Wireless home networks are widely deployed due to their low cost, ease of installation, and plug-and-play capabilities with consumer electronic devices. Participating devices may cache continuous media (audio and video clips) in order to reduce the demand for outside-the-home network resources and improve the average delay incurred from when a user references a clip to the onset of its display (startup latency). In this paper, we focus on a home network consisting of a handful of devices. A device may manage its cache at the granularity of either a clip or individual blocks of a clip by employing either a clip-based or a block-based cache replacement technique. It may employ a technique in either stand-alone mode to enhance a local metric such as cache hit rate or make its state dependent on other devices in the home network to enhance a global metric such as average startup latency. These two variants are named greedy and cooperative, respectively. The primary contribution of this paper is to evaluate these alternatives using realistic specifications of a wireless home network. Our key finding is as follows. Our proposed cooperative block-based technique enhances startup latency when compared with clip-based alternatives. With multiple devices where each device acts greedy, clip-based caching is superior. In addition, we show our proposed block-based replacement policy materializes appropriate fraction of each clip across devices dynamically, eliminating the need to compute and place prefetch portions statically.

Abstract — Efficient video-on-demand (VoD) is a highly desired service for media and telecom providers. VoD allows subscribers to view any item in a large media catalog nearly-instantaneously. However, systems that provide this service currently require large amounts of centralized resources and significant bandwidth to accommodate their subscribers. Hardware requirements become more substantial as the service providers increase the catalog size or number of subscribers. In this paper, we describe how cable companies can leverage deployed hardware in a peer-topeer architecture to provide an efficient alternative. We propose a distributed VoD system, and use real measurements from a deployed VoD system to evaluate different design decisions. Our results show that with minor changes, currently deployed cable infrastructures can support a video-on-demand system that scales to a large number of users and catalog size with low centralized resources. I.

Abstract—Nowadays, sharing multimedia contents in P2P systems has become a common practice. In general, it is done in ”download ” mode as any other type of data. However, due to their huge volume and real time constraint, multimedia data present another consumption mode which is the streaming mode. In this mode, rather than downloading the entire data before viewing it, it initiates a stream from the source peer to the destination one. The advantages of this mode are numerous such as latency reduction, copyright preservation (in the sense that the consumption peer does not hold the multimedia content), etc. Nevertheless, due to the unpredictable departure of peers and the real time constraint of the consumed data, the problem of maintaining data availability remains open. In this paper we propose to set up a cache to save the ”last part ” of the ”now playing ” videos. We demonstrate that such a cache can enhance the data availability in these systems while introducing an acceptable additional traffic. Additionally, we develop a cache management policy that automatically adapts the size of the cached data to the system workload. The simulations we conducted show the effectiveness of our approach. Index Terms—Data Availability, Suffix Caching, P2P Streaming I.

Wireless mesh home networks are deployed widely due to their ease of installation and economical prices. A typical network may consist of a handful of devices such as PCs, laptops, wireless consumer electronic devices, and game consoles. Devices may share data by making the state of their caches dependent on one another using a cooperative caching technique such as Domical. This sharing of data at the edges of the network reduces the load on the infrastructure outside of the household, freeing it to service other requests. In this paper, we analyze three local cache replacement techniques designed to enhance average startup latency of streaming media. All three pre-stage a fraction of a clip on a device in anticipation of its future reference in order to display the prefetch portion while streaming its remainder in the background. We use a simulation study of a realistic home network to compare these three techniques with one another when deployed with Domical. Obtained results show show one technique, named urgency-worthiness, is superior to the others when storage is abundant.

Wireless home networks are widely deployed due to their low cost, ease of installation, and plug-and-play capabilities with consumer electronic devices. Participating devices may cache continuous media (audio and video clips) in order to reduce the demand for outside-the-home network resources and enhance the average delay incurred from when a user references a clip to the onset of its display (startup latency). In this paper, we focus on a home network consisting of a handful of devices configured with a mass storage device to cache data. A cooperative caching technique may manage the available cache space at the granularity of either a clip or individual blocks of a clip. The primary contribution of this paper is to evaluate these two alternatives using realistic specifications of a wireless home network, identifying factors that enable one to outperform the other. 1

Domical is a recently introduced cooperative caching technique for streaming media (audio and video clips) in wireless home networks. It employs asymmetry of the available link bandwidths to control placement of data across the caches of different devices. A key research question is what are the merits of this design decision. To answer this question, we compare Domical with DCOORD, a cooperative caching technique that ignores asymmetry of network link bandwidths in its caching decisions. We perform a qualitative and quantitative analysis of these two techniques. The quantitative analysis focuses on startup latency defined as the delay incurred from when a device references a clip to the onset of its display. Obtained results show Domical enhances this metric significantly when compared with DCOORD inside a wireless home network. The qualitative analysis shows DCOORD is a scalable technique that is appropriate for networks consisting of many devices. While Domical is not appropriate for such networks, we do not anticipate a home network to exceed more than a handful of wireless devices. 1