In recent years, adaptive HTTP streaming protocols have become the de-facto standard in the industry for the distribution of live and video-on-demand content over the Internet. This paper presents SmoothCache 2.0, a distributed cache platform for adaptive HTTP live streaming content based on peer-to-peer (P2P) overlays. The contribution of this work is twofold. From a systems perspective, to the best of our knowledge, it is the only P2P platform which supports recent live streaming protocols based on HTTP as a transport and the concept of adaptive bitrate switching. From an algorithmic perspective, the system describes a novel set of overlay construction and prefetching techniques that realize: i ) substantial savings in terms of the bandwidth load on the source of the stream, and ii ) CDN-quality user experience in terms of playback latency and the watched bitrate. In order to support our claims, we conduct a methodical evaluation on thousands of real consumer machines.

Peer-to-peer (P2P) technology has long been considered a natural complement to standard CDN infrastructure for video distribution since it greatly reduces costs and improves quality of user experience. However, P2P solutions have traditionally required the installation of additional software or plugins to be deployed, which significantly hinders adoption. In this paper, we present Hive.js, a browser-based plugin-less distributed caching platform for video streaming. Hive.js is layered over WebRTC, a new set of HTML5 APIs for direct browser-to-browser communication, and it is designed to transport adaptive HTTP streaming protocols, specifically MPEG-DASH. Initial results obtained by evaluating Hive.js in a controlled test environment show that our approach significantly reduces the load on CDN infrastructure and does not sacrifice quality of user experience.

The WebRTC framework enables direct browser-to-browser communication for VoIP, video-conferencing but also generic javascript-based peer-to-peer (P2P) applications. In this paper, we address a major deficiency of the framework, that is the absence of a background transfer protocol. In doing so, we enable the development of un-intrusive data-intensive P2P applications, such as file-sharing and content delivery, on top of WebRTC. This work constitutes the first step in the process of studying and improving WebRTC's network stack in the context of a real use-case, that is a commercial distributed content delivery application.

In peer-to-peer applications deployed on the Internet, it is common to assign greater responsibility to supernodes, which are usually peers with high computational power, large amount of memory, or high network bandwidth capacity. In this paper, we describe a practical solution to the problem of supernode selection, that is the process of discovering the best peers in the network by some application-specific metric. We provide a distributed heuristic that allows to identify the best K nodes in the P2P overlay, by taking into consideration the realities of actual deployments, such as the presence of NATs. Our approach consists of an epidemic protocol which does not require new connections to be established, but rather relies on established connections, such as the ones provided by a NAT-resilient peer sampling framework. We support our claims with a thorough evaluation of our solution in simulation and in a real deployment on thousands of consumer machines.

The WebRTC framework enables direct browser-to-browser communication for VoIP, video-conferencing but also generic javascript-based peer-to-peer (P2P) applications. In this paper, we address a major deficiency of the framework, that is the absence of a background transfer protocol. In doing so, we enable the development of un-intrusive data-intensive P2P applications, such as file-sharing and content delivery, on top of WebRTC. This work constitutes the first step in the process of studying and improving WebRTC's network stack in the context of a real use-case, that is a commercial distributed content delivery application.

State of the art gossip protocols for the Internet are based on the assumption that connection establishment between peers comes at negligible cost. Our experience with commercially deployed P2P systems has shown that this cost is much higher than generally assumed. As such, peer sampling services often cannot provide fresh samples because the service would require too high a connection establishment rate. In this paper, we present the wormhole-based peer sampling service (WPSS). WPSS overcomes the limitations of existing protocols by executing short random walks over a stable topology and by using shortcuts (wormholes), thus limiting the rate of connection establishments and guaranteeing freshness of samples, respectively.We show that our approach can decrease the connection establishment rate by one order of magnitude compared to the state of the art while providing the same levels of freshness of samples. This, without sacrificing the desirable properties of a PSS for the Internet, such as robustness to churn and NAT-friendliness. We support our claims with a thorough measurement study in our deployed commercial system as well as in simulation.

In this work, we present a distributed caching solution which addresses the problem of efficient delivery of HTTP live streams in large private networks. With our system, we have conducted tests on a number of pilot deployments. The largest of them, with 3000 concurrent viewers, consistently showed that our system saves more than 90% of traffic towards the source of the stream while providing the same quality of user experience of a CDN. Another result is that our solution was able to reduce the load on the bottlenecks in the network by an average of 91.6%.

This paper presents the design and implementation of the Dynamic Transport Library (DTL), a UDP-based reliable transport library, initially designed for - but not limited to - peer-to-peer applications. DTL combines many features not simultaneously offered by any other transport library including: i) Wide scope of congestion control levels starting from less-than-best-effort to high-priority, ii) Prioritization of traffic relative to other non-DTL traffic, iii) Prioritization of traffic between DTL connections, iv) NAT-friendliness, v) Portability, and vi) Application level implementation. Moreover, DTL has a novel feature, namely, the ability to change the level of aggressiveness of a certain connection at run-time. All the features of the DTL were validated using a controlled environment as well as the Planet Lab testbed.

