Over the last few years, with the rise of phablet devices, there has been a tremendous growth in mobile data demand, followed by increased investments in development of the network infrastructures. However, rapid development of technologies to support the inevitable traffic surge poses new challenges to mobile network operators (MNOs) in terms of revenue. If MNOs want to turn the intriguing opportunity of such a large market into revenue uplift, they should scramble on multiple fronts such as cutting the operational expenditure costs, increasing the average revenue per user for the current customers and enhancing the acquisition of new customers, etc. In this thesis, a variety of deterministic and stochastic optimization techniques aimed at providing innovative solutions for suc...[ Read more ]

Over the last few years, with the rise of phablet devices, there has been a tremendous growth in mobile data demand, followed by increased investments in development of the network infrastructures. However, rapid development of technologies to support the inevitable traffic surge poses new challenges to mobile network operators (MNOs) in terms of revenue. If MNOs want to turn the intriguing opportunity of such a large market into revenue uplift, they should scramble on multiple fronts such as cutting the operational expenditure costs, increasing the average revenue per user for the current customers and enhancing the acquisition of new customers, etc. In this thesis, a variety of deterministic and stochastic optimization techniques aimed at providing innovative solutions for such challenges are explored.

In particular, we focus on tractable theoretical methods to analyze, model and dynamically optimize network planning and operation for the heterogeneous cellular networks in the following problems: (i) resource allocation, (ii) base stations management and, (iii) user association. Unlike most of existing literature, we address practical implications and limitations in our analytical modelings. Our theoretical findings provide insightful theories and novel closed-form expressions for system parameters that can be integrated in other general models developed for cellular networks. The key contributions can be summarized as (i) optimally solving instantaneous resource allocation problem previously identified as NP-hard under different formulation (ii) stochastic queue modeling of heterogeneous users followed by a tractable and tight convex approximation for it (iii) joint optimization of 3-tuple system parameters to manage base stations in a large-scale heterogeneous network (iv) deriving closed-form expression for cell-load which provides significant insights for base station sleep strategies (v) developing an optimal distributed user association that bypasses need for a centralized unit. Our analytical solutions are corroborated by extensive numerical simulations.