Search

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2010.01.013 Byline: P. Casas (a)(c), S. Vaton (a), L. Fillatre (b), I. Nikiforov (b) Keywords: Network Monitoring and Traffic Analysis; Traffic Matrix; Network Traffic Modeling; Optimal Volume Anomaly Detection and Isolation Abstract: Recent studies from major network technology vendors forecast the advent of the Exabyte era, a massive increase in network traffic driven by high-definition video and high-speed access technology penetration. One of the most formidable difficulties that this forthcoming scenario poses for the Internet is congestion problems due to traffic volume anomalies at the core network. In the light of this challenging near future, we develop in this work different network-wide anomaly detection and isolation algorithms to deal with volume anomalies in large-scale network traffic flows, using coarse-grained measurements as a practical constraint. These algorithms present well-established optimality properties in terms of false alarm and miss detection rate, or in terms of detection/isolation delay and false detection/isolation rate, a feature absent in previous works. This represents a paramount advantage with respect to current in-house methods, as it allows to generalize results independently of particular evaluations. The detection and isolation algorithms are based on a novel linear, parsimonious, and non-data-driven spatial model for a large-scale network traffic matrix. This model allows detecting and isolating anomalies in the Origin-Destination traffic flows from aggregated measurements, reducing the overhead and avoiding the challenges of direct flow measurement. Our proposals are analyzed and validated using real traffic and network topologies from three different large-scale IP backbone networks. Author Affiliation: (a) Telecom Bretagne, Brest, France (b) Universite de Technolgie de Troyes, Troyes, France (c) Universidad de la Republica, Montevideo, Uruguay Article History: Received 10 August 2009; Revised 13 January 2010; Accepted 23 January 2010 Article Note: (miscellaneous) Responsible Editor: A. Popescu

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2009.06.002 Byline: Ian W.C. Lee, Abraham O. Fapojuwo Keywords: Wireless TCP; Long-range dependence; Multifractals; Cascades; IP; Heavy-tailed distribution Abstract: In this paper we analyzed and modeled wireless TCP/IP traffic. Specifically, we focused on the interarrival times of TCP flows and the number of packets within a flow. We show that the marginal distribution of the flow interarrival times is piecewise Weibull distributed. Second and higher order statistics show that the flow interarrival times are long-range dependent and exhibit multifractal scaling. Taking these higher order properties into consideration, we proposed a multinomial canonical cascade with 3 stages to model the flow interarrival times. Looking at the IP layer, we find that the number of packets in a flow is heavy-tailed distributed. Especially interesting is that in 2 of our data sets, the number of packets in a flow possesses infinite mean. The interarrival time of packets within a flow is highly correlated, bursty, and its statistical characteristics vary from flow to flow. Author Affiliation: Department of Electrical and Computer Engineering, University of Calgary, Calgary, AB, Canada T2N 1N4 Article History: Received 23 July 2007; Revised 10 March 2009; Accepted 8 June 2009 Article Note: (miscellaneous) Responsible Editor: R. Sivakumar

This paper presents a link model which captures the queue dynamics in response to a change in a transmission control protocol (TCP) source's congestion window. By considering both self-clocking and the link integrator effect, the model generalizes existing models and is shown to be more accurate by both open loop and closed loop packet level simulations. It reduces to the known static link model when flows' round trip delays are identical, and approximates the standard integrator link model when there is significant cross traffic. We apply this model to the stability analysis of fast active queue management scalable TCP (FAST TCP) including its filter dynamics. Under this model, the FAST control law is linearly stable for a single bottleneck link with an arbitrary distribution of round trip delays. This result resolves the notable discrepancy between empirical observations and previous theoretical predictions. The analysis highlights the critical role of self-clocking in TCP stability, and the proof technique is new and less conservative than existing ones. Index Terms--Acknowledgments (ACKs), fast active queue management scalable transmission control protocol (FAST TCP).

