Peer-to-Group Media Broadcast

Overview

Requirements

Download

FAQs

Forum

Reviews

What is Peer-to-Group Media Broadcast?

Peer-to-Group Media Broadcast provides an application-layer virtual tunnel for delivering continuous data to thousands of receivers at the same time. For example, by combining it with a live streaming system, thousands of employees can watch a company's executive speech on their desktop PC screens in almost real time, even when they are in remote branch offices.

This technology does not require any special network infrastructure such as multicast-enabled routers, but it is a complete end-point solution, as well as a server and client model, which realizes low-cost installation, subscription, and support and easy deployment. In addition, network traffic is well-distributed and semi-optimized over the receiver nodes, which also work as relaying points.

How does it work?

• Peer-to-group unicast: The virtual tunnel has an entrance and one or more exits. A station is located at the entrance and it continuously receives data segments (or packets) from a streaming source server such as Windows Media Encoder. At the same time, the station chooses one of the receivers, which are located at each exit, for each packet and forwards the packet to not all of them but to only the chosen one. We call this a Peer-to-Group (P2G) Unicast, because the station continuously sends a single set of data packets with no duplicates towards a group of receivers by using unicast methods.
• Grid delivery: When a receiver gets a data packet from the station, then it forwards the segment to its peer receivers in the group. Seen from a receiver, partial data comes directly from the station and the rest of the data from its peers. It is a multiple-source download and is thus called a Grid Delivery.
• Traffic optimization: Each receiver also works as a forwarder of data packets by using its uplink. This helps the station reduce its downlink to a single data rate. Each receiver always gets data packets in an original data rate, which can reproduce the original data as long as no packet is lost in the real network. It also sends data packets almost in the original data rate for peers, as delivery paths change for each data segment. This "one-in/one-out" traffic distribution helps the delivery system stay well-balanced, especially when an uplink bandwidth is lower than a downlink one, whereas a fixed delivery path approach in the conventional application-layer multicast systems requires much higher bandwidth in uplinks than in downlinks because it's based on "one-in/multi-out" traffic.
• Independence from outer streaming system: At each exit of the tunnel, a streaming client such as Windows Media Player receives and plays the reproduced original streaming data. Thus, the data comes not directly from the streaming source server but through the virtual tunnel. However, the streaming client acts as if it receives the data directly from the streaming server, so no modification is necessary for the outer streaming system.
• Scalability: A scalability issue arises when the number of receivers in a group becomes large, for example, a thousand, because the receivers are supposed to forward packets to all their peers in the above scenario. In order to solve this situation, this technology introduces a multiple-step peer-to-group delivery chain by using network hierarchy: When a receiver gets a data packet from the station, then it acts as the second-level station for the same and lower level of groups. It forwards the packet to each of the groups in a P2G Unicast way, and the same steps continue until all receivers get the data packet. Even in the hierarchical approach, all receivers still work in an equal basis, that is, there are no representatives fixed in the hierarchy; no fixed roll is assigned to any receiver. They are upstream in some delivery paths, but may be downstream in other paths. Traffic is still well-distributed in that any level of group always receives packets in a single data rate with no duplicates, independent of the number of receivers included in each group, because of the varying path. This is also the same for uplink usages from each group.
• Stability: Routing is dynamically determined by not only the station but also the receivers. The station determines the first destination receiver and then the receiver determines the next-level destination peers. It reduces resource usage for computing paths in the station, and it makes the computing load shared in the peers. This dynamic and decentralized routing also helps the system select stable paths detouring around low-performance receivers, which makes the whole system stable even when one or more receivers is not reliable.