Toward Wi-Fi AP-Assisted Content Prefetching for an On-Demand TV Series: A Learning-Based Approach

Abstract

The emergence of smart Wi-Fi APs (Access Point), which are equipped with huge storage space, opens a new research area on how to utilize these resources at the edge network to improve users' quality of experience (QoE) (e.g., a short startup delay and smooth playback). One important research interest in this area is content prefetching, which predicts and accurately fetches contents ahead of users' requests to shift the traffic away during peak periods. However, in practice, the different video watching patterns among users, and the varying network connection status lead to the time-varying server load, which eventually makes the content prefetching problem challenging. To understand this challenge, this paper first performs a large-scale measurement study on users' AP connection and TV series watching patterns using real-traces. Then, based on the obtained insights, we formulate the content prefetching problem as a Markov Decision Process (MDP). The objective is to strike a balance between the increased prefetching&storage cost incurred by incorrect prediction and the reduced content download delay because of successful prediction. A learning-based approach is proposed to solve this problem and another three algorithms are adopted as baselines. In particular, first, we investigate the performance lower bound by using a random algorithm, and the upper bound by using an ideal offline approach. Then, we present a heuristic algorithm as another baseline. Finally, we design a reinforcement learning algorithm that is more practical to work in the online manner. Through extensive trace-based experiments, we demonstrate the performance gain of our design. Remarkably, our learning-based algorithm achieves a better precision and hit ratio (e.g., 80%) with about 70% (resp. 50%) cost saving compared to the random (resp. heuristic) algorithm.Index Terms-Wi-Fi AP, content prefetching, learning-based approach.

Content Prefetching of Mobile Caching Devices in Cooperative D2D Communication Systems

“…When discussing the caching schemes of nomadic/fixed helpers, we assume that UEs can obtain the short-term content preferences through content prediction and content recommendation algorithms [40][41][42][43][44]. Suppose that highcapacity base stations capable of supplying sufficient amounts of data are installed at a distance of BS d along the road, and mobile helpers move in both directions on the xaxis, as shown in Fig.…”

“…In this paper, in order to overcome the problems associated with conventional D2D caching technology, we propose a system in which the cache of a helper is updated by predicting the contents that will be accessed in the near future by the UEs around the helper. Considering the average characteristics of all UEs, it may be difficult to predict which content will be requested in the distant future, but it may be possible for each UE to accurately predict the content that will be consumed in the near future [39][40][41][42][43][44]. For example, if a long video such as a movie is divided into multiple chunks and a user is currently viewing some chunk, it is likely that the next chunk will be requested after a short time.…”

“…If a user is currently watching a short video, such as a YouTube video, there is a high probability that she or he will request one of the relevant contents recommended by the content recommendation system next time [40][41]. If a user is watching an episode of a series, the content to be consumed next may be the next episode [42]. Compared to the content caching based on the long-term content popularity [31][32][33], a higher hit ratio may be achievable using content prefetching based on the short-term content preference of individual UE [40][41][42][43][44] predicted using the currently and recently consumed contents in addition to the long-term popularity.…”

“…If a user is watching an episode of a series, the content to be consumed next may be the next episode [42]. Compared to the content caching based on the long-term content popularity [31][32][33], a higher hit ratio may be achievable using content prefetching based on the short-term content preference of individual UE [40][41][42][43][44] predicted using the currently and recently consumed contents in addition to the long-term popularity. However, at the peak time, it is difficult to provide additional data due to almost saturated access link, so the UE, the base station, and the helpers around the UE cannot directly prefetch content through the access link.…”

“…As opposed to the conventional D2D caching scheme which relies on content popularity for a long time and a large number of UEs, the content prefetching scheme proposed in this paper uses content prediction for a short time and individual UE. In general, content prefetching is mainly performed for short startup delay, smooth playback, seamless handover, etc., by a UE or a base station that is in the reception path of the network [41][42][43][44]. The proposed content prefetching scheme, however, is a part of a D2D caching system for data offloading, and content prefetching is performed by another helper that is far from the UE and thus not in the normal reception path.…”