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2008.07.015 Byline: Hwangnam Kim (a), Jennifer C. Hou (b) Keywords: Network calculus; Performance modeling and simulation Abstract: In this paper, we propose to speed up the simulation performance for TCP-operated networks by incorporating network calculus-based models in a simulation framework. In the simulation framework, we describes the operational properties of TCP congestion control, additive increase and multiplicative decrease (AIMD) and slow start, together with the queue discipline in the domain of network calculus. In order to determine network calculus models for TCP congestion, we first devise a simple TCP throughput model which approximately determines the range of per-flow throughput that one TCP can attain in a given interval, given the number of packet losses in the interval. We then exploit the TCP model to define a collection of network calculus theorems that regulate TCP flows in the range between the upper- and lower-limits on the TCP per-flow throughput in the corresponding interval. Finally, we incorporate the derived rules (theorems) into ns-2 to obtain a network calculus-based (NC-based) simulation, and carry out both the NC-based and the packet-level simulation to evaluate the performance gain and accuracy of the NC-based simulation, where the former is represented in perspective of the execution time (wall time) incurred in conducting the simulations and the later is evaluated in terms of the difference between results obtained in both the packet-level and NC-based simulation. The simulation results indicate that an order of magnitude or more (maximally 30 times) improvement in the execution time is achieved and the performance improvement becomes more salient as the network size increases in perspective of network-link capacities and the number of flows. The discrepancy observed between the NC-based simulation and the packet-level simulation, on the other hand, is minimally 1-2% and maximally 8-12% in a wide spectrum of network topologies and traffic loads. Additionally the results also indicate that the NC-based simulation outperforms a fluid-model-based simulation realized with the use of the time-stepped hybrid simulation technique, and that the performance improvement of the NC-based simulation is still held in IEEE 802.11-based wireless networks and also immune to the type of the simulation platform consisting of ns-2, operating system, and hardware specification. Author Affiliation: (a) School of Electrical Engineering, Korea University, Seoul 136-713, Republic of Korea (b) Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, IL 61801, United States Article History: Received 24 July 2007; Revised 20 March 2008; Accepted 12 July 2008 Article Note: (footnote) [star] A preliminary version of this manuscript appeared in IEEE INFOCOM 2004.

In this paper, we propose a novel probabilistic method for evaluating concurrent search in all-Internet protocol (IP) networks. It is expected that the next-generation Internet will consist of a variety of wireless-access networks. To be independent of wireless-access technology and to reduce the cost of mobility management in a heterogeneous environment, we propose integrating IP Paging with concurrent search, which was originally proposed for homogeneous cellular networks. We focus on the case in which the call-arrival process is general and has stationary independent increments. Since a number of calls could arrive simultaneously, the blocking probability is not equal to the probability with the system that is full. Therefore, we resort to renewal theory to accurately derive the blocking probability. Furthermore, we obtain the throughput, the average system size, and the average system delay. Our analytical results are consistent with simulation results. Index Terms--All-IP networks, concurrent search, mobility management, paging, performance analysis.

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2006.04.013 Byline: G. Carofiglio, M. Garetto, E. Leonardi, A. Tarello, M. Ajmone Marsan Keywords: Stochastic processes; Queueing theory; Simulation; IP networks; TCP mice; Fluid models Abstract: Fluid models of IP networks are based on a set of ordinary differential equations, that provide an abstract deterministic description of the average network dynamics. When IP networks operate close to saturation, fluid models were proved to provide reliable performance estimates. Instead, when the network load is well below saturation, standard fluid models lead to wrong performance predictions, since all buffers are forecasted to be always empty, so that the packet discard probability is predicted to be zero. These incorrect predictions are due to the fact that fluid models, being deterministic in nature, do not account for the random traffic variations that may induce temporary congestion of some network elements. In this paper we discuss three different approaches to describe random traffic variations in fluid models, considering randomness at both the flow and packet levels. With these approaches, fluid models allow reliable results to be obtained also in the case of IP networks that operate well below their saturation load. Numerical results are presented to prove the accuracy and the versatility of the proposed approaches, considering both stationary and non-stationary traffic regimes. Author Affiliation: Dipartimento di Elettronica, Politecnico di Torino, Italy Article History: Received 7 September 2005; Accepted 10 April 2006 Article Note: (miscellaneous) Responsible Editor: I. Matta

