On the deployment of picocellular wireless infrastructure

Abstract

Wireless LANs are becoming increasingly common in both home and office networks. Additionally, wireless Internet service providers have begun installing public WLANs in airports, hotels, and other frequently trafficked areas. This article analyzes deployment strategies for such networks, specifically for possible future systems utilizing bands of up to 60 GHz. A major part of the installation cost stems from network planning, wiring, and manpower for setting up base stations. It is therefore crucial to find ways to simplify the network installation and reduce the deployment costs, while at the same time maintaining the desired system performance. We show that even simple installation rules can often achieve adequate coverage results. However, proper network planning is in some situations necessary to achieve adequate coverage. Therefore, we examine different combinatorial optimization methods for obtaining close to optimal positioning of WLAN access points and compare the performance of the proposed algorithms to the simple installation methods. The optimization algorithms used in this article evaluate an objective function that aims to maximize both the coverage area and the overall signal quality over a discrete search space. Random search algorithms can yield very good solutions, but often exhibit difficult convergence properties. Successive removal algorithms, such as pruning, converge in polynomial time, but usually produce suboptimal solutions. We therefore propose a combination of the two approaches, using pruning to obtain an initial set of the base station positions and refining these by using neighborhood search or simulated annealing.