Optimal Project Scheduling and Staffing with Hybrid SFLA-Tabu Algorithm: A Performance Evaluation

Abstract

Planning a large-scale software project involves the objectives of optimal ordering of a set of activities and allocation of staff to activities. Current adopted method presents difficulty in reaching optimal good solutions when the two objectives are combined. This study proposes a hybrid SFLA-TABU search algorithm for solving project scheduling and staffing problem with the two objectives combined. The hybrid algorithm retains the frame work of SFLA algorithm but within the local memetic evolution, employs the neighborhood structure method of TABU search and its avoidance of already explored area in the solution space to move towards optimal solution. The performance of the algorithm was evaluated on 12 standard problem instances using hit rate, average fitness and lenght. Experimental results showed that the proposed algorithm outperforms original SFLA and TABU search in 100% of the considered instances based on hit rate and quality of solution (measured as average fitness). It how ever consumes more time than the other two algorithms in about 60% of the problem instances considered. Our findings revealed that the algorithm can be applied to handle various sizes of software projects if given enough time. The results prove the effectiveness, robustness and scalability of the proposed approach