Protocols for High-Speed Networks V

A mechanism for the estimation of the available bandwidth between two end-points of a best-effort network is presented. The estimation is obtained by a simple statistical analisys of the effect which the network has on a synchronous train of packets. The possibility of exploiting self-similar characteristics of the delay jitters is also discussed.

The purpose of call admission control in Integrated Services Networks is to offer a guarantee that Quality of Service (QoS) bounds are not violated due to the admission of new calls into the network. This is typically accomplished using the declared worst-case traffic descriptors for incoming calls, a solution which results in poor bandwidth utilization. A measurement-based admission scheme is an appealing alternative: not only does it offer adaptivity to changing traffic conditions, it also allows statistical multiplexing gains to be exploited. In this paper, we examine the problem of determining which traffic characterization a measurement-based admission control algorithm should require of sources requesting access. Building on the work of Jamin et al. (1995), we also propose an adaptive measurement-based admission control algorithm that simplifies the estimation process, and show that it can achieve a high level of utilization without violating its delay-based QoS guarantees.

In this paper we study the TCP and XTP transport protocols in light of their possible use for applications running over B-ISDN in a IJBR-like context. In particular, we investigate the performance of file transfer applications taking into account both the impact of the ATM layer protocols and the features of the adopted transport protocol. The investigation is performed by simulation, using CLASS, a simulation tool for the study of ATM networks at the cell and burst levels, and considering quite a simple topology, known as the bottleneck topology, where TCP or XTP connections share a bottleneck link with some interfering background traffic, that is taken to have either Poisson or ON-OFF characteristics. Although the two transport protocols have quite different features, their performances are rather similar in almost all the considered scenarios. However, when XTP and TCP connections are mixed within the same network, the TCP performance is dramatically worse than that achieved by XTP, due to a more aggressive policy of XTP in the access to the transmission medium: a behavior that should raise some concern about the performance achievable when mixing different transport protocols on the same network.

This paper addresses the problem of transmission of digital Video communications over B-ISDN and provides a solution based on a good knowledge of both video system design and broadband networks capabilities. We propose a picture-oriented QoS control framework to ensure a high endto-end Quality of Picture (QoP) level for MPEG-encoded Video-On-Demand (VOD) applications carried over ATM networks. Our approach overcomes the difficulty imposed by random cell discarding due to the bursty and Variable Bit Rate transmission nature of compressed video, by using an accurate cell discrimination strategy and destination data recovery mechanisms. It is assumed that there are four hierarchical types of cell flows according to MPEG coding. Each flow has a different impact on the destination picture recovery process. Therefore, this scheme aims to minimize cell loss probability for critical data and thus, reduce the destination picture quality degradation, by adjusting the priority level of cells according to the carried data type. Despite these preventive measures, cell loss or errors may still occur. Therefore, complementary actions are cooperatively taken at the source and destination equipment (i.e., codecs), in conformity with the temporal requirements of MPEG video streams. In the one hand a fast cell-error detecting and recovery algorithm is proposed for protecting critical picture system information. In the other hand, a block-based bit interleaving schemes are designed to spatially disperse cell loss effect on reference frames. Finally, an extended AAL type 5 is presented to support such a high performance picture delivery framework.

We present and discuss detailed measurements showing how TCP behaves in an ATM LAN where a Variable Bit Rate (VBR) traffic contract is enforced by a Usage Parameter Control (UPC) mechanism. These measurements show that TCP has a lot of difficulties to adapt its behaviour to such a traffic contract. In our environment, TCP only achieved a 10% utilisation of a 10 Mbps VBR VC. We then discuss and evaluate possible solutions to improve this low utilisation of the reserved bandwidth.

The Network Interface Framework (NIF) is an object-oriented software architecture for providing networking services in the Choices object-oriented operating system. The NIF supports multiple client subsystems, provides clients with low-latency notification of received packets, and imposes no particular structure on clients. By contrast, traditional BSD UNIX-style networking does not meet the last two requirements, since it forces clients to use software interrupts and queueing. BSD UNIX cannot accomodate a process-based protocol subsystem such as the x-Kernel, whereas the NIF can. We have ported the x-Kernel to Choices by embedding it into the NIF. Using the standard x-Kernel protocol stack with NIF yields Ethernet performance comparable to BSD networking. The NIF is also flexible enough to support services that cannot easily be supported by traditional BSD, such as quality-of-service for multimedia. Preliminary performance results for asynchronous transfer mode (ATM) networks show that the NIF can be used to minimize jitter for continuous media data streams in the presence of non-realtime streams.

These latest years, technological improvements in Computer and Telecommunication areas led to the emergence of a new kind of distributed multimedia and co-operative applications, whose features and requirements are both multiple and diversified. In order to tackle these needs, two major architectural approaches are currently pursued, one of them arguing for an improvement of the application software, the other one promoting a more sophisticated network support. Following this last approach, this paper aims at providing implementation results performed around the design of a multimedia Transport architecture based on the partial order connection (POC) concept. Design principles of a multimedia partial order Transport connection (MM-POC) are first introduced. Main mechanisms of the associated protocol are then detailed and implementation investigations using the STREAM concept are exposed. Experimentation results are finally given.

Packet level forward error correction can be implemented by transmitting M redundancy packets after each set of N regular packets, so that all packets can be reconstructed if at least N out of N + M are received. The idea is usually dismissed because the gain is not worth the additionnai transmission overhead and the increased computation load, but further analysis shows that this dismissal may be questioned.

