Collaborative Filtering-Based In-Network Content Caching for 5G Networks

As the number of wireless devices, the demand for high data rates, and the need for always-on connectivity are growing and becoming more stringent with the evolvement and emergence of 5G systems, network engineers and researchers are being faced with new unique challenges that need to be addressed. Among many challenges, traffic congestion bottleneck at backhaul links arising from the massive connections emerges as one key challenge that 5G systems need to tackle. One solution approach that has been investigated as a key enabler for addressing such traffic bottlenecks is in-network content caching, where frequently-accessed content is placed closer to end users at the network edges so that the amounts of traffic that need to traverse core network and backhaul links are reduced.

In this thesis, we propose a content placement and caching technique that leverages collaborative filtering and k-means clustering to make efficient content placement decisions, thereby reducing downloading time and backhaul traffic. We simulate the proposed technique and compare it with two other existing prefetching techniques, and show that our approach outperforms existing ones by achieving higher hit ratios, lesser backhaul traffic, and lesser download times. The proposed technique improves the users' quality of experience by minimizing network latency and alleviating back-haul traffic congestion.

Major Advisor: Bechir Hamdaoui
Committee: Anita Sarma
Committee: Lizhong Chen
GCR: Maggie Niess