A Mahalanobis Surrogate-Assisted Ant Lion Optimization and Its Application in 3D Coverage of Wireless Sensor Networks

Area of Science:

• Engineering
• Computer Science
• Optimization Algorithms

Background:

• Metaheuristic algorithms are crucial in engineering but often require extensive computation time for solutions.
• Existing algorithms face challenges with computationally expensive problems, demanding more efficient approaches.

Purpose of the Study:

• To develop a novel Mahalanobis sampling surrogate model to assist the Ant Lion optimization algorithm (MSAAO).
• To enhance the efficiency and performance of metaheuristic algorithms for computationally expensive problems.

Main Methods:

• Integration of three surrogate models: global, Mahalanobis sampling, and local surrogate models.
• Utilizing Mahalanobis distance to mitigate variable correlation interference and identify optimal samples.
• Sorting ant distances and selecting samples for model training.

Main Results:

• MSAAO demonstrated superior competitiveness against other algorithms on seven benchmark functions.
• Simulations showed MSAALO improved average coverage by 2.122% (radii) and 1.718% (nodes).

Conclusions:

• The proposed MSAALO effectively accelerates metaheuristic computations for expensive problems.
• Mahalanobis distance sampling is a viable strategy for improving optimization algorithm efficiency and coverage.