Decentralized Evolutionary Optimization Approach to the p-Median Problem

摘要

The facility location problem also known as p-median problem concerns the positioning of facilities such as bus-stops, broadcasting stations or supply stations in general. The objective is to minimize the weighted distance between demand points (or customers) and facilities. In general there is a trend towards networked and distributed organizations and their systems, complicating the design, construction and maintenance of distributed facilities as information is scattered among participants while no global view exists. There is a need to investigate distributed approaches to the p-median problem. This paper contributes to research on location problems by proposing an agent oriented decentralized evolutionary computation (EC) approach that exploits the flow of money or energy in order to realize distributed optimization. Our approach uses local operators for reproduction like mutation, recombination and selection finally regulated by market mechanisms. This paper presents two general outcomes of our model: how adaptation occurs in the number and strategies of agents leading to an improvement at the system level. The novelty of this approach lies in the biology-inspired bottom-up adaptation method for inherent distributed problems. It is applied to the uncapacitated p-median problem but is also intended to be general for a wide variety of problems and domains, e.g. wireless sensor networks.