The ant colony metaphor for multiple knapsack problem

Abstract

This paper presents an Ant Colony (AC) model for the Multiple Knapsack Problem (MKP). The ant colony metaphor, as well as other evolutionary metaphors, was applied successfully to diverse heavily constrained problems. An AC system is also considered a class of multiagent distributed algorithm for combinatorial optimisation. The principle of an AC system is adapted to the MKP. We present some results regarding its performance against known optimum for different instances of MKP. The obtained results show the potential power of this particular evolutionary approach for optimisation problems.