Yavatkar, Rajendra e Manoj, Leelanivas (1993) "Optimistic Strategies for Large-Scale Dissemination of Multimedia Information", First of the ACM Int. Conf. on Multimedia `93.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Moreover the arrivals of receiver reports may well be synchronized, raising the source s burden to an unbearable level [29] A number of schemes have been devised to alleviate this problem. Probabilistic approaches require each receiver to observe some random delay before sending a receiver report [30]. 33 This makes it possible to assign unequal importance to receivers by changing some weights in the random process, but the delays before a congestion indication is taken into account by the source may be prohibitive. Another approach is for the source to perform progressive, topological ....

....most cases this 34 translates into selecting the lowest common rate between all indicated values, or a rate low enough that a vast majority of receivers will not experience congestion. The corollary is reduced quality for all receivers, whether they actually experience congestion or not [34] 29] [30]. When the feedback scheme is devoted to error protection, as in our mode selection approach, then the source has to take into account the worse error conditions encountered by the different receivers. The corollary in this case is suboptimal rate distorsion ratio, since most forward error ....

R. Yavatkar and L. Manoj. Optimistic strategies for large-scale dissemination of multimedia information. In ACM Multimedia'93, Anaheim, CA, August 1993.

....RTP timestamp, packet count, and octet count. The reception component contains performance reports for up to 31 senders. Each report includes the source s identifier, cumulative packets received, cumulative packets expected, inter arrival jitter, and the timestamp of the last report. Yavatkar [104] investigates rate adjustment, error control, and feedback strategies for large scale distribution of time sensitive data. The following rate adjustment strategies are evaluated: 1) majority policy the rate is set to the majority of the receivers, 2) universal policy the rate is set to the ....

R. Yavatkar and Leelanivas Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. ACM Multimedia 1993, pages 1--8, August 1993.

....scarce, while noncongested paths may transmit both the high and low priority cells, achieving high quality video on paths where bandwidth is plentiful. Other solutions to the problems of feedback implosion and available bandwidth variation have been proposed. These include probabilistic feedback [7, 8] and hierarchical acknowledgements [9] Probabilistic feedback ameliorates feedback implosion by having each destination return a feedback signal with a probability less than or equal to one. In the context of multicast video, however, probabilistic schemes do not allow for the rapid detection of ....

R. Yavatkar and L. Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. Proc. of ACM Multimedia, pages 13--20, August 1993.

.... packet level admission control and buffer reservation in the network( 5] and the use of a tree like hierarchy of multicast groups to control the sender rate ( 6] Real time applications like video have received particular attention in the study of multicast rate control in wide area networks ([7, 8, 9]) 7] describes mechanisms for soliciting receiver response in a scalable manner and adjusting video transmission rate based on it. The issue of how to satisfy receivers with diverse requirements (which we consider here) was left open in [7] 8] has proposed algorithms for handling ....

....rate control in wide area networks ( 7, 8, 9] 7] describes mechanisms for soliciting receiver response in a scalable manner and adjusting video transmission rate based on it. The issue of how to satisfy receivers with diverse requirements (which we consider here) was left open in [7] [8]) has proposed algorithms for handling heterogeneous receivers throttling the sender according to the needs of the majority of receivers, according to the bottleneck receiver, etc. However. similar approaches for non real time loss sensitive approaches are largely unexplored. Multilevel coding ....

R. Yavatkar, L. Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. In Proceedings of ACM Multimedia, 1993.

....is not true for video, where once playback begins the frames must be displayed at fixed intervals. We believe that variable frame rate video is annoying to the users and therefore we do not advocate this option. Other researchers have relied on FEC for error control for multimedia applications [27]. However, as discussed earlier, we consider retransmission based error control as preferable to FEC because of its lower cost. 4. IMPLOSION IN MULTICAST ERROR CONTROL In this section we first describe implosion and how it occurs in multicast error control. Then, using a simple analytic model we ....

