Variable-bit-rate compressed video can exhibit significant, multiple-time-scale bit rate variability. In this paper we consider the transmission of stored video from a server to a client across a network, and explore how the client buffer space can be used most effectively toward reducing the variability of the transmitted bit rate. Two basic results are presented. First, we show how to achieve the greatest possible reduction in rate variability when sending stored video to a client with given buffer size. We formally establish the optimality of our approach and illustrate its performance over a set of long MPEG-1 encoded video traces. Second, we evaluate the impact of optimal smoothing on the network resources needed for video transport, under two network service models: Deterministic Guaranteed service [1, 31] and Renegotiated CBR (RCBR) service [9]. Under both models the impact of optimal smoothing is dramatic. 1 Introduction A broad range of applications is enabled by the capac...

Proxies are emerging as an important way to reduce user-perceived latency and network resource requirements in the Internet. While relaying traffic between servers and clients, a proxy can cache resources in the hope of satisfying future client requests directly at the proxy. However, existing techniques for caching text and images are not appropriate for the rapidly growing number of continuous media streams. In addition, high latency and loss rates in the Internet make it difficult to stream audio and video without introducing a large playback delay. To address these problems, we propose that, instead of caching entire audio or video streams (which may be quite large), the proxy should store a prefix consisting of the initial frames of each clip. Upon receiving a request for the stream, the proxy immediately initiates transmission to the client, while simultaneously requesting the remaining frames from the server. In addition to hiding the latency between the server and the proxy, st...

In networks that support Quality of Service (QoS), an admission control algorithm determines whether or not a new traffic flow can be admitted to the network such that all users will receive their required performance. Such an algorithm is a key component of future multi-service networks as it determines the extent to which network resources are utilized and whether the promised QoS parameters are actually delivered. Our goals in this paper are threefold. First, we describe and classify a broad set of proposed admission control algorithms. Second, we evaluate the accuracy of these algorithms via experiments using both on-off sources and long traces of compressed video; we compare the admissible regions and QoS parameters predicted by our implementations of the algorithms with those obtained from trace-driven simulations. Finally, we identify the key aspects of an admission control algorithm necessary for achieving a high degree of accuracy and hence a high statistical multiplexing gain...

Measurement-based admission control (MBAC) is an attractive mechanism to concurrently offer quality of service (QoS) to users, without requiring a priori traffic specification and on-line policing. However, several aspects of such a system need to be clearly understood in order to devise robust MBAC schemes, i.e., schemes that can match a given QoS target despite the inherent measurement uncertainty, and without the tuning of external system parameters. We study the impact of measurement uncertainty, flow arrival, departure dynamics, and of estimation memory on the performance of a generic MBAC system in a common analytical framework. We show that a certainty equivalence assumption, i.e., assuming that the measured parameters are the real ones, can grossly compromise the target performance of the system. We quantify the improvement in performance as a function of the length of the estimation window and an adjustment of the target QoS. We demonstrate the existence of a critical time scale over which the impact of admissin decisions persists. Our results yield new insights into the performance of MBAC schemes, and represent quantitative and qualitative guidelines for the design of robust schemes.

Abstract 1 Retransmission-based error recovery has been, in general, considered inappropriate for continuous media (CM) applications, because of its latency. However, retransmission is still attractive because it requires minimal network bandwidth, (less than either peak-bandwidth allocation or FEC for bursty streams) and processing cost. Despite its latency, we believe that retransmission can be used successfully in many cases, especially if playout buffering is employed. We have designed and implemented a retransmissionbased error control scheme for CM applications, which aims to provide the best possible reliability at a minimal cost, without violating the application’s timing constraints. We have enhanced selective-repeat retransmission with: (1) playout buffering to increase the time available for recovery, (2) gap-based rather than timer-based loss detection to minimize loss detection latency, (3) implicit expiration of sender retransmission buffers to eliminate acknowledgments, (4) conditional retransmission requests to avoid triggering late, unnecessary retransmissions, and (5) data integrity information delivery to the application to aid in concealment. Experimental results show that the mechanism significantly reduces observed loss without violating the application’s delay constraints. 1.

