Coverage planning for outdoor wireless LAN systems

Abstract

Wireless LANs (WLANs) are becoming increasingly popular for providing high data rate network access to mobile computers. Most of the currently deployed systems operate in the 2.4 GHz unlicensed frequency band. However, increasing demand for higher data rates and network capacities has led to new system standards for the 5 GHz band. We examine methods for obtaining a close-to-optimal positioning of WLAN access points (APs) and evaluate their performance in a typical downtown or campus environment. The system performance is evaluated using an objective function which aims to maximize both the coverage area and the overall signal quality. The optimization algorithms evaluate this objective function over a discrete search space, thereby considerably reducing the inherent complexity of the problem, while at the same time providing a reasonable approximation to the continuous optimization problem. Numerical results show that random search algorithms, such as simulated annealing, can yield very good solutions. However, the convergence speed of simulated annealing strongly depends on the fine-tuning of simulation parameters and a good choice of the initial set of transmitter positions. Successive removal algorithms, such as pruning, though usually producing sub-optimal solutions, converge in polynomial time. We therefore propose a combination of the two approaches - using pruning for obtaining an initial set of transmitter positions and refining these by using either neighborhood search or simulated annealing.