Novel Adaptive Crossover Mechanism for Genetic Algorithm on Integrated Circuit Cell Placement

The increasing complexity of the placement problem, combined with constraints on design time, has led to a heightened reliance on heuristic methods for its resolution. These methods, while managing complexity and addressing the lack of analytical relationships between parameters, also significantly reduce the time required to find solutions. Genetic algorithms (GAs) have emerged as a powerful optimization tool for addressing the complex issue of cell placement within integrated circuits (ICs). This task involves determining the optimal positions for cells on a chip to minimize factors such as signal delay, power consumption, and chip area, and is crucial for the development of efficient and high-performance ICs. A significant advantage of GAs is their ability to maintain and leverage population diversity, which is essential for avoiding premature convergence on suboptimal solutions. Among the components of a GA, the crossover mechanism plays a pivotal role in transferring knowledge across generations and exploring the solution space. This paper introduces a novel approach to the crossover mechanism specifically designed for the cell placement problem. This approach enhances knowledge transfer and more effectively considers the relationships between cells. The results demonstrate improved performance compared to conventional crossover mechanisms.