Link Buffer Research Articles

On the limitation of fluid-based approach for Internet congestion control

Fluid models have been the main tools for Internet congestion control. By capturing how the average rate of each flow evolves, the fluid model proves to be useful as it predicts the equilibrium point to which system trajectory converges and also provides conditions under which the convergence is ensured, i.e., the system is stable. However, due to inherent randomness in the network caused by random packet arrivals or random packet marking, the actual system evolution is always of a stochastic nature. In this paper, we show that we can be better off using a stochastic approach toward the congestion control. We first prove that the equilibrium point of a fluid model can be quite different from the true average rate of the corresponding stochastic system. After we describe the notion of stability for two different approaches, we show that a stable fluid model can impose too much restriction on our choice of system parameters such as buffer size or link utilization. In particular, under fluid models, we show that there exists a fundamental tradeoff between the link utilization and buffer size requirement for large systems, while in a more realistic setting with stochastic models, there is no such tradeoff. This implies that the current congestion control design can be much more flexible, to the benefit of efficient usage of network resources.

Radio Link Buffer Management and Scheduling for Wireless Video Streaming

In this paper we compare strategies for joint radio link buffer management and scheduling for wireless video streaming. Based on previous work in this area [8], we search for an optimal combination of scheduler and drop strategy for different end-to-end streaming options including timestamp-based streaming and ahead-of-time streaming, both with variable initial playout delay. We will show that a performance gain versus the two best drop strategies in Liebl et al. [8], i.e. drop the HOL packet or drop the packet with the lowest priority starting from HOL, is possible: Provided that some basic side-information on the structure of the incoming video stream is available, a more sophisticated drop strategy removes packets from an HOL group of packets in such a way that the temporal dependencies usually present in video streams are not violated. This advanced buffer management scheme yields significant improvements for almost all investigated scheduling algorithms and streaming options. In addition, we will demonstrate the importance of fairness among users when selecting a suitable scheduler, especially if ahead-of-time streaming is to be applied: Given a reasonable initial playout delay at the streaming media client, both the overall achievable quality averaged over all users, as well as the individual quality of users with bad channel conditions can be increased significantly by trading off fairness with maximum throughput of the system.

Long range dependence in network traffic and the closed loop behaviour of buffers under adaptive window control

We consider an Internet link carrying http-like traffic, i.e., transfers of finite volume files arriving at random time instants. These file transfers are controlled by an adaptive window protocol (AWP); an example of such a protocol is TCP. We provide analysis for the auto-covariance function of the AWP-controlled traffic into the link’s buffer; this traffic, in general, cannot be represented by an on–off process. The analysis establishes that, for TCP-controlled transfer of Pareto-distributed file sizes with infinite second moment, the traffic into the link buffer is long range-dependent (LRD). We also develop an analysis for obtaining the stationary distribution of the link buffer occupancy under an AWP-controlled transfer of files sampled from some distribution. For any AWP, the analysis provides us with the Laplace–Stieltjes transform (LST) of the distribution of the link buffer occupancy process in terms of the functions defining the AWP and the file size distribution. The analysis also provides a necessary and a sufficient condition for the finiteness of the mean link buffer content; these conditions again have explicit dependence on the AWP used and the file size distribution. This establishes the sensitivity of the buffer occupancy process to the file size distribution. Combining the results from the above analyses, we provide various examples in which the closed loop control of an AWP results in finite mean link buffer occupancy even though the file sizes are Pareto-distributed (with infinite second moment), and the traffic into the link buffer is long range-dependent (with Hurst parameters which would suggest an infinite mean queue occupancy under open loop analysis). We also study the effect of window reductions due to active queue management and find that window reductions lead to further lightening of the tail of buffer occupancy distribution. The significance of this work is three-fold: (i) by looking at the window evolution as a function of the amount of data served and not as a function of time, this work provides a new framework for analysing various processes related to the link buffer under AWP-controlled transfer of files with a general file size distribution; (ii) it indicates that the buffer behaviour in the Internet may not be as poor as predicted from an open loop analysis of a queue fed with LRD traffic; and (iii) it shows that the buffer behaviour (and hence the throughput performance for finite buffers) is sensitive to the distribution of file sizes.

The lightening effect of adaptive window control

We consider adaptive window (e.g., TCP) controlled transfer of http-like traffic (i.e., finite volume file transfers starting at random time instants) over a link, and develop an analysis for obtaining the stationary distribution of the link buffer occupancy. The significance of this work is that it provides a framework for analyzing the "closed-loop" behavior of the link buffer under adaptive window controlled transfer of files with a general file size distribution. We find that the tail of the stationary distribution of the buffer occupancy is lighter than may be expected from an open loop analysis. We also find that the tail of the buffer distribution is sensitive to the distribution of the file sizes, which provides insight into the sensitivity of TCP performance to file size distribution.

