International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol. 1 Issue 6, August - 2012 Study and analysis on Optimal scheduling Data-Driven Peer-to-Peer Streaming 1 2 B.Kundan D.Jamuna currently emerging throughout society, enabled by Abstract: Internet technologies in general. We propose a new architecture for on-demand media streaming centered around the peer-to-peer (P2P) The basic idea of data-driven streaming paradigm. The key idea of the architecture is that peers protocol is very simple and similar to that of Bit- share some of their resources with the system. As peers Torrent. The protocol contains two steps. In the first contribute resources to the system, the overall system step, each node independently selects its neighbors so capacity increases and more clients can be served.. In as to form an unstructured overlay network, called the these applications, each node independently selects gossip-style overlay construction or membership some other nodes as its neighbors (i.e., gossip style management. The second step is named block overlay construction) and exchanges streaming data scheduling: The live media content is divided into with the neighbors (i.e., data scheduling). To improve blocks (or segments or packets), and every node the performance of such protocol, many existing works announces what blocks it has to its neighbors. Then focus on the gossip-style overlay construction issue. each node explicitly requests the blocks of interest from However, few of them concentrate on optimizing the its neighbors according to their announcement. In this streaming data scheduling to maximize the throughput project, we present our analytical model and of a constructed overlay. In this paper, we analytically corresponding solutions to tackle the block scheduling study the scheduling problem in data-driven streaming problem in data-driven protocol. We first model this system and model it as a classical min-cost network scheduling problem as a classical min-cost network flow problem. We then propose both the global optimal flow problem and propose a global optimal solution in scheduling scheme and distributed heuristic algorithm order to find out the ideal throughput improvement in to optimize the system throughput. Furthermore, we theory. Since this solution is centralized and requires introduce layered video coding into data-driven global knowledge, based on its basic idea, we then protocol and extend our algorithm to deal with the end- propose a heuristic algorithm that is fully distributed host heterogeneity. and asynchronous with only local information exchange. Furthermore, we employ layered video Keywords: Peer-to-peer systems, optimal scheduling, coding to encode the video into multiple rates and Distributed hash tables extend our algorithm to improve the satisfaction of the users with heterogeneous access bandwidth. `1.INTRODUCTION: In Existing system we using two approaches A peer-to-peer, commonly abbreviated to P2P, for online media streaming called unicast based distributed network architecture is composed of approaches and multicast based approach. There are participants that make a portion of their resources (such two approaches that use unicast for on-demand as processing power, disk storage or network streaming centralized content distribution networks bandwidth) directly available to other network (CDN). In the centralized approach we have participants, without the need for central coordination maintaining the powerful server, this approach is easy instances (such as servers or stable hosts). Peers are to deploy and manage. But the scalability process leads both suppliers and consumers of resources, in contrast to the failure in the network. There are two other to the traditional client-server model where only servers critical, but less obvious, disadvantages of the supply, and clients consume. Peer-to-peer was centralized approach: high cost and load on the popularized by file sharing systems like Napster. Peer- backbone network. On the negative side, this approach to-peer file sharing networks have inspired new requires deploying and managing proxies at many structures and philosophies in other areas of human locations. While deploying proxies increases the overall interaction. In such social contexts, peer-to-peer as a system capacity, it multiplies the cost. The capacity is meme refers to the egalitarian social networking that is still limited by the aggregate resources of the proxies. In multicast they have following two techniques www.ijert.org 1 International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol. 1 Issue 6, August - 2012 network level and application level in both they constructing the network in the tree and client act as a leaf. Centralized approach: high cost and load on the backbone network. On the negative side, this approach requires deploying and managing proxies at many locations. While deploying proxies increases the overall system capacity, it multiplies the cost. The capacity is still limited by the aggregate resources of the proxies. In multicast they have following two techniques network level and application level in both they constructing the network in the tree and client act as a leaf. The main purpose of the proposed system the scheduling problem in data-driven streaming system and model it as a classical min-cost network flow problem. We then propose both the global optimal scheduling scheme and distributed heuristic algorithm to optimize the system throughput. Furthermore, we Fig: 2.1 Peer-to-peer systems Architecture introduce layered video coding into data-driven protocol and extend our algorithm to deal with the end- host heterogeneity. We present our analytical model Peer-to-peer systems often implement an and corresponding solutions to tackle the block Application Layer overlay network on top of the native scheduling problem in data-driven protocol. We first or physical network topology. Such overlays are used model this scheduling problem as a classical min-cost for indexing and peer discovery. Content is typically network flow problem and propose a global optimal exchanged directly over the underlying Internet solution in order to find out the ideal throughput Protocol (IP) network. Anonymous peer-to-peer improvement in theory. Since this solution is systems are an exception, and implement extra routing centralized and requires global knowledge, based on its layers to obscure the identity of the source or basic idea, we then propose a heuristic algorithm that is destination of queries. In structured peer-to-peer fully distributed and asynchronous with only local networks, connections in the overlay are fixed. They information exchange. This scheduling problem as a typically use distributed hash table-based (DHT) classical min-cost network flow problem and propose a indexing, such as in the Chord system (MIT). global optimal solution in order to find out the ideal Unstructured peer-to-peer networks do not provide any throughput improvement in theory. Since this solution algorithm for organization or optimization of network is centralized and requires global knowledge, based on connections. In particular, three models of unstructured its basic idea, we then propose a heuristic algorithm that architecture are defined. In pure peer-to-peer systems is fully distributed and asynchronous with only local the entire network consists solely of equipotent peers. information exchange. There is only one routing layer, as there are no preferred nodes with any special infrastructure function. Hybrid peer-to-peer systems allow such infrastructure 2. Peer-to-Peer systems: nodes to exist, often called supernodes. In centralized peer-to-peer systems, a central server is used for A peer-to-peer, commonly abbreviated to P2P, indexing functions and to bootstrap the entire system. distributed network architecture is composed of Although this has similarities with a structured participants that make a portion of their resources (such architecture, the connections between peers are not as processing power, disk storage or network determined by any algorithm. The first prominent and bandwidth) directly available to other network popular peer-to-peer file sharing system, Napster, was participants, without the need for central coordination an example of the centralized model. Gnutella and instances (such as servers or stable hosts).Peer-to-peer Freenet, on the other hand, are examples of the networks are typically formed dynamically by ad-hoc decentralized model. Kazaa is an example of the hybrid additions of nodes. In an 'ad-hoc' network, the removal model. P2P networks are typically used for connecting of nodes has no significant impact on the network. The nodes via largely ad hoc connections. Sharing content distributed architecture of an application in a peer-to- files (see file sharing) containing audio, video, data or peer system provides enhanced scalability and service anything in digital format is very common, and real robustness. Figure shows architecture of peer-to-peers. www.ijert.org 2 International Journal of Engineering Research & Technology (IJERT) ISSN: 2278-0181 Vol. 1 Issue 6, August - 2012 time data, such as telephony traffic, is also passed using it, there is no guarantee that flooding will find a peer P2P technology. that has the desired data. Flooding also causes a high amount of signaling traffic in the network and hence A pure P2P network does not have the notion such networks typically have very poor search