Publications
Network Congestion Papers and Presentations
Fischer, M. J., D. M. B. Masi, and J. F. Shortle, “Approximating Low Latency Queueing Buffer Latency,” The Fourth Advanced International Conference on Telecommunications (AICT 2008), Athens, Greece, June 8-13, 2008.
Abstract [ - ]
Low Latency Queueing (LLQ) is an Internet Protocol (IP) router discipline that is being used to ensure that performance-sensitive high priority traffic, such as voice and video, receive their high level of performance, while allowing less performance-sensitive traffic, such as e-mail or best-effort IP, to receive some portion of the bandwidth. In this paper, we develop a simple analytic approximation for the buffer latency (expected buffer delay) for each traffic class using the LLQ system. The approximation is validated via a simulation model.
Low Latency Queueing (LLQ) is an Internet Protocol (IP) router discipline that is being used to ensure that performance-sensitive high priority traffic, such as voice and video, receive their high level of performance, while allowing less performance-sensitive traffic, such as e-mail or best-effort IP, to receive some portion of the bandwidth. In this paper, we develop a simple analytic approximation for the buffer latency (expected buffer delay) for each traffic class using the LLQ system. The approximation is validated via a simulation model.
Masi, D. M. B., M. J. Fischer, and D. A. Garbin, “Video Frame Size Distribution Analysis,” 9th INFORMS Telecommunications Conference—Telecommunications Modeling, Policy, and Technology, College Park, MD, March 27-29, 2008.
Abstract [ - ]
The purpose of this research is to determine what statistical distributions are most appropriate to represent the video traffic, for use in congestion models—analysis emphasis is on video frames, although some research was also done on the packets within the frames.
The purpose of this research is to determine what statistical distributions are most appropriate to represent the video traffic, for use in congestion models—analysis emphasis is on video frames, although some research was also done on the packets within the frames.
Fischer, M. J., D. M. B. Masi, and J. F. Shortle, “On the Relationship Between the Low Latency Queueing and the Non-Preemptive Priority Queueing Model,” 9th INFORMS Telecommunications Conference—Telecommunications Modeling, Policy, and Technology, College Park, MD, March 27-29, 2008.
Abstract [ - ]
The purpose of this presentation is to present our observations in the development of performance models for the Low Latency Queueing (LLQ) system and how the standard Non-Preemptive Priority Queueing (NPPQ) system plays an important role in this development.
The purpose of this presentation is to present our observations in the development of performance models for the Low Latency Queueing (LLQ) system and how the standard Non-Preemptive Priority Queueing (NPPQ) system plays an important role in this development.
Shortle, J., M. Fischer, P. Brill. “Waiting-time distribution of M/DN/1 queues through numerical Laplace inversion, “INFORMS Journal on Computing, Winter, 2007
Abstract [ - ]
This paper considers an M/G/1 queue where the service time for each customer is a discrete random variable taking one of N values. We call this queue anM/DN/1 queue. There are potential numerical problems inverting Laplace transforms associated with this queue because the service distribution is discontinuous. The purpose of this paper is to investigate the performance of numerical transform inversion methods in analyzing this queue. We first derive continuity properties of the steady-state distribution of wait in the M/DN/1 queue. Then, we show analytically how continuity properties affect the performance of the Fourier method for inverting transforms. In particular, continuity is not required in all derivatives for best performance of the method. Then, we present a new inversion method specifically for the M/DN/1 queue. Finally, we give numerical experiments comparing these and four other inversion methods. Although no method clearly dominates,the recursion method performs very well in all examples.
This paper considers an M/G/1 queue where the service time for each customer is a discrete random variable taking one of N values. We call this queue anM/DN/1 queue. There are potential numerical problems inverting Laplace transforms associated with this queue because the service distribution is discontinuous. The purpose of this paper is to investigate the performance of numerical transform inversion methods in analyzing this queue. We first derive continuity properties of the steady-state distribution of wait in the M/DN/1 queue. Then, we show analytically how continuity properties affect the performance of the Fourier method for inverting transforms. In particular, continuity is not required in all derivatives for best performance of the method. Then, we present a new inversion method specifically for the M/DN/1 queue. Finally, we give numerical experiments comparing these and four other inversion methods. Although no method clearly dominates,the recursion method performs very well in all examples.
