Energy-Efficient Context-Aware Matching for Resource Allocation in Ultra-Dense Small Cells

With the explosive growth of mobile data traffic and rapidly rising energy price, how to implement caching at small cells in an energy-efficient way is still an open problem and requires further research efforts. In this paper, we study the energy-efficient context-aware resource allocation problem, which falls into the category of mixed integer nonlinear programming (MINLP) and is NP-hard. To provide a tractable solution, the MINLP problem is decoupled and reformulated as a one-to-one matching problem under two-sided preferences, which are modeled as the maximum energy efficiency that can be achieved under the expected matching. An iterative algorithm is developed to establish preference profiles by employing nonlinear fractional programming and Lagrange dual decomposition. Then, we propose an energy-efficient matching algorithm based on the Gale-Shapley algorithm, and provide the detailed discussion and analysis of stability, optimality, implementation issues, and algorithmic complexity. The proposed matching algorithm is also extended to scenarios with preference, indifference, and incomplete preference lists by introducing some tie-breaking and preference deletion rules. The simulation results demonstrate that the proposed algorithm achieves significant performance and satisfaction gains compared with the conventional algorithms.