Optimization models and algorithms for downlink UMTS radio planning

The problem of planning third generation UMTS networks with a W-CDMA radio interface in investigated. In previous work, we have proposed discrete optimization models and algorithms for supporting the decisions on where to locate base stations in which antenna configuration to select considering quality constraints for the uplink (mobile to base station) direction. The signal-to-interference ratio (SIR) is considered as quality measure and we aim at a trade-off between maximizing coverage and minimizing installation costs. In this paper we present two mathematical programming models for locating directive base stations considering downlink (base station to mobile) direction and assuming a power-based as well or a SIR-based power control mechanism. The downlink direction is expected to be particularly relevant in the presence of asymmetrical traffic deriving, for instance, from data service. A randomized greedy procedure as well as a Tabu search algorithm is adapted to find good approximate solution of the resulting NP-hard downlink BS location problem. Experimental results obtained for realistic instances with voice as well as data traffic are reported and they are compared with those provided by the uplink models and algorithms.