Stochastic Leader Gravitational Search Algorithm for Enhanced Adaptive Beamforming Technique.

Area of Science:

• Artificial Intelligence
• Optimization Algorithms
• Computational Intelligence

Background:

• Standard Gravitational Search Algorithm (SGSA) can converge to suboptimal results due to its deterministic nature.
• Randomization is crucial for enhancing the global search capabilities of optimization algorithms.
• Existing GSA variants may not fully address the trade-off between global exploration and rapid convergence.

Purpose of the Study:

• To introduce a novel optimization algorithm, the Stochastic Leader Gravitational Search Algorithm (SL-GSA), designed to overcome the limitations of SGSA.
• To enhance the global search ability and convergence rate of the Gravitational Search Algorithm through a randomized approach.
• To validate the efficacy of SL-GSA on benchmark functions and a real-world application.

Main Methods:

• The proposed SL-GSA algorithm incorporates randomized selection of 'k' agents to improve global exploration.
• A mechanism for gradually reducing the agent population by eliminating underperforming agents is implemented for faster convergence.
• Performance evaluation involves testing SL-GSA on six standard benchmark functions and the Minimum Variance Distortionless Response (MVDR) beamforming technique.

Main Results:

• SL-GSA demonstrated superior performance compared to SGSA and its variants across various benchmark functions.
• The algorithm exhibited a significantly improved convergence rate and solution quality.
• Application to MVDR beamforming confirmed SL-GSA's effectiveness in solving real-world optimization challenges.

Conclusions:

• The SL-GSA effectively balances global search and rapid convergence, outperforming traditional GSA.
• The randomized approach integrated into SL-GSA enhances its robustness and efficiency.
• SL-GSA presents a promising alternative for complex optimization problems in diverse fields.