An obstacle detoured routing algorithm based on the enhanced ACS

Abstract

Listen

Translate article

The ACS (ant colony system) algorithm is quite suited for solving the obstacle detoured routing (ODR) problem in the physical design of VLSI circuits. But with the rapid growth of the number of circuit modules, the solution space of primitive ACS is substantially increased. In this paper, we present an heuristic algorithm based on an enhanced ACS to reduce the search space and accelerate the convergence. Firstly, this algorithm initializes the pheromone of a grid graph of a BBL (building block layout) mode layout respectively with an out of usual strategy. After that, we complete the main routing process based on the ACS algorithm with the heuristic of initial pheromone and obtain a feasible solution. In order to accelerate the convergence of primitive ACS, we developed a fast approach called the PEAK-CUT algorithm to shorten the path length. Taking advantage of its merits, we are able to get a local optimum solution closest to that feasible solution quickly. Comparative simulation experiments with respect to the original ACS demonstrate that our enhanced approach is superior to the ACS-only approach in several aspects, such as yielding better quality of solution, faster convergence and resulting in higher complete rate of routing.