LTE radio network planning under uncertainty -

Abstract

Radio network planning and optimization (RNPO) is an essential process for cellular operators and has a significant impact on the operation and cost of the resulting network. The conventional approach for planning cellular networks does not take into account the uncertainty aspects present in the network and considers a deterministic model instead. For instance, the traffic distribution of the users is usually taken at hours of peak demand and the impact of channel variation is modeled using fixed power budgets. Neglecting these uncertain parameters in the planning process leads to performance variation and, thus, requires notable efforts for post-deployment optimization. In this thesis, we adopt a stochastic optimization approach for optimizing cellular network planning under uncertainty. We propose two problem formulations for planning LTE networks taking into account the uncertainty in the location and number of users in addition to the uncertainty in the signal and interference levels. In the first part, an optimization framework is developed for planning LTE cellular networks under demand uncertainty. A two-stage deterministic equivalent of the problem is formulated and solved to optimality. Moreover, a dynamic on-off switching algorithm is developed in order to minimize the energy consumption at off-peak hours. In the second part, a chance constraint approach is adopted for solving the problem of LTE radio network planning under the uncertainty of the signal and interference levels. Site selection and site placement formulations are developed. Both problems are shown to be NP-hard and, thus, a heuristic algorithm is proposed that can achieve notable performance gains compared to conventional approaches with relatively low computational complexity. Performance results are presented and assessed for several cellular network scenarios in order to highlight the effectiveness of the proposed algorithms.