Ultra Dense Edge Caching Networks With Arbitrary User Spatial Density

Cache-enabled small cells can be an effective solution to deliver contents to mobile users with much lower power and latency. While the trend for getting smaller and denser cells is clear, interference will soon become unmanageable and an obstacle when the number of content requests is massive. Moreover, content request is seldom a spatially homogeneous process due to physical impediments (e.g., buidings) and social activities, which makes resource allocation for content delivery more challenging. In this paper, we consider an ultra-dense network (UDN) in which content requests are served by cache-enabled access nodes which can either be active for delivering contents to users, or inactive to reduce interference and network energy consumption. Our aim is to devise an approach that can locally adapt the caching node density and content caching probabilities to accommodate any arbitrary user density and content request for maximizing the network's successful content delivery probability (SCDP). With a non-homogeneous spatial distribution for user equipments (UEs), we find that user-load, a parameter at the access node, plays a major role in the overall optimization. Simulation results illustrate that the proposed method can obtain superior performance against the considered benchmarks, with up to 150-160% increase, and our optimized solutions effectively adapt to the spatial-dependent user density.