There are in fact at least three environment where the reduced error rate proves very valuable. When multicasting data toward large groups, even a small individual error rate per recipient may result in large retransmission rates for the whole group and the use of redundancy will result in dramatic efficiency gains. In the case of long transmission delays, the use of redundancy helps maintaining the delivery delays within acceptable limits, even in presence of errors. When the receivers do not have enough memory resources to implement sophisticated retransmission techniques, forward error correction can compensate the relative inefficiency of cheap algorithms of the go-back N family.

Packet level forward error correction is not very difficult to implement. The level of redundancy can easily be tuned as a function of the network’s characteristics. The additional robustness obtained through packet level redundancy helps implementing feedback control algorithms. In short, this seldom used technology could easily improve the performance of current transmission control protocols.

Multicast (1:N) is now supported by a number of networks and communication protocols. A problem in this context is how to provide fully reliable data transmission to the receiver group (or specified sub-group) in a heterogeneous environment. To achieve this an algorithm is required which is independent of any underlying network and multicast scheme. In this paper we present such an algorithm which provides fully reliable data transfer over Internet style networks as well as over ATM networks. The algorithm was initially inspired by mechanisms proposed for XTP 4.0 and proves to work well using XTP 4.0 protocol mechanisms. To evaluate its feasibility and performance it was tested with a series of simulations and implemented over MBone. The results presented in this paper show that it can provide fully reliable multicast to a relatively large number of receivers on top of different networks, protocols and protocol architectures.

The MBONE provides an infrastructure for multicast communication on the internet based on IP. Several proposals have been made to create reliable multicast transport on top of the MBONE structure. Nearly all research on reliable multicast protocols for the MBONE focuses on ARQ error recovery. The counterpart of ARQ, FEC, does not guarantee 100% reliability but increases the reliability. The aim of this paper is to determine the best place in a multicast tree for the use of FEC. We develop a framework that allows us to model analytically the impact of FEC on the average number of transmissions necessary to transmit a packet to all members of the multicast group. We look on different multicast tree topologies, different degrees of correlated loss in the multicast tree, different multicast group sizes and show the effect of FEC in terms of transmissions needed to achieve 100% reliability. We find that the shared part of the multicast tree is not always the best part to employ FEC.

This paper analyzes the performance of a reliable multicast transport protocol and discusses experimental test results. The Reliable Multicast Transport Protocol has been proposed to support “reliable” information delivery from a server to thousands of receivers over unreliable networks via IP-multicast. The protocol provides high-performance for most receivers through the advantage of IP multicast while also supporting temporarily unavailable or performance impaired receivers. Its applicability to large scale delivery is examined using an experimental network and the backoff time algorithm which avoids ACK implosion. The two types of flow control with the protocol are also examined. Separate retransmission is used to offset the local performance decline limited to a small number of receivers. Monitor-based rate control is used to offset the global performance declines due to causes such as network congestion.

We explore the performance effects of applying ILP to data manipulation functions with varying characteristics. The functions are generated from a set of parameters including input and output block size, state size and number of instructions. We present experimental data for varying function state sizes, number of integrated functions and instruction counts.

The results clearly show that the aggregated state of the functions must fit in registers for ILP to be competitive.

ALF (application level framing) and ILP (integrated layer processing) are protocol design and implementation concepts applied in high-performance communication architectures, e.g. to support multimedia applications. Writing ILP code is rather complex and, therefore, ILP code generation tools can reduce the time to develop efficient ILP protocol code significantly. This paper presents a tool, which allows automated ILP code generation based on high-level specifications of user data types. The tool is based on a stub compiler and integrates ILP loops into (un)marshalling routines automatically.

The long awaited ‘new environment’ of high speed broadband networks and multimedia applications is fast becoming a reality. However, few systems in existence today, whether they be large scale pilots or small scale test-beds in research labs, offer a fully integrated and flexible environment where multimedia applications can maximally exploit the quality of service (QoS) capabilities of supporting networks and end-systems. In this paper we describe the implementation of an adaptive transport system that incorporates a QoS oriented API and a range of mechanisms to assist applications in exploiting QoS and adapting to fluctuations in QoS. The system, which is an instantiation of the QoS Architecture (QoS-A), is implemented in a multi ATM switch network environment with Linux based PC end systems and continuous media file servers. A performance evaluation of the system configured to support Video-on-Demand application scenario is presented and discussed. Emphasis is placed on novel features of the system and on their integration into a complete prototype. The most prominent novelty of our design is a ‘distributed QoS adaptation’ scheme which allows applications to delegate to the system responsibility for augmenting and reducing the perceptual quality of video and audio flows when resource availability increases or decreases.

The rapid development towards networked multimedia applications leads to new and high requirements on the network as well as on the systems involved. Especially guaranteed Qualityof-Service (QoS) is required for certain applications with isochronous data streams or with highly interactive characteristics. Therefore, an integrated systemwide QoS management is needed. Among others, issues such as mapping of QoS parameters between fundamentally different levels of the system (user, application, network) and QoS monitoring are very important. This paper presents a framework for QoS management. Requirements of networked multimedia applications as well as several QoS management issues are discussed. Moreover, a QoS monitor as part of the QoS framework and its implementation is presented.