B. Dempsey, J. Liebeherr, and A.C. Weaver. "On RetransmissionBased Error Control for Continuous Media Traffic in Packet-Switching Networks". Computer Networks and ISDN Systems, 28(5):719--736, March 1996.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... preceding or following a talk spurt should be transmitted to avoid that perceptually important but lowenergy speech material is not transmitted ( clipping , Sch92, San95, MM98] The lengths of the talk spurts vary dependent on the speaker and on the speech material (Dempsey et al. report in [DLW96] that using a length of 400ms is in accordance with the measurements they conducted) See also [JS00a] for more details on silence detection. ffl encoder decoder (section 2.1.3) ffl delay adaptation (section 2.2.1) ffl real time transport protocol: in addition to conventional transport ....

....parts (seen on the time and or frequency axis) of the original signal. We do not discuss retransmission (ARQ, CB97a, CSS00] as an applicable method, as typically the delay constraints (section 2.2. 1) together with the delay conditions in the network do not allow to apply this method (though [DLW96] reports the usefulness for voice for some (local area) network scenarios) Recent work in the context of video ( Rhe98] has shown that retransmission can be used also for realtime transmission to avoid the effect of error propagation (thus a retransmitted packet might not be usable for direct ....

B. Dempsey, J. Liebeherr, and A.C. Weaver. "On RetransmissionBased Error Control for Continuous Media Traffic in Packet-Switching Networks". Computer Networks and ISDN Systems, 28(5):719--736, March 1996.

....for homogeneous traffic scenarios, for instance, but worked well for bursty video sources. ARQ schemes, however, should not be ruled out for several reasons. One is that for LANs, MANs and even country wide WANs, the propagation delay is small enough to permit retransmissions. In a recent study [DeLW94], the applicability of retransmission based error recovery schemes to packet voice streams was demonstrated for a wide range of end to end delays. Furthermore, in the most general multipoint applications, some participants will be local (vis a vis the transmitter(s) others remote, and it would ....

....the lower resolution information, when using multi resolution techniques (e.g. Ghan92] Performing error concealment on a corrupt I frame could lead to a very significant degradation of several subsequent frames. As for voice, even small amounts of packet loss can be perceptible to the human ear [DeLW94]. We will show in Sections 4 and 5 that retransmissions are both practical and useful for real time media. 3 Overview of Literature As in many light weight transport protocols for high speed networks, ARMS92] and [BRAU93] propose a NACK only, block basis scheme, to improve data throughput. Our ....

[Article contains additional citation context not shown here]

B. J. Dempsey, J. Liebeherr, and A. C. Weaver, "On Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switching Networks," submitted for possible publication to the Computer Networks and ISDN Systems Journal, 1994.

....and multicast backbone (MBone) audio packets [25] With the exception of the last study, neither the setup, nor the type of probe traffic of prior work represented realistic real time streaming scenarios. Among the studies that specifically sent audio video traffic over the Internet (e.g. 4] [8], 22] 23] the majority of experiments involved only a few Internet paths, lasted for a short period of time, and focused on analyzing the features of the proposed scheme rather than the impact of Internet conditions on real time streaming. The methodology used in prior large scale TCP ....

....received by the decoder buffer at the time of its deadline, the video frame is discarded and an underflow event is registered. In order to compensate for one way delay jitter and allow retransmissions, many real time streaming protocols pre buffer video data before starting the decoding process [8], 20] The duration of such pre buffering is called the startup delay of a Note that many video coding schemes require that an additional startup delay be applied to the decoder buffer. This additional delay is called the ideal startup delay and is derived from the encoder buffer model [19] ....

[Article contains additional citation context not shown here]

B. Dempsey, J. Liebeherr, and A. Weaver, "On Retransmissionbased Error Control for Continuous Media Traffic in Packetswitching Networks," Computer Networks and ISDN Systems, vol. 28, no. 5, March 1996.