In this paper, we present SmoothCache, a peer-to-peer live video streaming (P2PLS) system. The novelty of SmoothCache is threefold: i) It is the first P2PLS system that is built to support the relatively-new approach of using HTTP as the transport protocol for live content, ii) The system supports both single and multi-bitrate streaming modes of operation, and iii) In Smoothcache, we make use of recent advances in application-layer dynamic congestion control to manage priorities of transfers according to their urgency. We start by explaining why the HTTP live streaming semantics render many of the existing assumptions used in P2PLS protocols obsolete. Afterwards, we present our design starting with a baseline P2P caching model. We, then, show a number of optimizations related to aspects such as neighborhood management, uploader selection and proactive caching. Finally, we present our evaluation conducted on a real yet instrumented test network. Our results show that we can achieve substantial traffic savings on the source of the stream without major degradation in user experience.

Peer2View is a commercial peer-to-peer live video streaming (P2PLS) system. The novelty of Peer2View is threefold: i) It is the first P2PLS platform to support HTTP as transport protocol for live content, ii) The system supports both single and multi-bitrate streaming modes of operation, and iii) It makes use of an application-layer dynamic congestion control to manage priorities of transfers. Peer2View goals are to achieve substantial savings towards the source of the stream while providing the same quality of user experience of a CDN.

In this paper we present what are, in our experience, the best practices in Peer-To-Peer (P2P) application development and how we combined them in a middleware platform called Mesmerizer. We explain how simulation is an integral part of the development process and not just an assessment tool. We then present our component-based event-driven framework for P2P application development, which can be used to execute multiple instances of the same application in a strictly controlled manner over an emulated network layer for simulation/testing, or a single application in a concurrent environment for deployment purpose. We highlight modeling aspects that are of critical importance for designing and testing P2P applications, e.g. the emulation of Network Address Translation and bandwidth dynamics. We show how our simulator scales when emulating low-level bandwidth characteristics of thousands of concurrent peers while preserving a good degree of accuracy compared to a packet-level simulator.

In this paper we present an exploration of central coordination as a way of managing P2P live streaming overlays. The main point is to show the elements needed to construct a system with that approach. A key element in the feasibility of this approach is a near real-time optimization engine for peer selection. Peer organization in a way that enables high bandwidth utilization plus optimized peer selection based on multiple utility factors make it possible to achieve large source bandwidth savings and provide high quality of user experience. The benefits of our approach are also seen most when NAT constraints come into play.

When evaluating Peer-to-Peer content distribution systems by means of simulation, it is of vital importance to correctly mimic the bandwidth dynamics behaviour of the underlying network. In this paper, we propose a scalable and accurate flow-level network simulation model based on an evolution of the classical progressive filling algorithm which follows the max-min fairness idea. We build on top of the current state of the art by applying an optimization to reduce the cost of each bandwidth allocation/deallocation operation on a node-based directed network model. Unlike other works, our evaluation of the chosen approach focuses both on efficiency and on accuracy. Our experiments show that, in terms of scalability, our bandwidth allocation algorithm outperforms existing directed models when simulating large-scale structured overlay networks. In terms of accuracy we show that allocation dynamics of our proposed solution follow those of the NS-2 packet-level simulator by a small and nearly constant offset for the same scenarios. To the best of our knowledge, this is the first time that an accuracy study has been conducted on an improvement of the classical progressive filling algorithm.

In this paper we modify a fast heuristic solver for the Linear Sum Assignment Problem (LSAP) for use on Graphical Processing Units (GPUs). The motivating scenario is an industrial application for P2P live streaming that is moderated by a central node which is periodically solving LSAP instances for assigning peers to one another. The central node needs to handle LSAP instances involving thousands of peers in as near to real-time as possible. Our findings are generic enough to be applied in other contexts. Our main result is a parallel version of a heuristic algorithm called Deep Greedy Switching (DGS) on GPUs using the CUDA programming language. DGS sacrifices absolute optimality in favor of low computation time and was designed as an alternative to classical LSAP solvers such as the Hungarian and auctioning methods. The contribution of the paper is threefold: First, we present the process of trial and error we went through, in the hope that our experience will be beneficial to adopters of GPU programming for similar problems. Second, we show the modifications needed to parallelize the DGS algorithm. Third, we show the performance gains of our approach compared to both a sequential CPU-based implementation of DGS and a parallel GPU-based implementation of the auctioning algorithm.

This paper deals with solving large instances of the Linear Sum Assignment Problems (LSAPs) under realtime constraints, using Graphical Processing Units (GPUs). The motivating scenario is an industrial application for P2P live streaming that is moderated by a central tracker that is periodically solving LSAP instances to optimize the connectivity of thousands of peers. However, our findings are generic enough to be applied in other contexts. Our main contribution is a parallel version of a heuristic algorithm called Deep Greedy Switching (DGS) on GPUs using the CUDA programming language. DGS sacrifices absolute optimality in favor of a substantial speedup in comparison to classical LSAP solvers like the Hungarian and auctioning methods. We show the modifications needed to parallelize the DGS algorithm and the performance gains of our approach compared to a sequential CPU-based implementation of DGS and a mixed CPU/GPU-based implementation of it.