In this paper, we define a class of generalized Guaranteed Rate (GR) scheduling algorithms that includes algorithms which allocate variable rate to packets of a flow. We define work-conserving generalized Virtual Clock, Packet-by-Packet Generalized Processor Sharing, and Self-Clocked Fair Queuing scheduling algorithms that can allocate variable rate to the packets of a flow. We also define scheduling algorithms suitable for servers where packet fragmentation may occur. We demonstrate that if a class of rate controllers is employed for a flow in conjunction with any scheduling algorithm in GR, then the resulting non-work-conserving algorithm also belongs to GR. This leads to the definition of several non-work-conserving algorithms. We then present a method for deriving the delay guarantee of a network of servers when: 1) different rates are allocated to packets of a flow at different servers along the path and the bottleneck server for each packet may be different, and 2) packet fragmentation and/or reassembly may occur. This delay guarantee enables a network to provide various service guarantees to flows conforming to any specification. We illustrate this by utilizing delay guarantee to derive delay bounds for flows conforming to Leaky Bucket, Exponentially Bounded Burstiness, and Flow Specification. Our method for determining these bounds is valid in internetworks and leads to tighter results.

In this paper, we examine the transport and storage of video compressed with a variable bit rate (VBR). We focus primarily on networked video, although we also briefly consider other applications of VBR video, including satellite transmission (channel sharing), playback of stored video, and wireless transport. Packet video research requires careful integration between the network and the video systems; however, a major stumbling block has resulted because commonly used terms are often interpreted differently by the video and networking communities.

Variable bit rate traffic that requires a boundeddelay network service is one of the most important types of traffic in future integrated services networks. In this paper, we introduce a new deterministic traffic model called Deterministic Bounding INterval-length Dependent (D-BIND) to capture the important multiplexing properties of bursty streams. With the D-BIND model, clients specify their traffic to the network via multiple rate-interval pairs, (Rk ; I k ), where a rate Rk is a bounding or worst-case rate over every interval of length I k . The model captures the intuitive property that over longer interval lengths, a source may be bounded by a rate lower than its peak rate and closer to its long-term average rate. We analyze the new model in the context of a deterministic service, and we quantify its performance benefits using a set of experiments with traces of MPEG-compressed video. We show that D-BIND's more accurate characterization of traffic streams leads to substantial im...

Distributed weighted fair scheduling schemes for QoS support in wireless networks have not yet become standard. In this paper we propose an Admission Control and Dynamic Bandwidth Management scheme that provides fairness in the absence of distributed link level weighted fair scheduling. In case weighted fair scheduling becomes available, our system assists it by supplying the scheduler with weights and adjusting them dynamically as network and traffic characteristics vary. To obtain these weights, we convert the bandwidth requirement of the application into a channel time requirement. Our Bandwidth Manager then allots each flow a share of the channel time depending on its requirement relative to the requirements of other flows in the network. It uses a max-min fairness algorithm with minimum guarantees. The flow controls its packet transmission rate so it only occupies the channel for the fraction of time allotted to it by the Bandwidth Manager. As available bandwidth in the network and the traffic characteristics of various flows change, the channel time proportion allotted also dynamically varies. Our experiments show that, at the cost of a very low overhead, there is a high probability that every flow in the network will receive at least its minimum requested share of the network bandwidth.

The burstiness of compressed video complicates the provisioning of network resources for emerging multimedia services. For stored video applications, the server can smooth the variable-bit-rate stream by prefetching frames into the client playback buffer in advance of each burst. Drawing on a priori knowledge of the frame lengths and client buffer size, such bandwidth smoothing techniques can minimize the peak and variability of the rate requirements while avoiding underflow and overflow of the playback buffer. However, in an internetworking environment, a single service provider typically does not control the entire path from the stored-video server to the client buffer. To develop efficient techniques for transmitting variable-bit-rate video across a portion of the route, we investigate bandwidth smoothing across a tandem of nodes, which may or may not include the server and client sites. We show that it is possible to compute an optimal transmission schedule for the tandem system by...