Yavatkar, R., Manoj, L., "Optimistic Strategies for Large Scale Dissemination of Multimedia Information," Proceedings of ACM Multimedia `93, Anaheim, CA, pp. 1-8, Aug. 1993.

....for the sake of brevity. 6.3. Hop by Hop Approach One of the main drawbacks of end to end schemes using a feedback mechanism is the problem of feedback implosion at the source, under a large multicast scenario. This has led many researchers to reject closed loop schemes altogether (e.g. [81]) Others have used a combination of probabilistic querying (where receivers send feedback information with some probability) randomly delayed responses (where receivers randomly delay their feedback control messages) and expanding scoped search (where the transmitter gradually increases the ....

R. Yavatkar and L. Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. In ACM Multimedia '93, pages 12--20, Anaheim, CA, August 1993. 178

....in [7] see [12] for details) 3 Hop by Hop Approach One of the main drawbacks of end to end schemes using a feedback mechanism is the problem of feedback implosion at the source, under a large multicast scenario. This has led many researchers to reject closed loop schemes altogether (e.g. [15]) Others have used a combination of probabilistic querying (where receivers send feedback information with some probability) randomly delayed responses (where receivers randomly delay their feedback control messages) and expanding scoped search (where the transmitter gradually increases the ....

R. Yavatkar and L. Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. In ACM Multimedia '93, pages 12--20, Anaheim, CA, August 1993.

.... of packet level admission control and buffer reservation in the network( 3] and the use of a tree like hierarchy of multicast groups to control the sender rate ( 4] Real time applications like video have received particular attention in the study of multicast rate control in wide area networks ([5, 6, 7]) 5] describes mechanisms for soliciting receiver response in a scalable manner and adjusting video transmission rate based on it. The issue of how to satisfy receivers with diverse requirements (which we consider here) was left open in [5] 6] has proposed algorithms for handling ....

....rate control in wide area networks ( 5, 6, 7] 5] describes mechanisms for soliciting receiver response in a scalable manner and adjusting video transmission rate based on it. The issue of how to satisfy receivers with diverse requirements (which we consider here) was left open in [5] [6]) has proposed algorithms for handling heterogeneous receivers throttling the sender according to the needs of the majority of receivers, according to the bottleneck receiver, etc. However. similar approaches for non real time loss sensitive approaches are largely unexplored. Multilevel coding ....

R. Yavatkar, L. Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. In Proceedings of ACM Multimedia, 1993.

....growth in demand for wide area bandwidth [26] Congestion based losses can be reduced through dynamic rate control of the multicast source. By sending to the source feedback messages indicating network performance, receivers adapt the data rate of the source to the prevailing network capacity [6, 17, 34]. The data rate is adjusted by changing the compression ratio or selectively dropping information, e.g. skipping video frames. While lower data rates reduce the quality of the audiovisual stream, the technique protects the network from being swamped by a single transmitter and provides good ....

....swamped by a single transmitter and provides good quality relative to the current network capacity. To provide a rate control algorithm that avoids the implosion problem for large receiver sets, feedback messages to the source must be time multiplexed in some manner, e.g. probabilistic polling [6, 34]. Time multiplexed responses imply a long duration feedback loop, which yields imperfect knowledge of the network performance at all receivers and limits the rate at which the source can adapt to changing network conditions. Note that rate controlling the source is an orthogonal issue to that of ....

[Article contains additional citation context not shown here]

R. Yavatkar and Leelanivas Manoj. Optimistic Strategies for Large-Scale Dissemination of Multimedia Information. ACM Multimedia 1993, pages 1--8, August 1993.

....problems in getting feedback from the receivers. It is important to get timely notification of congestion, but if the congestion is close to the source then all receivers will detect the congestion and will send a notification to the source generating an implosion of messages at the source [133,134]. For this reason, a control mechanism such as that described in Chapter 5 is not possible. Instead, we require a mechanism for soliciting feedback information in a scalable way from the receivers. Given this feedback information, it is important to relate the state to the entire group of ....

