This paper considers the support of real-time applications in an

The Internet has been a startling and dramatic success. However, multimedia applications, with their novel traffic characteristics and service requirements, pose an interesting challenge to the technical foundations of the Internet. In this paper we address some of the fundamental architectural design issues facing the future Internet. In particular, we discuss whether the Internet should adopt a new service model, how this service model should be invoked, and whether this service model should include admission control. These architectural issues are discussed in a nonrigorous manner, through the use of a utility function formulation and some simple models. While we do advocate some design choices over others, the main purpose here is to provide a framework for discussing the various architectural alternatives.

Recent interest in supporting packet-audio applications over wide area networks has been fueled by the availability of low-cost, toll-quality workstation audio and the demonstration that limited amounts of interactive audio can be supported by today's Internet. In such applications, received audio packets are bu ered, and their playout delayed at the destination host in order to compensate for the variable network delays. In this paper we investigate the performance of four different algorithms for adaptively adjusting the playout delay of audio packets in an interactive packet-audio terminal application, in the face of such varying network delays. We evaluate the playout algorithms using experimentally-obtained delay measurements of audio tra c between several different Internet sites. Our results indicate that an adaptive algorithm which explicitly adjusts to the sharp, spike-like increases in packet delay which we observed in our traces can achieve a lower rate of lost packets for both a given average playout delay and a given maximum buffer size.

In packet audio applications, packets are buffered at a receiving site and their playout delayed in order to compensate for variable network delays. In this paper, we consider the problem of adaptively adjusting the playout delay in order to keep this delay as small as possible, while at the same time avoiding excessive “loss ” due to the arrival of packets at the receiver after their playout time has already passed. The contributions of this paper are twofold. First, given a trace of packet audio receptions at a receiver, we present efficient algorithms for computing a bound on the achievable performance of any playout delay adjustment algorithm. More precisely, we compute upper and lower bounds (which are shown to be tight for the range of loss and delay values of interest) on the optimum (minimum) average playout delay for a given number of packet losses (due to late arrivals) at the receiver for that trace. Second, we present a new adaptive delay adjustment algorithm that tracks the network delay of recently received packets and efficiently maintains delay percentile information. This information, together with a “delay spike ” detection algorithm based on (but extending) our earlier work [RKTS94], is used to dynamically adjust talkspurt playout delay. We show that this algorithm outperforms existing delay adjustment algorithms over a number of measured audio delay traces and performs close to the theoretical optimum over a range of parameter values of interest.

Integrated Services Packet Networks (ISPN) are designed to integrate the network service requirements of a wide variety of computer-based applications. Some of these services are delivered primarily through the packet scheduling algorithms used in the network switches. This paper addresses two questions related to these scheduling algorithms. The first question is: what scheduling services should an ISPN offer? In answer, we propose a scheduling service model for ISPN's which is based on our projections about future application and institutional service requirements. Our service model includes both a delay-related component designed to meet the ergonomic requirements of individual applications, and also a hierarchical link-sharing component designed to meet the economic needs of resource sharing between different entities. The second question we address is: what implications does this service model have for the packet scheduling algorithms? We answer this question by construc...

The Internet is currently being confronted with service demands from a new generation of applications. Supporting these applications effectively and efficiently will require extending the current Internet "best-effort" service model to one that offers an integrated suite of services. The purpose of this memo (which is derived primarily from [34]) is to describe a proposed "core" service model for an integrated services Internet. In the Appendix we discuss the process by which such a service model could be standardized by the Internet.

Measurement and analysis of the network behavior are crucial to understanding the Internet performance and designing appropriate control mechanisms for better performance. End hosts and their applications, however, have a limited capability in accessing and acquiring information about the network behavior. To them, end-to-end measurement of the network behavior is usually the only available information. This thesis focuses on two fundamental measures of network performance: end-to-end packet delay and loss.

We address interactive voice services over Best E ort packet networks where tra c is subject to unpredictable congestions. The quality ofvoice services can be signi cantly a ected by network delay and jitter due to the rejection of late packets. LIFO has proven to be an interesting overload strategy in delay constrained systems. We propose the use of the LIFO discipline in the context of congested packet networks where voice packets are subject to delay bounds. We analyze how delay accumulates in congested queues and show that serving voice packets in LIFO improves user-level quality. An extensive simulation study shows that LIFO scheduling in congested nodes signi cantly reduces the fraction of late packets compared to the FIFO discipline. The enhancement of user-level quality is emphasized by using the ITU-T P.861 standard for perceptual evaluations of reconstructed voice. Finally, weshowhow the proposed mechanism can be easily used in the Internet.

A multimedia system combines audio, video, graphics, and text into one presentation. Each of these multimedia data types has distinct temporal characteristics. For example video has a specific number of frames that must be displayed per second. There are also temporal relationships that exist between the media. In a movie application, the audio and video streams must be synchronized to achieve a lip syncing effect. In our system, we manage these temporal requirements through the scheduling of the communication channel; multimedia data is retrieved across the network at the appropriate time so that temporal presentation requirements are met. This real-time scheduling forms a basis for the limited a priori (LAP) scheduler. The scheduler assumes that it knows enough about the system a priori to schedule the next period or limited portion of the presentation. By considering only one period at a time, the scheduler can adapt to dynamic user input or changing communication channel character...

Traditional telephone conferencing has been accomplished by way of a centralized con-ference bridge. An Internet Protocol (IP)-based conference bridge is subject to speech distortions and substantial computational demands due to the tandem arrangement of high compression speech codecs. Decentralized architectures avoid the speech distortions and delay, but lack strong control and have a key dependence on silence suppression for endpoint scalability. One solution is to use centralized speaker selection and forwarding, and decentralized decoding and mixing. This approach eliminates the problem of tandem encodings and maintains tight control, thereby improving the speech quality and scalability of the conference. This thesis considers design options and solutions for this model, and evaluates performance through live conferences with real conferees. Conferees found the speaker selection of the new conference model to be transparent, and strongly preferred the resulting speech quality to that of a centralized IP-based conference bridge. i Sommaire