....audio packets [24] With the exception of the last one, similar observations apply to these studies neither the setup, nor the type of probe traffic represented realistic real time streaming scenarios. In addition, among the studies that specifically sent video traffic over the Internet [5] [10], 21] 22] the majority of experiments involved only a few Internet paths, lasted for a very short period of time, and focused on analyzing the features of the proposed scheme rather than the impact of Internet conditions on real time streaming. In this paper, we present the methodology and ....

....) and reflects the values of the expected (at the beginning of a session) delay jitter and round trip delay during the length of the session. Note that in the context of Internet streaming, it is common to call D budget simply startup delay and to completely ignore the ideal startup delay (e.g. [10]) From this point on, we will use the same convention. In all our experiments, we used D budget equal to 2,700 ms, which was manually selected based on preliminary testing. Consequently, the total startup delay (observed by an 5 We will not elaborate further on the ideal startup delay, except ....

[Article contains additional citation context not shown here]

B. Dempsey, J. Liebeherr, and A. Weaver, "On retransmissionbased error control for continuous media traffic in packetswitching networks," Computer Networks and ISDN Systems, vol. 28, no. 5, March 1996.

....both ACK and NACK based real time streaming applications do not possess a common (i.e. agreedupon) retransmission scheme that is shown to perform well under heterogeneous Internet conditions. In fact, many proposed real time streaming schemes do not specify the choice of an RTO estimator [3], 4] 13] do not deal with real time decoding deadlines of individual frames, ignore the probability of packet reordering [3] 14] 19] and often neglect to set the limit on the maximum number of retransmission requests (where the limit could be based on the lost packet s decoding deadline, ....

....that is shown to perform well under heterogeneous Internet conditions. In fact, many proposed real time streaming schemes do not specify the choice of an RTO estimator [3] 4] 13] do not deal with real time decoding deadlines of individual frames, ignore the probability of packet reordering [3], 14] 19] and often neglect to set the limit on the maximum number of retransmission requests (where the limit could be based on the lost packet s decoding deadline, some fixed integer number, or both) Papadopoulos et al. 14] proposed a real time retransmission scheme in which the receiver ....

B. Dempsey, J. Liebeherr, and A. Weaver, "On retransmission-based error control for continuous media traffic in packet-switching networks," Computer Networks and ISDN Systems, vol. 28, no. 5, March 1996.

....protocol for retransmission based error recovery of continuous media data is presented by Dempsey [37] The slack retransmission request (S ARQ) scheme extends the initial buffering at the receiver (the slack time) to enable receivers to recover some lost messages via retransmission. Figure 5. 1 [36] illustrates the S ARQ method for a unicast voice application. In the figure, the network drops the second voice packet in a talkspurt. Upon detecting the lost packet (as triggered by the reception of an out of order packet) the receiver initiates a retransmission request and recovers the data ....

B. J. Dempsey. Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switched Networks. PhD thesis, Department of Computer Science, University of Virginia, May 1994.

....tradeoff characteristics is AOTP [10] which trades reliability for speed using a receiverbased approach to decide whether or not retransmission is required in order to conform to the user prescribed QoS level. Similar approaches, but centered on sender based transport services, are described in [2, 3, 4, 5, 6]. 3COM has recently designed a set of protocols to support messaging over TCP IP for handheld devices using Palm OS [13] The protocol attempts to save energy by limiting the transmitted data: it transmits 500 bytes during each communication phase under the assumption that the data of interest to ....

B. Dempsey, "Retransmission-based Error Control For Continuous Media Traffic in Packet Switched Networks", Ph.D. Thesis, University of Virginia, 1995

....are further divided into retransmission based techniques and Forward Error Correction (FEC) techniques. Retransmission based techniques are only suited for short and very high speed links where the round trip delay is small enough to support a resend of lost packets as in the S ARQ scheme in [7]. Retransmission is avoided in FEC methods. FEC based methods send redundant information along with the original information (e.g. 2] Prioritybased encoding methods can also be used [1] In media independent FEC based methods, the redundant information can be sent in the form of parity ....

