This paper explains why long Distributed Queue Dual Bus (DQDB) networks without bandwidth balancing can have fairness problems when several nodes are performing large file transfers. The problems arise because the network control information is subject to propagation delays that are much longer than the transmission time of a data segment. Bandwidth balancing is then presented as a simple solution. By constraining each node to take only a certain fraction of the transmission opportunities offered to it by the basic DQDB protocol, bandwidth balancing gradually achieves a fair allocation of bandwidth among simultaneous file transfers. We also propose two ways to extend this procedure effectively to multi-priority traffic. __________________ This paper appears in IEEE Transactions on Communications, Vol. 40, No. 7, July 1992, pp. 1192-1204. -- -- - 2 - 1. INTRODUCTION The Distributed Queue Dual Bus (DQDB) [1], [2] is a metropolitan area network that has recently been standardized by ...

The most important part of an internet protocol is its addressing information---that is, the information that affects routing of an internet packet. While there has been much research of routing in internetworks, there has not been a comprehensive study of addressing information in internet protocols per se. In this thesis, we examine the use of addresses in internet protocols. We start with a taxonomy of addressing functions. Using this taxonomy, we then give a comprehensive description of addressing modes in internetwork protocols. Finally, we present two designs for addressing in internetwork protocols, one based on current internet protocol syntax (SIPP), and one based on a new syntax (SPip). Both of these designs exploit the notion that virtually all routing and addressing semantics can be achieved through the loose source route mechanism, though SPip does this more generally than SIPP. We analyze the capabilities and costs of SIPP and SPip, and compare them with those of OSI's in...

DTM is a fiber-optic network, based on bandwidth reservation and with support for dynamic reallocation of bandwidth. It is designed for real-time multimedia applications and for high-speed computer communication. DTM uses a novel medium-access technique and provides a multicast connection-oriented service. Several DTM networks can be connected into one large network. A prototype implementation and testbed is being constructed.

This paper deals with the distributed queue dual bus (DQDB) (IEEE 802.6) network, and makes two independent contributions. First, the unfairness problem of DQDB is addressed, and several alternative solutions that can improve the network’s fairness are proposed. They include (1) the proportional assignment scheme (PR); (2) the (multiple-request) FCFS-message-queue-based DQDB scheme (MD); and (3) a combination of MD and PR, denoted by MP. Implementation methods that require simple additional hardware on top of the regular DQDB interface are outlined. Simulation examples are employed to compare the performance of the above schemes and to gain insights into their characteristics. The performance of these schemes are also compared with those of regular DQDB and bandwidth balancing DQDB. The second contribution of this paper is the development of an analytical model of the DQDB network. By employing some constrained assumptions for analytical tractability, a Markov chain model for (an earlier version of) the entire DQDB network is formulated. The analytical model can predict an individual station’s throughput and mean packet delay for known (possibly-asymmetric) loading patterns. Also, it can be relatively easily extended to approximately model both bandwidth balancing DQDB and the PR scheme. The model is verified via simulation.

A discrete-time queueing system operating under a two-level, consistent priority service policy is studied in this paper. The consistency of the policy guarantees that no low priority customer will be served before a previously (or simultaneously) arrived high priority one. Unlike the well known head of the line priority policy (which is consistent), the considered policy provides for limited service to low priority customers, even in the presence of high priority ones. The proposed policy may be viewed as a consistent version of the straightforward gated/limited service priority policy. It may also be viewed as a compromise between the head of the line priority policy and the straightforward gated/ limited priority service policy. The customer service time is assumed to be deterministic and equal to one time unit, which renders the queueing model applicable to a packetized communication network environment; potential relevant applications are presented. Based on renewal arguments, the theory of infinite dimensional linear equations and a workconservation law, a general methodology is developed for the derivation of arbitrarily tight bounds on the induced mean packet delay.

Several important modifications to the distributed queue dual bus (DQDB) (IEEE 802.6) protocol for metropolitan area networks (MANs) are proposed. These modifications are aimed at improving not only the throughput and delay performance of the network but also its fairness performance. The first proposal exploits the continuation-of-message (COM) information in a DQDB segment in order to minimize the mean overhead for multi-packet messages. This reduces the net offered load to the system and thus improves its throughput and delay performance. Then, a scheme for combining the COM information along with the spatial reuse of slots is proposed, thereby achieving even higher throughput and lower delay. Simulation models are used to study the performance gains achievable by using the COM and the slot reuse approaches. Although slot reuse has already been proposed before to improve the DQDB protocol’s performance, the unfairness of such a system has been given little attention so far. Moreover, since the unfairness becomes more pronounced when slots are reused, the study of schemes that can achieve fairness in such environments becomes all the more important. This paper also proposes several schemes to achieve different fairness criteria in a DQDB network employing slot reuse. In particular, the following fairness criteria are studied: (a) an individual station’s throughput proportional to the number of stations downstream from itself, (b) equal throughput at all stations, and (c) an individual station’s throughput proportional to its own offered load. Analytical results corresponding to these schemes are developed, and they are verified via simulation. The proposed schemes can be implemented with just one additional counter and one register (on top of the regular DQDB hardware) at each station.

Packet switching technology promises to allow improvement of video quality by efficiently supporting variable rate video coding. Its inherent multiplexing of multiple streams also allows more efficient multi-destination delivery for N-way conferencing. However, most commercial video codecs are designed to work with circuits, not packets, in part because these benefits are accompanied by some problems. This paper describes a packet video system implementation in which commercial codecs were adapted to exploit the benefits of packet switching while addressing the problems as follows: 1. Clock synchronization was obviated by asynchronous operation; 2. Delay was reduced by bandwidth reservation and fast packet forwarding; 3. Packet loss was reduced by bandwidth reservation and forward error correction. An overview of the system is followed by sections addressing each of the problems and benefits, plus future directions for expansion of the system. System Description The Multimedia Co...

In this paper a multiple access protocol is proposed for a system consisting of many high-speed bursty traffic stationa via an optical passive star coupler. Each u.w has access over a range of wavelengths, thus resulting in a wavelength division multiplexed communication. A station has b reserve a wavelength first then transmits the data on that wavelength. The performance has been modeled and analyzed with the Equilibrium Point Analysis (EPA) technique. Numerical results show that low delay and high throughput (larger than the electronic speed of a single user) can be achieved. The analysis also showa that the best performance is obtained when the capacities of the reservation channels and the data channels are balanced. 1