.... create a so called implosion problem, in which a potentially large amount of feedback information is sent almost synchronously from the receivers back to the source [133] Solutions to this problem have included probabilistic querying, randomly delayed responses, and an expanding scoped search [134]. In a probabilistic scheme, a receiver responds to the request from a source with a given probability. Typically, the request is sent again by the source after some timeout interval if no reply has been received. The probabilistic scheme is easy to implement, but it has limitations. For example, ....

R. Yavatkar & L. Manoj, "Optimistic strategies for large-scale dissemination of multimedia information," Proc. ACM Multimedia '93 (Aug. 1993).

.... (4) is to send redundant information that allows lost data to be reconstructed provided a subset of the original information is sent, a technique called forward error correction (FEC) 8,72] Yavatkar and Manoj study two FEC strategies (XOR and replication) in a mutlicasting simulation study [85]. The Priority Encoded Transmission (PET) project [2] implements FEC in a particularly novel way. Each frame in a group of frames is assigned a numerical priority in the range (0. 1] with smaller values yielding higher priorities. The group is then encoded into packets that are sent to the ....

....network is delayed. UDP resends uses strategy (3) to reduce the rate at which frames are generated at the source. HeiTP [19] uses media specific protocols to compensate for network congestion in response to feedback messages from the destination. And in the simulation study by Yavatkar and Manoj [85], several strategies for flow control based on destination feedback messages are investigated. TCP IP is the only system I know of that uses window based flow control. The Xphone system [20] which sends audio and motion JPEG data using TCP IP, also uses window based flow control by extension. ....

[Article contains additional citation context not shown here]

R. Yavatkar, L. Manoj, Optimistic Strategies for Large Scale Dissemination of Multimedia Information, Proceedings of ACM Multimedia 93, (Anaheim, CA, August 1-6, 1993). Association for Computing Machinery Press, New York, 1993, pp. 13-20

....problems in getting feedback from the receivers. It is important to get timely notification of congestion, but if the congestion is close to the source then all receivers will detect the congestion and will send a notification to the source generating an implosion of messages at the source [8, 31]. To prevent this, we require a mechanism for soliciting feedback information in a scalable way from the receivers. Given this feedback information, it is important to relate the state to the entire group of receivers within the context of the application. If only a single receiver is suffering ....

.... might create a so called implosion problem, in which a potentially large amount of feedback information is sent almost synchronously from the receivers back to the source [8] Solutions to this problem have included probabilistic querying, randomly delayed responses, and an expanding scoped search [31]. In a probabilistic scheme, a receiver responds to the request from a source with a given probability. Typically, the request is sent again by the source after some timeout interval if no reply has been received. The probabilistic scheme is easy to implement, but it has limitations. For example, ....

R. Yavatkar, L. Manoj, "Optimistic strategies for large-scale dissemination of multimedia information", Proc. ACM Multimedia '93, Anaheim, CA, pp. 1-8, Aug. 1993.

....information dissemination: High speed network applications such as remote visualization and on demand multimedia services [1] involve disseminating delay sensitive multimedia information from a single source to multiple receivers. Multicast transport services designed to support such applications [2] must incorporate a mechanism for keeping track of the receiver group membership. 2. Reliable multicasting: Reliable multicast has been proposed as a basic mechanism for communication in distributed systems [3,4] Here, multicast messages must be delivered reliably from one or more sources to a ....

....only of those members in V m , G , messages from m to G are targeted at only those members. Using a view identification method, members in a group accept a message only if they belong to the view being addressed. For instance, with the availability of group views, a multicast transport protocol [6,2] may proceed as follows. A source attempts to transfer multicast messages reliably to the set of members in its current view. It either succeeds in transferring the messages to each member in the view or it obtains a new view. If the view has changed, it repeats the same process with the new view. ....