Fischer, M.J., D.M. Masi and P. McGregor, “Making an Efficient Integrated Services Enterprise Network,” IT-Pro, pgs. 19-26, September-October 2007.
Abstract [ - ]
Packet switching Internet Protocol (IP) technology can efficiently support integrated voice and data services, but doing so for the enterprise requires different engineering than for the public infrastructure. At public network core bandwidths, very high utilization with the simplest First Come, First Served (FCFS) queueing discipline can yield a very efficient implementation that still meets Quality of Service (QoS) objectives. However, at typical enterprise access speeds, efficiently achieving QoS during the busy hour requires use of more sophisticated queueing disciplines such as Priority Queueing (PQ) and Weighted Fair Queueing (WFQ). We found that for OC-3 (155 Mb/s) speeds and higher, i.e., public network speeds, the FCFS discipline achieves QoS with utilizations greater than 95%. For T-3 (45 Mb/s) speeds, changing from FCFS to PQ or WFQ is required to achieve QoS above 90% utilization. For T-1 speeds (1.5 Mb/s) the change is required to achieve QoS above 70% utilization.
Packet switching Internet Protocol (IP) technology can efficiently support integrated voice and data services, but doing so for the enterprise requires different engineering than for the public infrastructure. At public network core bandwidths, very high utilization with the simplest First Come, First Served (FCFS) queueing discipline can yield a very efficient implementation that still meets Quality of Service (QoS) objectives. However, at typical enterprise access speeds, efficiently achieving QoS during the busy hour requires use of more sophisticated queueing disciplines such as Priority Queueing (PQ) and Weighted Fair Queueing (WFQ). We found that for OC-3 (155 Mb/s) speeds and higher, i.e., public network speeds, the FCFS discipline achieves QoS with utilizations greater than 95%. For T-3 (45 Mb/s) speeds, changing from FCFS to PQ or WFQ is required to achieve QoS above 90% utilization. For T-1 speeds (1.5 Mb/s) the change is required to achieve QoS above 70% utilization.
Garbin, D.A. , P. McGregor, D. M. Masi , “Using Event Simulation to Evaluate Internet Protocol Enhancements for Special Service,” Invited Paper, Winter Simulation Conference 2007, Washington DC, December 2007
Abstract [ - ]
Disasters can cause extraordinary service demand by thepublic, while concurrently causing outages that reducenetwork capacity to serve the surging demand. As the public network infrastructure for telecommunications services transforms into one based on packet switching and the Internet Protocol (IP), it is imperative that services supporting management of disaster response continue to perform with minimal degradation during such events.Mechanisms exist within IP-based networks to provide preferential treatment for real-time services such as voice and video using Differentiated Services Code Points (DSCP) in the packet headers and corresponding different Per Hop Behaviors (PHB) in the routers. However, there is currently no way to identify voice and video packets supporting response management and to ensure their timely delivery during periods of extreme network overload. The authors and contributors are particularly concerned with ensuring that essential network services for managing disaster response are assured under all circumstances. We have applied event simulation to evaluate the potential benefit of two additional DSCP markings to be applied to such voice and video packets. Several router configurations were investigated for their ability to utilize the new code points and assure the performance of the associated packets. The results demonstrate significant value of the additions in preserving disaster response management performance even when significant aberration in demand and/or capacity causes ordinary voice and video performance to significantly degrade.
