Showing papers on "Packet loss published in 1983"

Techniques for Packet Voice Synchronization

Abstract: Packet switching has been proposed as an effective technology for integrating voice and data in a single network. An important aspect of packet-switched voice is the reconstruction of a continuous stream of speech from the set of packets that arrive at the destination terminal, each of which may encounter a different amount of buffering delay in the packet network. The magnitude of the variation in delay may range from a few milliseconds in a local area network to hundreds of milliseconds in a long-haul packet voice and data network. This paper discusses several aspects of the packet voice synchronization problem, and techniques that can be used to address it. These techniques estimate in some way the delay encountered by each packet and use the delay estimate to determine how speech is reconstructed. The delay estimates produced by these techniques can be used in managing the flow of information in the packet network to improve overall performance. Interactions of packet voice synchronization techniques with other network design issues are also discussed.

A Layered Network Protocol for Packet Voice and Data Integration

Abstract: This paper discusses considerations in the design of packet protocols suitable for interactive voice and interactive data communication, and then outlines a potential layered protocol architecture for the internal communication of a long haul network that might support packet voice and packet data transport. Following the protocol description, the paper compares potential delays for two voice/data packet network architectures: one using only link retransmission, the other using only edge retransmission for data (as included by the voice/data protocol). The underlying data traffic loads offered to the network are the same for the two methods, although they give rise to different traffic patterns. This preliminary analysis shows that the average delay using an edge-to-edge recovery discipline can be made comparable to the delays introduced with a link-by-link recovery discipline, if the network uses high speed transmission facilities (e.g., over 1 Mbits/s) having good error characteristics (e.g., one or less packets corrupted in each 1000), and sends up to 128 bytes of customer data as a single packet.

Stability of Adaptive Congestion Controls in Packet Networks.

Abstract: Some congestion controls used in packet networks are static, while others are adaptive, i.e., they respond on the basis of estimated load determined from measurements. Adaptive controls pose a potential stability problem. If the network reacts too strongly to overload, its load level may oscillate widely. In this paper, we examine the effect of offered overload and measurement error on the response of the network. We also study the effect of smoothing measurements. The model requires non-zero time to respond after a measurement of the load is taken. Results include conditions for convergence to a desired load, rate of convergence, and variance of system load as time goes to infinity as a function of system parameters.

A bang-bang principle for real-time transport protocols

Abstract: A real-time protocol is usually concerned with the transportation of a real-time data stream over a packet switched network. Among the major issues distinguishing real-time protocols from ordinary transport protocols is the problem of trading delay for loss. That is, if some loss of packets may be acceptable, and usually inevitable, the objective of the protocol is to minimize the delay of packets, subject to constraints on the acceptable loss. This is unlike usual transport protocols, which are designed to guarantee no loss at the expense of increased delays.This paper presents a model for the delay-loss tradeoffs in a family of real-time transport protocols. It is demonstrated that, under very general assumptions, an optimal real-time protocol is a bang-bang protocol: there exists a threshold queue length such that as long as the packet queue length at the sender is less than the threshold the protocol should be an ordinary positive acknowledgement with retransmissions transport protocol. However, as soon as the threshold queue length is exceeded, a newly arriving packet causes the first (or any other) packet in the queue to be discarded. Closed form expressions for the threshold buffer size are obtained and analyzed in terms of the given parameters of the system.

Design and performance evaluation of a packet voice system

Abstract: A packet voice network has been designed using adaptive delta modulators. Tests were performed to examine the effects of packet length and packet loss rate on digital voice intelligibility. The packet voice network was simulated using a SLAM network model to find queue size requirements as a function of packet size, delay distribution, and expiring time out. The queuing strategy is investigated in order to minimize total information loss during channel acquisition delays.

Communication system between processes

Abstract: PURPOSE: To receive packets without lowering the processing efficiency of a computer in case of transmission to another process, by providing a message transmission instruction processing means to perform the processing where packets in a packet storage area are transferred to a process. CONSTITUTION: If a packet is not queued in a process control block 1-n, (0) is written in a data packet entry column 5. If the block 1-n points a packet storage area 3-J and the storage area 3-J points a packet storage area 3-K, (0) is written in a pointer column because no link packets exist in the storage area 3-K. When a packet in the storage area 3-J is written in a data area of a destination process in this state, the block 1-n points the storage area 3-K. In this manner, the presence or the absence of an arrival packet is checked simultaneously with transmission when a packet transmission instruction is issued, and the arrival packet is transferred to the process if it is present, and thus, packets are received without lowering the processing efficiency of the computer. COPYRIGHT: (C)1984,JPO&Japio

Theoretical performance analysis of sliding window link level flow control for a local area network

Abstract: A transmitter sends packets over a local area network to a receiver. The receiver has a finite amount of storage capacity for buffering messages. A sliding window protocol is employed to insure all packets are delivered without error in order of arrival. Mean throughput rate and packet delay statistics are studied as a function of model parameters. Simulation data are presented.

Accurate method for analysis of a packet speech multiplexer

Abstract: A packet speech multiplexer is analysed using a fluid approximation for the flow of packets in and out of the multiplexer. Delay distributions and fractional packet loss are determined with a small amount of computation. Comparisons with a simulation using real speech show the analysis to be accurate.

On the design of prioritized multiplexing systems

Abstract: Systems in which many data sources are multiplexed over a single communication channel are considered. Data from all the sources are generated in fixed-length packets, and are stored in a common buffer with finite capacity. Packets overflowed or removed from the buffer prior to transmission are lost. The system performance measure is the set of packet loss probabilities associated with the sources. Queueing disciplines vary depending on the stringency of prioritization and the utilization of system resources. The set of all possible performances is characterized as the set of all queueing disciplines is spanned. Whether a given performance is possible can be deduced. Strategies that achieve the minimum overall loss probability are identified. The extreme disciplines are specified, and their performances are calculable by means of a given algorithm.

H-network-a high speed distributed packet switching local computer network

Abstract: The h-network is a high-speed distributed local computer network designed to facilitate user interaction and control of the homogeneous multiprocessor. The topology of the network and its access protocol are discussed. Performance measures based on analysis of the protocol and the hardware used are presented. 3 references.