Multiplex communications – Data flow congestion prevention or control
Reexamination Certificate
1998-01-30
2001-04-10
Pham, Chi H. (Department: 2731)
Multiplex communications
Data flow congestion prevention or control
C370S235000, C370S465000, C370S256000, C709S235000
Reexamination Certificate
active
06215766
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the transfer of video information over the internet or over intranet systems. More particularly, the present invention is directed to a method and apparatus for providing, in the transmission of a video signal, efficient adding and dropping of video layers received by receivers in a data communication system to reduce data interference associated with realtime video distribution.
DISCUSSION OF RELATED ART
In recent years, there has been a rapid expansion of the internet and intranets, and significant increase in both computer processing power and network bandwidth. Such infrastructure improvements have introduced opportunities for new multimedia applications over networks, such as video conferencing, distance learning, remote presentation, and media on demand, applications that typically involve realtime video distribution. Approaches proposed to handle the realtime aspect of video distribution over networks fall into two categories: (1) the use of a network capable of resource reservation to provide performance guarantees, and (2) the use of adaptive control to adjust multimedia traffic characteristics to meet the network capacity. Both Research Reservation Protocol (RSVP) and Asynchronous Transfer Mode (ATM), which offer network-level reservations, are not yet available for ubiquitous use in real-time video distribution. Even when reservations are available, there are two reasons to apply adaptation techniques: (1) it is difficult to plan particularly accurate reservations so that some adaptation is required to allow tolerance in reservation accuracy; and (2) in view of the cost of resource reservation, it is more efficient to reserve only enough resources to provide the basic required video quality and to then transmit and add on other enhancement layers with best-effort network support and adaptive control.
Due to the heterogeneity of the internet, multicasting of the same video stream to a multiplicity of receivers distributed on a wide range may cause a significant problem. In particular, the video stream may congest certain low-capacity or heavily loaded portions or segments of the network, while some other high-capacity or lightly loaded portions remain under-utilized. However, in an ideal system, each receiver should receive a video stream of a quality commensurate with its processing power and the bandwidth capacity of the path leading to that receiver.
One way of ensuring fair distribution is to multicast different layers of video (which individually contain progressive enhancement layers) using different multicast addresses and to allow the individual receivers to decide which multicast group(s) to subscribe to, i.e. which enhancement layers to receive and add to the signal. An alternative approach is to multicast replicated streams and use destination set grouping (DSG) protocols for improving the fairness of video distribution in a heterogeneous network with only a small bandwidth cost overhead. Layered Video Multicast with Retransmissions (LVMR) addresses the network congestion and heterogeneity problem using layered video coding techniques by allowing each receiver to subscribe to a subset of the video layers as a function of its processing power and network bandwidth availability. LVMR also deploys an error recovery scheme using smart retransmission and adaptive playback point.
There have been two principal approaches employed to address the rate control problem in video multicast: sender-initiated control, and receiver-initiated control. In the sender-initiated approach, the sender multicasts a single video stream whose quality is adjusted based on feedback information from the receivers. The receiver-initiated approach, on the other hand, is usually based on a layered video coding scheme in which the sender multicasts several layers of video (typically a base layer and several enhancement layers) in different multicast groups, and a receiver subscribes to one or more of the layers based on its capabilities. This scheme is “receiver-initiated” in the sense that each receiver determines on its own whether to drop or add a particular enhancement layer.
Comparing LVMR with Receiver-driven Layered Multicast (RLM), both systems deploy layered video multicast schemes but differ in the mechanisms used for adding or dropping a layer. In RLM, a fully distributed approach is advocated in which a receiver, by itself, makes decisions to add or drop an enhancement layer. This decision is enhanced by a “shared learning” process in which information from experiments (i.e. prior attempts by receivers to add or drop enhancement layers) conducted by other receivers is used to improve performance. Shared learning, although providing an improvement to indiscriminate adding and dropping of layers, requires that each receiver maintain a variety of state information parameters that it may or may not require. In addition, the use of multicasting to exchange control information may decrease usable bandwidth on low speed links and lead to lower quality for receivers on these links.
In LVMR, however, a hierarchical approach in the receivers' dynamic rate control schemes is used to allow receivers to maintain minimal state information and decrease control traffic on the multicast session. LVMR also provides more functionality as compared to simple receiver-driven schemes such as RLM. In particular, it allows multiple experiments to be conducted simultaneously, and also helps to drop, in most cases, the correct enhancement layer(s) during periods of congestion, i.e. to drop those enhancement layers that will alleviate the congestion.
The network mechanism employed for video distribution is Internet Protocol (IP)-multicast which, by virtue of its use of the Internet Group Management Protocol (IGMP) and a multicast routing protocol such as Distance Vector Multicast Routing Protocol (DVMRP), Protocol Independent Multicast (PIM) or Core-Based Tree (CBT), sets up a multicast tree spanning all receivers. The multicast tree is defined by the IP-address of the sender and a Class-D IP-address of the group. Receivers can dynamically join and leave a multicast group, thereby resulting in dynamic reconfiguration of the multicast tree. The sender of a message or signal to the multicast group need not know the identity of the receivers, and can therefore continue to send IP packets using the Class-D IP-address representing the group irrespective of any ongoing reconfiguration. On the other hand, it is the responsibility of the receivers to inform the nearest router(s) that those receivers are a part of a specific group.
SUMMARY OF THE INVENTION
The deficiencies of the prior art are addressed herein by a method and apparatus for providing hierarchical data flow rate control of a video signal transmitted through a network to a plurality of receivers. The video signal contains layers of data and, in particular, a base layer and various enhancement layers which are arranged according to a hierarchy such that the base layer is necessary to obtain the video information, a first enhancement layer can be received and decoded along with the base layer, a second enhancement layer can be received and decoded with the base layer and the first enhancement layer, etc. Each receiver receiving a particular video signal may receive different layers of the signal depending, for example, on the particular receiver's capability. Thus, a first receiver may receive the base layer, and a second receiver may receive the base layer and the first enhancement layer.
The inventive method and apparatus segregate the network into domains and subnets, with the receivers divided among the subnets based on criteria such as receiver location. Each domain is assigned an intermediate agent (IA) and each subnet is assigned a subnet agent (SA). Add-layer experiment history data derived from prior attempts made by receivers in the network to add video enhancement layers is obtained. The experiment history data contains information concerning successful and failed add-layer
Ammar Mostafa
Li Xue
Pancha Pramod
Paul Sanjoy
Lucent Technologies - Inc.
Pham Chi H.
Tran Maikhanh
LandOfFree
Hierarchical rate control of receivers in a communication... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Hierarchical rate control of receivers in a communication..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Hierarchical rate control of receivers in a communication... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2526989