Disasters can cause extraordinary service demand by thepublic, while concurrently causing outages that reducenetwork capacity to serve the surging demand. As the public network infrastructure for telecommunications services transforms into one based on packet switching and the Internet Protocol (IP), it is imperative that services supporting management of disaster response continue to perform with minimal degradation during such events.Mechanisms exist within IP-based networks to provide preferential treatment for real-time services such as voice and video using Differentiated Services Code Points (DSCP) in the packet headers and corresponding different Per Hop Behaviors (PHB) in the routers. However, there is currently no way to identify voice and video packets supporting response management and to ensure their timely delivery during periods of extreme network overload. The authors and contributors are particularly concerned with ensuring that essential network services for managing disaster response are assured under all circumstances. We have applied event simulation to evaluate the potential benefit of two additional DSCP markings to be applied to such voice and video packets. Several router configurations were investigated for their ability to utilize the new code points and assure the performance of the associated packets. The results demonstrate significant value of the additions in preserving disaster response management performance even when significant aberration in demand and/or capacity causes ordinary voice and video performance to significantly degrade.
Masi, D.M. , M.J. Fischer, D.A. Garbin , “Modeling the Performance of Low Latency Queueing for Emergency Telecommunications,” Invited Paper, Winter Simulation Conference 2007, Washington DC, December 2007
Abstract [ - ]
Event simulation and analytic modeling are used to evaluate the performance of Low Latency Queueing (LLQ), a queueing discipline available in some Internet packet switching routers for integrated services performance assurance. LLQ combines priority queueing with Class-Based Weighted Fair Queueing (CBWFQ). Priority queueing is used to ensure satisfying tight delay constraints for real-time traffic, whereas CBWFQ is used to ensure acceptable throughput for traffic classes that are less sensitive to delay. Simulations are developed both using a commercial product, OPNET Modeler, and also custom simulators that we developed. Our custom simulators model two different approaches to CBWFQ; and comparisons between the approaches and that of the commercial simulator are conducted. Our computational experiences (central processing unit [CPU] times for model execution and postprocessing) in using the simulators are described. This work is an important first step in the ability to model a proposed enhancement to LLQ which may be beneficial to Emergency Telecommunications Services.
Event simulation and analytic modeling are used to evaluate the performance of Low Latency Queueing (LLQ), a queueing discipline available in some Internet packet switching routers for integrated services performance assurance. LLQ combines priority queueing with Class-Based Weighted Fair Queueing (CBWFQ). Priority queueing is used to ensure satisfying tight delay constraints for real-time traffic, whereas CBWFQ is used to ensure acceptable throughput for traffic classes that are less sensitive to delay. Simulations are developed both using a commercial product, OPNET Modeler, and also custom simulators that we developed. Our custom simulators model two different approaches to CBWFQ; and comparisons between the approaches and that of the commercial simulator are conducted. Our computational experiences (central processing unit [CPU] times for model execution and postprocessing) in using the simulators are described. This work is an important first step in the ability to model a proposed enhancement to LLQ which may be beneficial to Emergency Telecommunications Services.
M.J. Fischer, D.M. Masi and D.A. Garbin, “A Discussion of Low Latency Queueing Performance Modeling,” Networking and Electronic Commerce Research Conference, Garda, Italy, October 2007.
Abstract [ - ]
Both enterprise networks and the public telecommunications infrastructure are moving to use packet switching Internet Protocol (IP) technology for efficient integrated voice, data and video services. But achieving efficiency for the enterprise requires different practices than engineering the public infrastructure. One such difference is the role of queueing disciplines to achieve different Quality of Service (QoS) objectives for different integrated services. The International Telecommunications Union (ITU), the international standards body for telecommunications, has specified three key figures of merit affecting QoS for different services [ITU 02]:
a) Delay – the end-to-end transfer delay for a packet from its availability for transfer until its completion of transfer
b) Jitter – the variation in delay
c) Loss – the probability of a packet being lost during the transfer
Delay is expressed in terms of its average. In our work we use the 99.9% quantile of packet buffer delay as the formal definition of jitter. Loss is evaluated as the percentage of packets that are lost in their transfer due to lack of queue capacity. Note that these QoS measures are end-to-end objectives, so the performance over a particular link must be less. The mapping of objectives to types of applications (voice, data, video …) varies and numerous interpretations are available, with the IETF giving one in [Bab 06]. Low Latency Queueing is a combination of Priority Queueing and Class Based Weighted Fair Queueing. The packets that require near real time service are served using Priority Queueing. Low Latency Queueing is a very complicated queue discipline to model and obtain QoS results for all the different packet types. It is further complicated by the fact that not all the arrival processes for each traffic class and their packet sizes come from the same probability distribution. This paper discusses the simulation and analytic methods we have used to solve the LLQ performance characterization problems.