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2005.09.003 Byline: Giovanni Neglia (a), Vincenzo Falletta (a), Giuseppe Bianchi (b) Keywords: TCP marking; Differentiated services; Models Abstract: In Differentiated Services networks, packets may receive a different treatment according to their Differentiated Services Code Point (DSCP) label. As a consequence, packet marking schemes can also be devised to differentiate packets belonging to a same TCP flow, with the goal of improving the performance experienced. This paper presents an analytical model for an adaptive packet marking scheme proposed in our previous work. The model combines three specific sub-models aimed at describing (i) the TCP sources aggregate, (ii) the marker, and (iii) the network status. Preliminary simulation results show quite accurate predictions for throughput and average queue occupancy. Besides, the research suggests new interesting guidelines to model queues fed by TCP traffic. Author Affiliation: (a) Dipartimento di Ingegneria Elettrica, DIE, Universita degli Studi di Palermo, 90128 Palermo, Italy (b) Dipartimento di Ingegneria Elettronica, DIE, Universita degli Studi di Roma, Tor Vergata, 00133 Roma, Italy

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.comnet.2004.04.005 Byline: Anirban Mahanti (a), Derek L. Eager (b), Mary K. Vernon (c) Abstract: This paper describes Adaptive Vegas Multicast Rate Control (AVMRC), an equation-based multirate congestion control protocol that uses a recently proposed TCP Vegas throughput model. The AVMRC protocol exhibits TCP Vegas-like behavior and has the key advantage of operating without inducing packet losses when the bottleneck link is lightly loaded. The AVMRC protocol incorporates a new technique for dynamically adjusting the Vegas threshold parameters based on measured characteristics of the network. This technique implements fair sharing of network resources with widely deployed versions of TCP such as TCP Reno and might be fruitfully incorporated in TCP Vegas itself to aid in its incremental deployment. To evaluate the benefits of Vegas-like congestion control, the performance of AVMRC is compared using simulations to that of an analogous protocol that is based on a TCP Reno throughput model. Additional design choices in AVMRC are evaluated along four primary dimensions, namely synchronization policy, delay measurement, data transmission policy, and protocol reactivity. Author Affiliation: (a) Department of Computer Science, University of Calgary, 2500 University Dr., N.W. Calgary, AB, Canada T2N 1N4 (b) Department of Computer Science, University of Saskatchewan, 57 Campus Dr., Saskatoon, SK, Canada S7N 5A9 (c) Department of Computer Sciences, University of Wisconsin, 1210 West Dayton Street, Madison, WI 53706-1685, USA Article Note: (footnote) [star] This work was partially supported by the Natural Sciences and Engineering Research Council of Canada, and by the National Science Foundation under grant ANI-0117810.

The challenges involved in supporting Session Initiation Protocol (SIP)-based real-time and multimedia services over heterogeneous networks are examined. A SIP-enabled all IP-architecture (SEAA) that leverages 3rd Generation Partnership Project/3rd Generation Partnership Project 2 (3GPP/3GPP2) IP Multimedia Subsystem (IMS) framework, and IETF's SIP profile to support access-independent next-generation networks (NGN) is described.

This paper proposes and analyzes the performance of a hybrid selective repeat (SR)/multicopy (MC) automatic repeat request (ARQ) scheme to transmit fragmented internet protocol (IP) packets. The ARQ scheme works in SR mode until the last IP packet fragment is transmitted. If a fragment is negatively acknowledged after the last fragment is transmitted, then the system goes in MC mode. In MC mode, multiple copies of the erroneous fragment are transmitted. After all IP fragments are received without error, the system goes back to SR mode. Performance of the proposed scheme is evaluated in terms of bit error rate (BER), IP packet size, and fragmentation size with and without Bose Chaudhuri Hocquenghem (BCH) error correction codes. Both the results are obtained under additive white Gaussian noise (AWGN) as well as fiat Rayleigh fading channels. The proposed scheme gives a throughput of 0.9, even at high BER conditions, for any IP packet size under AWGN channel while, an 8-dB improvement is achieved, when using BCH(63, 51, 2) for throughput of 0.9, over selective repeat + stutter scheme 2 (SR+ST 2) under a fiat Rayleigh fading channel. Index Terms - Fading channel, IP packet transmission, multicopy retransmission, selective-repeat ARQ, spread spectrum, wireless communication.