Loading...
Please wait, while we are loading the content...
Similar Documents
Overlay Multi-hop FEC scheme for Point-to-Point Video Streaming over Peer-to-Peer Networks
| Content Provider | Semantic Scholar |
|---|---|
| Author | Shan, Yufeng Kalyanaraman, Shivkumar |
| Copyright Year | 2003 |
| Abstract | ion of a large end-to-end point-to-point virtual pipe. Fundamental design challenges can be tackled by revisiting fundamental architectural principles (eg: the end-to-end principle). In this paper, we focus on the problem of providing lightweight support at selected intermediate peer forwarding nodes to achieve dramatically increased error resilience on a single peer-based path for point-to-point (not multicast!) videostreaming applications. Unlike traditional error correction that relies on end-to-end ARQ or FEC based upon the end-to-end error characteristics of the network path, our proposed scheme is a flexible scheme that also considers the error characteristics of each peer-based overlay hop. However, our scheme is not a heavyweight hop-by-hop error resilience scheme (like X.25); we segment the end-to-end overlay path into maximal sized "segments" and provide error resilience between the overlay nodes (i.e. peers or hosts) of those segments. Therefore we call our scheme an "overlay multi-hop FEC" (OM-FEC) scheme. Architecturally, this flexible design lies in between the end-to-end and hop-by-hop paradigms, and we argue that it is well suited to peer-based overlay networks. No support is expected from traditional Internet routers. In this paper, we do not focus on overlay path construction and routing problems. We focus on a fixed constructed peer-based overlay path and how to efficiently utilize this path. We will henceforth use the term “overlay path” to mean the constructed path over a peer-to-peer network. High-quality video streaming over the current best-effort Internet is a challenging proposition due to the characteristics of video data such as high bit rate requirement, delay and loss sensitivity. Streaming media distribution has been an intensively studied research topic in the past several years. A large amount of research has been done from all kind of aspects. From the network point of view, companies such as Akamai and Digital Island have deployed Content Delivery Networks (CDNs) by using a network edgebased architecture (edge servers) to achieve load balancing, lower latency and higher throughput. The content is duplicated to the edge servers in order to reduce the round trip time and avoid congestion in the Internet. Simultaneous use of multiple servers [1-2] and multi-paths [3-5] has been proposed in the context of video transmission over Internet. In [1] and [2], the authors propose the use of multiple servers to stream different components of the same content to a single client. Improvement in performance of the video transmission system due to reduction of burst losses is observed in both cases. Video transmission over multiple paths is discussed in [3-5][11]. The authors try to match the characteristic of video data with the path parameters, such as loss rate, delay and capacity, so that the video quality at receiver is maximized. From channel coding perspective, Forward Error Correction (FEC) and Automatic Retransmission reQuest (ARQ) schemes are intensively studied for video transmission. In case of network congestion (loss happens in the video receiver), error recovery scheme and congestion control scheme, such as FEC/ARQ and scheduling, may be deployed to recover the lost packets over the default network path. FEC/ARQ scheme calculates the end-to-end parameters of the default network transmission path and decides what kind of FEC/ARQ scheme should be deployed to combat this network condition. Most recently, peer-to-peer (P2P) architectures and overlay networks are gaining attention. Padmanabhan et al [6] discuss the problem of distributing streaming media content, both live and on demand, to a large number of receivers in a scalable way. They propose a solution called CoopNet, an approach for content distribution that combines aspects of infrastructure-based and peer-to-peer based content distribution, where clients cooperate to distribute content, thereby alleviating the load on the server. CoopNet builds multiple distribution trees spanning the source and all the receivers for its MDC coded media content. Yeo et al in their multicasting streaming paper [7] propose an application level multicast overlay using peering technology and a lightweight gossip mechanism to monitor prevailing network conditions and to improve the tree robustness. Client can dynamically switch to other parents if they experience a poor QoS. In paper [8], Chu etc explore the possibility of video conferencing applications using an overlay multicast architecture. A redesigned Narada [9] protocol is used in [8] to organize the participating nodes into overlay spanning tree for data delivery. The constructed overlay is optimized according to the measurement of available bandwidth and latency among users, and can be modified by the addition of good links and the dropping of poor links. Their results indicate that End System Multicast can meet the stringent bandwidth and latency demands of conferencing applications in heterogeneous and dynamic Internet environments. The main goal of RON [10] is to enable a group of nodes to communicate with each other in the face of problems with the underlying Internet paths connecting them. RON detects problems by aggressively probing and monitoring the paths connecting its nodes. If the underlying Internet path is the best one, that path is used and no other RON node is involved in the forwarding path. If the Internet path is not the best one, the RON will forward the packet by way of other RON nodes. 1.1 Scope and Assumptions Most of the above papers talk about massive video data distribution or video conferencing using application layer multicast based on overlay or peer-to-peer network. When network is congested, the network chooses another better route for packet transmission according to its measurement. In contrast, our objective is to revisit the fundamental problem of efficiently utilize the resources a single overlay path, constructed over peers, i.e. having a number of "hops" between peers. Our approach operates at small time-scales in the data-plane, and can be combined with overlay routing and topology management approaches that operate in the control-plane and in larger time-scales [9][10]. In this sense, error resilience using FEC is complementary (i.e. does not compete) with resilience provided using overlay routing methods. Router Router Overlay node (Bi, Pi, RTTi ) |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.ecse.rpi.edu/Homepages/shivkuma/research/papers/yufeng_jsac.pdf |
| Alternate Webpage(s) | http://www.ecse.rpi.edu/Homepages/shivkuma/research/papers/MMCN04d.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
