Queue-Aware Cell Activation and User Association for Traffic Offloading via Dual-Connectivity

With the objective of reducing energy cost, we study the stochastic optimization of traffic off-loading via dual-connectivity by joint cell activation and user association. Particularly, explicitly considering the dynamic effects of time-varying and random traffic arrivals, we formulate a stochastic problem that quantifies the tradeoff between energy cost and queuing delay. Employing two-time-scale Lyapunov optimization, the formulated problem is transformed as a user association problem at each time slot and a small-cell activation problem at each frame. However, the user association problem turns out to be a mixed-integer nonlinear programming problem and is difficult to be transformed into a convex problem, and traditional game-theoretic solutions, such as Nash equilibrium, cannot be used to solve it. Then, we introduce a two-sided many-to-one matching game and propose a distributed user association algorithm that converges to a local optimum. On the other hand, the cell activation problem belongs to the class of maximum facility location problems that are generally NP-hard. Additionally, its objective function is not a submodular one. Thus, we present an iterative and heuristic algorithm that finds the best set of active small-cell BSs by repeatedly solving the user association problem with different sets of active small cells in each iteration. Furthermore, an online two-time-scale joint cell activation and user association algorithm are developed. Finally, numerical results show the effectiveness of the matching-based user association algorithm on traffic off-loading improvement, the heuristic cell activation algorithm on trading off between energy cost reduction and network throughput enhancement, and the online joint algorithm on balancing between energy cost reduction and queuing delay performance.