Learning to Cache: Federated Caching in a Cellular Network With Correlated Demands

Abstract

In this paper, the problem of distributed content caching in a small-cell Base Stations (sBSs) wireless network that maximizes the cache hit performance is considered. Most of the existing works consider static demands, however, here, data at each sBS is considered to be correlated across time and sBSs. Federated learning (FL) based caching strategy is proposed which is assumed to be a weighted combination of past caching strategies of neighbouring base stations. A high probability generalization guarantees on the performance of the proposed federated caching strategy is derived. The theoretical guarantee provides following insights on obtaining the caching strategy: (i) run regret minimization at each sBS to obtain a sequence of caching strategies across time, and (ii) maximize an estimate of the bound to obtain a set of weights for the caching strategy which depends on the discrepancy. Also, theoretical guarantee on the performance of the least recently frequently used (LRFU) caching strategy is derived. Further, FL based heuristic caching algorithm is also proposed. Finally, it is shown through simulations using Movie Lens dataset that the proposed algorithm significantly outperforms the recent online learning algorithms.

Wireless Edge Caching and Content Popularity Prediction Using Machine Learning

“…The data is collected and stored at multiple edge nodes, and a model is trained from the distributed data without sending the data from the nodes to a central node. This variant of distributed learning (model training) from a federation of edge nodes is known as FL [9]. In FL based caching the users are collaboratively trained to increase the overall speed.…”

“…The weights w T +1 jb(b ′ ) as well as α b,t are chosen in such a way the average cache hit is maximized. Thus assuming structured cache placement, a high probability bound on the conditional average cache hit is derived using martingale difference equation (see [9]). The insights provided by the bounds further helps in designing iterative FL based caching strategies.…”

“…Building such a wireless network architecture will enable operators to optimize the 5G and beyond wireless networks, where large amount of data can be exploited by using ML tools for content popularity estimation and placement. Second, after an overview, (a) application of reinforcement learning jointly with recommendation in edge caching has been demonstrated, (b) content popularity prediction using Transfer Learning (TL) [12] (c) popularity prediction using Deep Learning (DL) and (d) Federated Learning (FL) based caching strategy is presented [9]. Finally, we present the future directions and challenges to conclude the article.…”

“…To strike a balance between increasing mobile traffic and user's quality-of-experience, edge caching and computing can be seen as invariable solution by bringing storage and computation close to the edge. Both edge caching and computation helps in storing and computation helps in the implementation of ML algorithms at the edge, hence making the edge intelligent and further reducing the burden on the backhaul (see [6], [7], [8], or [9]). The standard simplified algorithms such as least frequently used (LFU), least recently used (LRU), least frequently/recently used (LRFU) and other variants, can be inefficient when it comes to dynamic environments, since they do not take into account the correlation and non-stationarity of the demand requests.…”

A Survey of Edge Caching: Key Issues and Challenges

“…Since each SBS is just equipped with limited caching space, it is necessary for SBSs to adjust their caching targets based on the request record. Krishnendu et al [34] presented a novel edge caching strategy for the SBS caching, which addressed the LP problem of maximizing average CHR in an approximate method. Results showed that the designed algorithm has a better performance compared with traditional greedy caching, Least Recently Used (LRU), and Least Frequently Used (LFU).…”