Increasing the Capacity of Large-Scale HetNets through Centralized Dynamic Data Offloading

Typically, mobile users cluster around points of interest in dense urban environments such as city centers forming so-called data traffic hot spots and hot zones. To provide capacity to such users efficiently, mobile operators deploy small cells. However, the deployment of heterogeneous networks, which consist of overlaying macro cells and many co-channel small cells, entails many problems. One typical problem is that, more often than not, hot spot users are not covered by the small cells due to the spatially fluctuating nature of the traffic demand. Data offloading, meaning actively shifting macro cell users to small cells, is a promising approach to address this issue. In this paper, we extend a queuing- theoretic model based on the notion of elastic data flows in order to model data offloading, or more specifically, cell range expansion along with inter-cell interference coordination. The model explicitly considers mutual co-channel interference and enables predicting the performance of networks consisting of hundreds of cells with very low computational effort. Based on this model, we present a heuristic centralized data offloading algorithm, which, for a certain traffic demand, is able to increase the 5
^th
percentile of the data flow throughput by a factor of 4.5 and to halve the probability of service unavailability. Moreover, we show that the network capacity can be increased by about 41.3% if data offloading is performed.