A lyrebird optimizer with mimicry and territory protection mechanisms for spectrum sharing MIMO system with intelligent reflecting surface

Wireless communication growth and rising spectrum demand have highlighted the necessity for effective spectrum sharing, particularly in cognitive radio networks. Intelligent reflecting surface (IRS) can improve signal propagation and reduce interference for secondary users. However, non-convex restrictions and channel uncertainty continue to make simultaneous IRS setup and beamforming optimization difficult. This paper provides an Improved Lyrebird Optimization Algorithm (Improved LOA) to address these issues. The Improved LOA improves on the standard Lyrebird Optimization Algorithm (LOA) by incorporating two biologically inspired mechanisms: the mimicry mechanism, which boosts exploitation by allowing agents to recall and refine previously successful positions, and the territory protection mechanism, which encourages exploration by repelling agents from densely populated areas. These processes enhance convergence speed, solution variety, and robustness. Simulation results show that Improved LOA outperforms traditional algorithms (e.g., standard LOA, Particle Swarm Optimization (PSO), Differential Evolution (DE), Gradient-Based Optimizer (GBO), and Weighted Differential Evolution (WDE)) in terms of achievable rate, energy efficiency, and optimization reliability in IRS-assisted cognitive MIMO systems. Improved LOA outperforms traditional methods by up to 35% in maximum achievable rate and 30% in median rate, while maintaining energy-efficient performance with up to 85.4% transmitted power reduction compared to baseline techniques.