Maximizing Cache Value for Distributing Content via Small Cells in 5G

Abstract

This paper discusses distribution of content to mobile users via small cells (SCs) in 5G. Users may simultaneously access between one and some maximum number of SCs. Files are stored in remote locations as a collection of source symbols and some number of RaptorQ-encoded symbols generated from these source symbols are cached in the SCs. Each file is characterized by the number of encoded symbols required to reconstruct the file at the user equipment and its download preference probability. Users recover the file by downloading a number of encoded symbols slightly larger than the number of source symbols. If a sufficient number of encoded symbols are not available from the SCs, the remainder are backhauled. Presented in this paper is an algorithm of low complexity to maximize the average value of using cache to reduce backhaul. This paper contributes to the literature by presenting an n log n algorithm to solve the optimization problem exactly, extending previous results from constant files size for all files to arbitrary actual files sizes for all files, and extending distribution portions from continuous fractions of files to an integer number of symbols.