B. Dempsey, J. Liebeherr, and A. Weaver, "On Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switching Networks," Technical Report CS-94-09, Computer Science Department, University of Virginia, February 1994.

....in a frame cannot be recovered in time. While not applicable to cases where the RTT is large, it has been shown that retransmission is feasible in many cases where the RTT is relatively small (e.g. LANs and MANs) especially if a playout buffer is used to increase the time available for recovery [15,26]. Despite its latency drawback, we believe that retransmission based error control is still an attractive solution because of its modest bandwidth and processing costs. For example, unlike FEC, retransmission requires network bandwidth proportional to the loss rate, not the data rate. In addition, ....

Dempsey, B., Liebeherr, J., and Weaver, A., "On Retransmission-based Error Control for Continuous Media Traffic in Packet-switching Networks," Technical Report CS-94-09, University of Virginia, Charlottesville, Virginia, Feb. 1994.

....tradeoff characteristics is AOTP [10] which trades reliability for speed using a receiver based approach to decide whether or not retransmission is required in order to conform to the user prescribed QoS level. Similar approaches, but centered on sender based transport services, are described in [2, 3, 4, 5, 6]. 3COM has recently designed a set of protocols to support messaging over TCP IP for handheld devices using Palm OS [13] The protocol attempts to save energy by limiting the transmitted data: it transmits 500 bytes during each communication phase under the assumption that the data of interest to ....

B. Dempsey, "Retransmission-based Error Control For Continuous Media Traffic in Packet Switched Networks", Ph.D. Thesis, University of Virginia, 1995

....delivery, therefore increasing delay and jitter which are crucial for most multimedia communications (e.g. video transmission) Partial reliability partial reordering is one solution as proposed at [1, 6, 7, 8] emphasizing on a sender based decision as to the need for retransmission. Dempsey in [3] supports the use of retransmission as effective way to handle loss of packets even for continuous media applications. This approach is widely rejected for real time applications, since it increases the RTT delay, which is crucial for real time multimedia, especially for highbandwidth links. The ....

B. Dempsey. "Retransmission-Based Error Control For Continuous Media Traffic In Packet Switched Networks", Ph.D. Thesis, University of Virginia, 1994.

....and another including data recovered by the protocol; display the two streams for subjective evaluation. Page 8 3.5 Related work Little work has been done in designing error control mechanisms for CM applications. In most CM transport protocols, errors are generally not recovered [23] In [11] work has been done in developing an end to end model of packet voice transmission. However, in voice transmission the authors assume that the control time in talkspurts can be extended without any adverse effect on playback. The same assumption is not true for video, where once playback begins ....

Dempsey, B., Liebeherr, J., and Weaver, A., "On Retransmission-based Error Control for Continuous Media Traffic in Packet-switching Networks," Technical Report CS-94-09, Computer Science Department, University of Virginia, Charlottesville, Virginia, Feb. 1994.

....can be addressed by users at their own sites. Furthermore, their impact is expected to decrease as users become familiar with the tools and the tools themselves become more user friendly. 2 However, they would be appropriate in low delay environments, or with relatex end to end delay constraints [9]. 2 as more FEC information is used) on the quality of the redundant information, etc. The problem, then, becomes a constrained optimization problem, namely: given constraints of the rate control mechanisms (i.e. given a total rate at which the source can send) find the combination of main and ....

B. Dempsey, J. Liebeherr, A. Weaver, "On retransmission-based error control for continuous media traffic in packet-switching networks", Computer Networks and ISDN Systems, vol. 28, no. 5, pp. 719 - 736, March 1996.

....the source has perfect information of is shown. mation that the sender believes has been lost; 4 as a result they do not unnecessarily waste bandwidth when there is no packet loss, and they can easily adapt to changes in the loss rates. Thus studies have examined soft ARQ for both unicast [20, 11] and multicast [21, 31, 30] streaming multimedia. One way our work differs from all of these is that we assume there is an overall transmission rate limit, so that a retransmission of one message can come at the expense of the first transmission of another; these other works assume that enough ....

