Extreme learning machine-driven joint user mobility and content popularity-based proactive caching in multi-tier wireless networks

Abstract

Proactive caching in cache-enabled wireless networks for 5G and 6G is a promising technique for alleviating the immense data traffic on bandwidth-limited backhaul links while minimizing content latency and handovers, particularly in multi-tier cellular networks (MTCNs). In a MTCNs, caching nodes are densely deployed, reducing the distance between users and their associated base stations (BS), which significantly enhances the performance of the caching system. In this context, various efficient content caching techniques have been proposed for MTCNs. However, most of these methods do not consider user mobility, which can significantly impact the performance of content caching in MTCNs. This paper explores the impact of user mobility and the popularity of the content in cache-enabled MTCNs and proposes two proactive caching techniques tailored to these factors. The first method is the extreme learning machine-based mobility-aware Proactive (E-MAP) Caching Scheme, which focuses solely on the mobility of the user. The second method, the Extreme Learning Machine-based Mobility and Popularity Aware Proactive (E-MAPP) caching scheme, considers both user movement and content popularity. An optimization problem is formulated to minimize the average content latency for the caching system. Numerical simulations demonstrate that the proposed technique outperforms traditional methods in terms of average content latency and cache hit ratio (CHR).