High-level synthesis by dynamic ant

In this paper a new algorithm called Dynamic Ant is introduced. It was a combination of Ant Colony Optimization techniques and Dynamic Niche Sharing scheme. The interesting point of this algorithm is that it is simple to be implemented and could be well matched with existing design algorithms by adding the heuristic weights to speed up the algorithm. The algorithm uses the problem state structure as in reinforcement-learning algorithm, but the storage explosion is prevented by mean of the pheromone trail. This algorithm was investigated by data path design problem of High-Level Synthesis of which has a large number of design steps and design techniques.