Comparison of ant-inspired search techniques for software release planning

The software releasing planning has been proven to be a NP-complete problem. For a metaheuristic algorithm, too much emphasis on the exploitation of the search experience gathered so far can frequently converge to local optima. On the contrary, too much emphasis on the exploration of unvisited or relatively unexplored search space regions can cause an unstable state. On the basis of the principles of ant colony optimization (ACO), we therefore propose a constructive-oriented iterative algorithm which is named as MMAS_DRW-SRP to overcome the mentioned drawbacks. Experimental results indicate that MMAS_DRW-SRP is a significant improvement compared with the other ACO-based approaches AS-SRP, AS_Rank-SRP, ACS-SRP, and MMAS-SRP at solving the software release planning problem.