Channel routing based on ant colony adaptive behavior model

Abstract

Based on the comparative analysis of existing approaches and methods for solving routing, routing completion, and rerouting of connections in VLSI, multiagent intelligent optimization methods are used. They are based on the simulation of adaptive behavior of an ant colony. The problem considered in this paper is represented by a set of components of the ant colony algorithm. Heuristics governing the behavior of an ant as it moves in the search graph are developed. A distinctive feature of the proposed algorithm is its capability to take into account some characteristics, such as the distribution of resources, hindrance effects (blockings), the number of transitions between layers, the number of inflections, etc., which are very difficult to take into account when calculating how the wave propagates. The routing algorithm was tested and compared with other known algorithms on a set of benchmarks. Compared with the available algorithms, the quality of solutions is improved by up to 3%. A promising direction of improving the algorithm is the use of an extended routing domain that admits a small increase in the route length but minimizes the hindrances. The efficiency of the algorithm can be improved by an adaptive control of algorithm parameters.