Both enterprise networks and the public telecommunications infrastructure are moving to use packet switching Internet Protocol (IP) technology for efficient integrated voice, data and video services. But achieving efficiency for the enterprise requires different practices than engineering the public infrastructure. One such difference is the role of queueing disciplines to achieve different Quality of Service (QoS) objectives for different integrated services. The International Telecommunications Union (ITU), the international standards body for telecommunications, has specified three key figures of merit affecting QoS for different services [ITU 02]:
a) Delay – the end-to-end transfer delay for a packet from its availability for transfer until its completion of transfer
b) Jitter – the variation in delay
c) Loss – the probability of a packet being lost during the transfer
Delay is expressed in terms of its average. In our work we use the 99.9% quantile of packet buffer delay as the formal definition of jitter. Loss is evaluated as the percentage of packets that are lost in their transfer due to lack of queue capacity. Note that these QoS measures are end-to-end objectives, so the performance over a particular link must be less. The mapping of objectives to types of applications (voice, data, video …) varies and numerous interpretations are available, with the IETF giving one in [Bab 06]. Low Latency Queueing is a combination of Priority Queueing and Class Based Weighted Fair Queueing. The packets that require near real time service are served using Priority Queueing. Low Latency Queueing is a very complicated queue discipline to model and obtain QoS results for all the different packet types. It is further complicated by the fact that not all the arrival processes for each traffic class and their packet sizes come from the same probability distribution. This paper discusses the simulation and analytic methods we have used to solve the LLQ performance characterization problems.
Masi, D.M., M.J. Fischer, D.A. Garbin, “Modeling the Performance of Class Based Weighted Fair Queueing with OPNET and Custom Simulators", OPNETwork, August 2007, Washington DC
Abstract [ - ]
The use of OPNET Modeler and custom simulators to model routers using Class-Based Weighted Fair Queueing (CBWFQ) is investigated. The custom simulators implement two different CBWFQ scheduling approaches. Comparisons between the OPNET and the two custom simulators are conducted, in terms of the latency, packet loss, and jitter on the transmission link. Our computational experiences such as CPU times for model execution and post-processing time in using the simulators are described. This work is an important step in the development of efficient and flexible simulations of Emergency Telecommunications Services for our client, the National Communications System.
The use of OPNET Modeler and custom simulators to model routers using Class-Based Weighted Fair Queueing (CBWFQ) is investigated. The custom simulators implement two different CBWFQ scheduling approaches. Comparisons between the OPNET and the two custom simulators are conducted, in terms of the latency, packet loss, and jitter on the transmission link. Our computational experiences such as CPU times for model execution and post-processing time in using the simulators are described. This work is an important step in the development of efficient and flexible simulations of Emergency Telecommunications Services for our client, the National Communications System.
Fischer, M.J., D.A. Garbin D.M. Masi and P. McGregor, “Simulating the Performance of Low Latency Queueing for Emergency Telecommunications,” Invited talk at the International INFORMS Conference, July 2007, Puerto Rico
Abstract [ - ]
Expedited Forwarding (EF) is a telecommunications router feature designed to give certain packets performance measures that are better than the routine packets. The purpose of this talk is to report on the development and use of simulation models, both OPNET models and models developed in-house, to conduct analyses to determine the potential benefits of such features.
Expedited Forwarding (EF) is a telecommunications router feature designed to give certain packets performance measures that are better than the routine packets. The purpose of this talk is to report on the development and use of simulation models, both OPNET models and models developed in-house, to conduct analyses to determine the potential benefits of such features.