....data lifetime estimate rather than relying on the receiver to estimate the RTT and suppress retransmission 4 Whether this belief is accurate depends on the specifics of the protocol and network. requests. Retransmission schemes for interactive unicast nonlayered multimedia are also studied in [11], which focuses on the viability of impact of the playback delay on the effectiveness of streaming multimedia. Using an end to end voice transmission model and an empirical measurement study, the authors conclude that there are playback delays meeting the delay constraints of interactive audio ....

B. Dempsey, J. Liebeherr, and A. Weaver. On retransmission-based error control for continuous media traffic in packet-switching networks. Computer Networks and ISDN Systems Journal, 28(5):719--36, March 1996. 18

....policies for the same reliability class. On the sending side, the application must inform the protocol what policy should be used for each reliability class when data has been lost. Studies have shown that for some applications, packet loss can often be tolerated. Dempsey, Liebherr and Weaver [DLW94] provide some insight into the usefulness of allowing the application to determine when retransmission is a viable option. Recent work by Marasli, Amer and Conrad [MAC96] shows some analytical studies for retransmission based reliability. 13 6.2 Distributed Robot Simulation The class of ....

Bert Dempsey, Jorg Liebherr, and Alfred Weaver. On retransmission-based error control for continuous media traffic in packet-switching networks. Technical Report CS 94-09, Computer Science Department, University of Virginia, 1994.

....for homogeneous traffic scenarios, for instance, but worked well for bursty video sources. ARQ schemes, however, should not be ruled out for several reasons. One is that for LANs, MANs and even country wide WANs, the propagation delay is small enough to permit retransmissions. In a recent study [25], the applicability of retransmission based error recovery schemes to packet voice streams was demonstrated for a wide range of end to end delays. Furthermore, in the most general multipoint applications, some participants will be local (vis a vis the transmitter(s) others remote, and it would ....

....standards, and the lower resolution information, when using multiresolution standards. Performing error concealment on a corrupt I frame could lead to a very significant degradation of several subsequent frames. As for voice, even small amounts of packet loss can be perceptible to the human ear [25]. We will show in Sections 5.4. and 5.5. that retransmissions are both practical and useful for real time media. 5.3. Overview of Literature As in many light weight transport protocols for high speed networks, 4] and [13] propose a NACK only, block basis scheme, to improve data throughput. Our ....

[Article contains additional citation context not shown here]

B. J. Dempsey, J. Liebeherr, and A. C. Weaver. On Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switching Networks. submitted for possible publication to the Computer Networks and ISDN Systems Journal, 1994. 171

....for homogeneous traffic scenarios, for instance, but worked well for bursty video sources. ARQ schemes, however, should not be ruled out for several reasons. One is that for LANs, MANs and even country wide WANs, the propagation delay is small enough to permit retransmissions. In a recent study [DeLW94], the applicability of retransmission based error recovery schemes to packet voice streams was demonstrated for a wide range of end to end delays. Furthermore, in the most general multipoint applications, some participants will be local (vis a vis the transmitter(s) others remote, and it would ....

....the lower resolution information, when using multi resolution techniques (e.g. Ghan92] Performing error concealment on a corrupt I frame could lead to a very significant degradation of several subsequent frames. As for voice, even small amounts of packet loss can be perceptible to the human ear [DeLW94]. We will show in Sections 4 and 5 that retransmissions are both practical and useful for real time media. 3 Overview of Literature As in many light weight transport protocols for high speed networks, ARMS92] and [BRAU93] propose a NACK only, block basis scheme, to improve data throughput. Our ....

[Article contains additional citation context not shown here]

B. J. Dempsey, J. Liebeherr, and A. C. Weaver, "On Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switching Networks," submitted for possible publication to the Computer Networks and ISDN Systems Journal, 1994.