Design and Modeling of an Interval-based ABR Flow Control Protocol

A novel flow control protocol is presented for Availability Bit Rate (ABR) service in Asynchronous Transfer Mode (ATM) networks. This scheme features periodic explicit rate feedback that enables precise allocation of link bandwidth and buffer space on a hop-by-hop basis to guarantee maximum throughput, minimum cell loss, and high resource efficiency. With the inclusion of resource management cell synchronization and consolidation algorithms, this protocol is capable of controlling point-to-multipoint ABR services within a unified framework. The authors illustrate the modeling of single ABR connection, the interaction between multiple ABR connections, and the constraints applicable to flow control decisions. A loss-free flow control mechanism is presented for high-speed ABR connections using a fluid traffic model. Supporting algorithms and ATM signaling procedures are specified, in company with linear system modeling, numerical analysis, and simulation results, which demonstrate its performance and cost benefits in high-speed backbone networking scenarios.

Procedures and tools for analysis of network traffic measurements

We present procedures and tools for the analysis of network traffic measurements. The tools consist of stand-alone 1 The stand-alone software modules can be downloaded from http://www.ics.forth.gr/netgroup/msa; the web interface is available at http://trace.ucnet.uoc.gr 1 modules that implement advanced statistical analysis procedures and a flexible web-based interface through which a user can create, modify, save, and execute experiments using the statistical analysis modules. The tools do not assume a specific source traffic model, but rather process actual measurements of network traffic. Indeed, the theory that the tools are based on can identify the time-scales that affect a link’s performance, hence suggest the appropriate time granularity of traffic measurements. We present and discuss case studies that demonstrate the application of the tools for answering questions related to network management and dimensioning, such as the maximum link utilization when some quality of service is guaranteed, how this utilization is affected by the link buffer and traffic shaping, the acceptance region and the effects of the scheduling discipline, and the token (or leaky) bucket parameters of a traffic stream. The case studies involve actual traffic measurements obtained by a high performance measurement platform that we have deployed at the University of Crete network.

Threshold-based congestion control for the SS7 signaling network in the GSM digital cellular network

This paper considers the congestion control scheme for the SS7 signaling network in the group special mobile (GSM) digital cellular network. This congestion control scheme is based on monitoring the SS7 link buffer occupancy. In this scheme, a congestion onset message is sent to the user parts of the SS7 network when the buffer occupancy exceeds a certain threshold, and, subsequently, a congestion abatement message is sent when the buffer occupancy goes below another threshold. Upon receipt of the congestion onset message, the user parts are expected to "intelligently" throttle the user traffic (reduce the traffic rate) so as to yield speedy recovery from congestion. Subsequently, on receipt of the abatement message, the user traffic is restored to precongestion levels. This paper primarily proposes appropriate choice of throttles and an algorithmic procedure to size the thresholds so as to yield good performance during congestion. The paper also addresses some implementation issues related to the throttles. Finally, it considers the effects of delays for the onset and abatement messages in reaching the user parts on the performance and parameters of the congestion control scheme.

Methodology for derivation of network resources to support video-related services in ATM-based private wide-area networks

The statistical mode of operation of asynchronous-transfer-mode (ATM) networks can lead to congestion which, in turn, leads to a degradation in the quality of service (QOS) of calls. Normally, congestion manifests itself by overflow in the input or output controllers of switches. An outstanding research issue relating to ATM networks, therefore, is resource management; i.e. determining the optimum network resources, in terms of link bandwidth and buffer capacity in switches, that are required for a call which guarantees a defined QOS, even under high-network-loading conditions. The two most important QOS parameters associated with a cell are cell-loss rate and cell-delay variation. In practice, the parameters are interrelated and make conflicting demands on the available resources. Video services, for example, are sensitive to both parameters and hence it is necessary to consider the combined parameters when allocating resources for such services. The paper describes a new method for deriving the optimum resources required for video-related services in an ATM-based private network to guarantee specific QOS requirements.

Performance analysis of virtual circuit connections for bursty data sources in ATM networks

ATM is a packet-like transmission mode that has been proposed for BISDN. It is characterized by an asynchronous slotted transmission mechanism that provides a high bandwidth, low delay connection-oriented transport service to the end user. In this paper, we provide an analytical approach for determining the performance of a virtual circuit connection for data transmission in a high-speed ATM network with finite buffers at the network nodes. The analysis assumes that the network operates using thebest effort delivery strategy and that the end-to-end virtual circuit is responsible for guaranteeing the integrity of the connection. Since the normal Markovian assumptions do not apply, a concise exact solution is impossible to obtain. This provides motivation for developing approximate techniques such as those found in Whitt's QNA paper that allow us to use general distributions for the traffic streams and service times. However, even these techniques assume infinite buffer capacities and hence cannot model buffer overflow. We have therefore developed a hybrid model that allows us to incorporate finite buffers at the nodes. This enables us to study the effect on the performance of both link errors and buffer overflow in conjunction with an end-to-end packet loss recovery scheme.

Voice packet loss: destination versus internodal links

Voice packet loss behavior at both the destination and internodal links in a packet-switched network is investigated. The fractional loss and blocking time periods for both are derived using a bivariate Markov model. The numerical results show that blocking due to the delay constraint at the destination can result in long periods of consecutive packet loss, which seriously degrade voice quality. The authors' work indicates that packets with excessive delay should be discarded at the internodal links, instead of blocking them at the destination. The relation between the internodal link buffer size and end-to-end permissible queueing delay is established.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>