Garbin, D.A., D.M. Masi et al, “Performance Evaluation of Expedite Forward - ADMIT”, IETF conference, Prague, March 2007
Abstract [ - ]
A new Differentiated Services Code Point (DSCP), EF-ADMIT, has been recommended for a class of real-time traffic conforming to the Expedited Forwarding (EF) Per Hop Behavior (PHB) and admitted using a strong Call Admission Control (CAC) procedure incorporating capacity assurance, as compared to a class of real-time traffic conforming to the EF PHB but not subject to strong CAC. This document presents modeling results demonstrating that EF-ADMIT traffic will experience low packet drop rates even when lack of strong CAC results in EF traffic experiencing high packet drop rates. The modeling shows the performance benefit is material at low to medium network access speeds (e.g., 256 Kb/s to 1.5 Mb/s), but relatively inconsequential at high access speeds (e.g., 45 Mb/s) and, by inference, backbone speeds (100Mb/s and higher) where more bandwidth headroom is assumed. Furthermore, mixing relatively long packets (e.g., 1500 byte video packets) with relatively short packets (e.g., 200 byte voice packets) in EF PHB causes significant degradation to short packet performance at low to medium access speeds. Finally, the results show that implementation can be effective utilizing either one queue with combined EF and EF-ADMIT flows, or two queues with one forEF-ADMIT flows and one for EF flows, with the choice of approach mostly a matter of policy.
A new Differentiated Services Code Point (DSCP), EF-ADMIT, has been recommended for a class of real-time traffic conforming to the Expedited Forwarding (EF) Per Hop Behavior (PHB) and admitted using a strong Call Admission Control (CAC) procedure incorporating capacity assurance, as compared to a class of real-time traffic conforming to the EF PHB but not subject to strong CAC. This document presents modeling results demonstrating that EF-ADMIT traffic will experience low packet drop rates even when lack of strong CAC results in EF traffic experiencing high packet drop rates. The modeling shows the performance benefit is material at low to medium network access speeds (e.g., 256 Kb/s to 1.5 Mb/s), but relatively inconsequential at high access speeds (e.g., 45 Mb/s) and, by inference, backbone speeds (100Mb/s and higher) where more bandwidth headroom is assumed. Furthermore, mixing relatively long packets (e.g., 1500 byte video packets) with relatively short packets (e.g., 200 byte voice packets) in EF PHB causes significant degradation to short packet performance at low to medium access speeds. Finally, the results show that implementation can be effective utilizing either one queue with combined EF and EF-ADMIT flows, or two queues with one forEF-ADMIT flows and one for EF flows, with the choice of approach mostly a matter of policy.
Fischer, M. J. and D.M. Masi, “Analyzing Internet Packet Traces Using Lindley’s Recursion,” Winter Simulation Conference, December, 2006
Abstract [ - ]
Internet trace packet data for a given network link contains information on each packet’s arrival time and size. An important problem is to model the congestion packets experienced over the collection period. Recent research has utilized a relationship from Queueing Theory known as Lindley’s Recursion to model packet congestion. This relationship has existed for 50 years and has been quite beneficial in analyzing these traces. We report on our use of Lindley’s Recursion to analyze publicly-available link data from the Abilene Network, an Internet2 backbone network. We extend the use of Lindley's Recursion and include a discussion of the our computational problems, numerical evaluation of trace packet performance and potential modeling issues, and a statistical investigation of the independence of packet interarrival times. In addition, we show how Lindley’s Recursion can be used to extend the baseline analysis to interject Voice over Internet Protocol (VoIP) packets into the trace.
Internet trace packet data for a given network link contains information on each packet’s arrival time and size. An important problem is to model the congestion packets experienced over the collection period. Recent research has utilized a relationship from Queueing Theory known as Lindley’s Recursion to model packet congestion. This relationship has existed for 50 years and has been quite beneficial in analyzing these traces. We report on our use of Lindley’s Recursion to analyze publicly-available link data from the Abilene Network, an Internet2 backbone network. We extend the use of Lindley's Recursion and include a discussion of the our computational problems, numerical evaluation of trace packet performance and potential modeling issues, and a statistical investigation of the independence of packet interarrival times. In addition, we show how Lindley’s Recursion can be used to extend the baseline analysis to interject Voice over Internet Protocol (VoIP) packets into the trace.