solution at the final stage of the search process by means of the proposed dynamic gravitational damping coefficient. The performance of ECGSA has been evaluated by applying it to eight standard benchmark functions along with six complicated composite test functions. It is also applied to adaptive beamforming problem as a practical issue to improve the weight vectors computed by minimum variance distortionless response (MVDR) beamforming technique. The results of implementation of the proposed algorithm are compared with some well-known heuristic methods and verified the proposed method in both reaching to optimal solutions and robustness.
Highlights
Adaptive array techniques have been utilized to mitigate multipath fading and high co-channel interference in mobile, wireless, radar communications, satellite and many other similar applications to achieve robust performance and high data transmission rate [1]
The minimum variance distortionless response (MVDR) integrated with experience oriented-convergence improved gravitational search algorithm (ECGSA), GSA and particle swarm optimization (PSO) method will optimize the signal to interference noise ratio (SINR) via the complex weights, as described in detail previously [28]
An experience oriented-convergence improved gravitational search algorithm with dynamic gravitational damping coefficient, α, and searching through the best experiments is presented in this paper
Summary
Adaptive array techniques have been utilized to mitigate multipath fading and high co-channel interference in mobile, wireless, radar communications, satellite and many other similar applications to achieve robust performance and high data transmission rate [1]. ECGSA benefits from two modifications: one is to save the best fitness function evaluation of the agents during search process and to treat them as the agents’ effective positions in terms of applying force to other agents. Two distinct modifications, using the best found experiments as the agents’ present positions and introducing a dynamic gravitational damping coefficient value, are introduced to GSA as follows: A. The standard GSA does not retain the best fitness function evaluations that the agents find during search process This is in contradiction of other stochastic search algorithms such as PSO [29], GA [30], DE [31] and simulated annealing (SA) [32]. By this modification no optimal found trajectory is lost, and the movement of any agent is started from its best experiment at any iteration, which achieves good local search around the optimal found solutions
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
