A permutation-coded evolutionary strategy for multi-objective GSM network planning

Abstract The base station placement problem, with n potential candidate sites is NP-Hard with 2n solutions (Mathar and Niessen, Wirel. Netw. 6, 421–428, 2000). When dimensioned on m unknown variable settings (e.g., number of power settings + number of tilt settings, etc.) the computational complexity becomes (m+1)n (Raisanen, PhD. thesis, 2006). We introduce a novel approach to reduce the computational complexity by dimensioning sites only once to guarantee traffic hold requirements are satisfied. This approach works by determining the maximum set of service test points candidate sites can handle without exceeding a hard traffic constraint, TMAX. Following this, the ability of two evolutionary strategies (binary and permutation-coded) to search for the minimum set cover are compared. This reverses the commonly followed approach of achieving service coverage first and then dimensioning to meet traffic hold. To test this approach, three realistic GSM network simulation environments are engineered, and a series of tests performed. Results indicate this approach can quickly meet network operator objectives.