R. Yavatkar and L. Manoj, "Optimistic Strategies for Large-Scale Dissemination of Multimedia Information," Proc. ACM Multimedia '93, 1993.

....and implemented a destination set grouping protocol for window flow controlled connections using a binary feedback flow control method. The unfairness problem has also been identified by many researchers but we are not aware of any in depth proposal and investigation for a solution. The work in [14] hints (in a footnote) at a solution involving the use of separate video stream with lower quality of service (QoS) for low capacity receivers. The protocol in [11] also makes a passing reference to the establishment of a low bandwidth multicast group for the worst case receivers. In both cases ....

R. Yavatkar and L. Manoj, "Optimistic strategies for large-scale dissemination of multimedia information," in Proceedings of the 1st ACM Multimedia Conference, pp. 13--20, ACM, 1993.

....a poor idea to tailor the multicast transmission rate to the worst case users. One natural approach for handling the feedback implosion problem is the use of random sampling, whereby only a small subset of receivers transmits feedback to the source at any point in time. Yavatkar and Manoj [YM 93] define a Quasi reliable Multicast Transport Protocol (QMTP) which performs rate based flow control with selective receiver feedback employing this technique. Unfortunately, their solution does not completely solve the implosion problem, moreover the limited feedback received may not be an ....

....unscalable style of acknowledgment. One widely used technique for improving scalability in multicast is random sampling, whereby each packet arrival triggers an acknowledgment from only a small fraction of the receivers. While this technique has been used effectively for small sets of receivers [YM 93] it does not appear to scale well. The technique which we employ instead is acknowledgment aggregation, using a scheme analogous to the aggregation of control messages in our policy for multicast bandwidth allocation. In this scheme, each receiver reports an acknowledgment value for each source ....

R. Yavatkar and L. Manoj. Optimistic Strategies for Large-scale Dissemination of Multimedia Information. In Proceedings of ACM Multimedia '93, August 1993.

....possible random losses [21] which would be mistakenly considered as indicative of network congestion. 3 The source uses the Real Time Control Protocol [22] implemented in IVS to determine the number of receivers in a multicast session. receiver sends its QoS measure only with some probability [28]. This probability is chosen so that the load of the feedback traffic is approximately constant. One way to do this is to set the probability to be inversely proportional to the total number of receivers. Each receiver sends its measured average loss rate back to the source in the QoS field in the ....

R. Yavatkar, L. Manoj, "Optimistic strategies for largescale dissemination of multimedia information", Proc. ACM Multimedia '93, Anaheim, CA, pp. 1-8, Aug. 1993.

....its queuing delay back to the source. A controller uses this information to adjust the output rate of the source coder, allowing the source to react to queue buildup before packet loss occurs. These source based rate adaptation schemes are poorly matched to multicast environments. QMTP [34] and the IVS congestion control scheme [1] adapt by soliciting feedback from the receivers in a scalable fashion, but these schemes do not cope well with bandwidth heterogeneity. Either lowcapacity regions of the distribution are overwhelmed or highcapacity regions are underutilized. Shacham ....

YAVATKAR, R., AND MANOJ, L. Optimistic strategies for large-scale dissemination of multimedia information. In Proceedings of ACM Multimedia '93 (Aug. 1993), ACM, pp. 1--8.

No context found.

Yavatkar, Rajendra e Manoj, Leelanivas (1993) "Optimistic Strategies for Large-Scale Dissemination of Multimedia Information", First of the ACM Int. Conf. on Multimedia `93.

No context found.

R. Yavatkar and L. Manoj, "Optimistic Strategies for Large-Scale Dissemination of Multimedia Information," Proceedings of ACM Multimedia 1993.

No context found.

R. Yavatkar and L. Manoj. Optimistic strategies for large-scale dissemination of multimedia information. In ACM Multimedia'93, Anaheim, CA, August 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC