SortingHat: Holisitic Resource Optimization for Heterogeneous Cellular Networks

While heterogeneous cellular networks offer the potential to improve network performance by augmenting coverage and enhancing traffic capacity, resource allocation is extremely complicated due to the heterogeneous and unplanned nature of base station deployments. In this paper, we present SortingHat, a tractable optimization model which leverages the advantages of the different heterogeneous components in the network to achieve 3 main objectives: user-cell pairing, power control, and channel assignment. SortingHat achieves this by casting the resource allocation problem in a form which can be solved using iterative optimization algorithms such as the constrained convex-concave procedure (CCCP). While algorithms for user-cell pairing, power control, and channel assignment have been studied extensively in the past, this reformulation enables SortingHat to take a holistic approach to resource allocation and jointly optimize all three objectives simultaneously. Numerical results show greater than 30-100% improvement in median user throughput relative to prior works in which the objectives are considered separately, rather than jointly.