2018/05/17 by Θεόδωρος Γιαννακάς, Giannakas, Theodoros, Pavlos Sermpezis +3
Computer Science · #Caching and Content Delivery #Cooperative Communication and Network Coding #FOS: Computer and information sciences #Networking and Internet Architecture (cs.NI) #Recommender Systems and Techniques
paper · pdf · doi:10.48550/arxiv.1805.06670
openalex publication_date 2018/05/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Caching has been successfully applied in wired networks, in the context of\nContent Distribution Networks (CDNs), and is quickly gaining ground for\nwireless systems. Storing popular content at the edge of the network (e.g. at\nsmall cells) is seen as a `win-win' for both the user (reduced access latency)\nand the operator (reduced load on the transport network and core servers).\nNevertheless, the much smaller size of such edge caches, and the volatility of\nuser preferences suggest that standard caching methods do not suffice in this\ncontext. What is more, simple popularity-based models commonly used (e.g. IRM)\nare becoming outdated, as users often consume multiple contents in sequence\n(e.g. YouTube, Spotify), and this consumption is driven by recommendation\nsystems. The latter presents a great opportunity to bias the recommender to\nminimize content access cost (e.g. maximizing cache hit rates). To this end, in\nthis paper we first propose a Markovian model for recommendation-driven user\nrequests. We then formulate the problem of biasing the recommendation algorithm\nto minimize access cost, while maintaining acceptable recommendation quality.\nWe show that the problem is non-convex, and propose an iterative ADMM-based\nalgorithm that outperforms existing schemes, and shows significant potential\nfor performance improvement on real content datasets.\n