....in a frame cannot be recovered in time. While not applicable to cases where the RTD is large, it has been shown that retransmission is feasible in many cases where the RTD is relatively small (e.g. LANs and MANs) especially if a playout buffer is used to increase the time available for recovery [9,19]. Despite its latency drawback, we believe that retransmission based error control is still an attractive solution because of its modest bandwidth and processing costs. For example, unlike FEC, retransmission requires network bandwidth proportional to the loss rate, not the data rate. In addition, ....

Dempsey, B., Liebeherr, J., and Weaver, A., "On Retransmission -based Error Control for Continuous Media Traffic in Packet-switching Networks," Technical Report CS-94-09, University of Virginia, Charlottesville, Virginia, Feb. 1994.

No context found.

B. Dempsey. Retransmission-Based Error Control for Continuous Media Tra#c in Packet-Switched Networks. PhD thesis, CS-94-23, Computer Science, Universityof Virginia, May 1994.

....However, the retransmission approaches employed are largely inappropriate for multimedia applications since the temporal constraints of continuous media are not considered. A deadline driven protocol for retransmission based error recovery of continuous media data is presented by Dempsey [37]. The slack retransmission request (S ARQ) scheme extends the initial buffering at the receiver (the slack time) to enable receivers to recover some lost messages via retransmission. Figure 5.1 [36] illustrates the S ARQ method for a unicast voice application. In the figure, the network drops the ....

B. J. Dempsey, J. Liebeherr, and A. C. Weaver. On Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switching Networks. Computer Networks and ISDN Systems, 28(5):719--736, March 1996. Bibliography149

....packet voice, an important and demanding application domain. We obtain delay measurements for three specific paths in the network, and these measurements are used to examine the voice protocol parameters for this environment, including a novel retransmission based error control proposed in [9]. We conclude that voice distribution is generally feasible in our campus wide network, though we also find evidence of problematic delay behavior. Key Words: Continuous Media, Packet Voice, Retransmission, Error Control. 1 Introduction Recent technology trends have resulted in increasingly ....

....show that voice distribution is generally feasible. We explore the design trade offs for a packet voice protocol in this environment and quantify the achievable transmission quality using our data. In our protocol evaluation we consider a novel retransmission based error control scheme proposed in [9, 10]. Our study indicates that, contrary to current wisdom, retransmission can provide significant error coverage for packet voice while respecting the delay constraints of the voice stream. The remainder of this paper is structured as follows. Section 2 outlines the requirements of packet voice and ....

[Article contains additional citation context not shown here]

B. Dempsey. Retransmission-Based Error Control for Continuous Media Traffic in PacketSwitched Networks. PhD thesis, (unpublished draft), Computer Science, University of Virginia, 1994.

....deadline of the lost packets, and outstanding retransmission requests are ignored once this deadline passes. However, the time constraints introduced by playback deadlines are generally not problematic since packet buffers at each multicast receiver can provide relatively long holding times [10]. Note that receiver buffering to handle jitter in wide area distributions is typically in the range of a few hundred milliseconds. The retransmission scheme is, in effect, based on the dynamic construction of a recovery graph in the wide area topology. When a loss occurs, receivers contact each ....

B. Dempsey. Retransmission-Based Error Control for Continuous Media Traffic in Packet-Switched Networks. PhD thesis, Department of Computer Science, University of Virginia, May 1994.

No context found.

B. Dempsey, J. Liebeherr, and A. Weaver, "On retransmission-based error control for continuous media traffic in packet-switching networks," Computer Networks and ISDN Syst., vol. 28, no. 5, pp. 719--736, Mar. 1996.

No context found.

B. Dempsey et. al, "On Retransmission Based Error Control for Continuous Media Traffic in Packet Switching Networks" Technical Report, CS-94-09, University of Virginia, 1994

No context found.

Bert J. Dempsey. Retransmission-Based Error Control For Continuous Media Traffic In Packet-Swithced Networks. PhD thesis, University of Virginia, 1994.

First 50 documents

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