(Summary of Talk at Hewlett-Packard) Amarnath Mukherjee. College of Computing. Georgia Institute of Technology. Atlanta, GA
|
|
- Rosa Ford
- 6 years ago
- Views:
Transcription
1 Trac Signatures for Network Engineering (Summary of Talk at Hewlett-Packard) Amarnath Mukherjee College of Computing Georgia Institute of Technology Atlanta, GA March 20, Introduction Trac over campus-level networks [4, 22, 29], larger multiplexing points such as NSFNET core switches [18], and variable bit-rate video applications [6, 13, 28] exhibit a slowly decaying autocorrelation function. Beyond its observed presence in measured data, there is mounting evidence [2, 3, 11, 13, 16, 21] that a slowly-decaying correlation structure is a trac property that (i) has measurable and practical impact on queueing behavior, (ii) plays a dominant role in a number of packet trac engineering problems (e.g., buer sizing, admission control, trac policing), and (iii) if ignored, typically results in overly optimistic performance predictions and inadequate network resource allocations. Protocols need to exist in this environment and need to be built for this environment. Generating representative workloads is, therefore, an important step towards eective simulations. Similarly, taking advantage of its correlation structure is essential to eective trac control algorithms. The objective of this article is to briey overview our talk, and present it in the context of related results. It is organized as follows. Section 2 presents two trac signature models: (i) a non-parametric trace-sampling model, and (ii) parametric timeseries models (seasonal ARIMA models). Section 3 presents related work in the eld. Section 4 discusses predictive control briey. Section 5 presents some concluding remarks. 2 Trac signature models We present below a couple of examples that illustrate how accurate trac can be generated using (i) trace-sampling, and (ii) a parameterized, seasonal ARIMA model (i.e., an Auto-Regressive Moving-Average model after non-random periodicities, if any, are removed through appropriate dierencing). We will observe through the examples a signicant impact on queue statistics when the arrival process is correlated as in real trac. Research supported by the National Science Foundation under grant NCR This is joint work with A. Adas (GT), S. Basu (Departmental of Statistical Sciences, Southern Methodist University) and S. M. Klivansky (GT). 1
2 2.1 Extracting application signatures with trace-sampling Trace-sampling [17] creates a desired mix of trac traces with statistical properties that are expected to match those of a proled network. The approach is fairly straight-forward and general. By way of example, we present the method we used for reproducing long-range dependence in TCP trac in the NSFNET Simulation Platform[17]. For generating TCP trac with a given proportion of dierent applications (nntp, ftp-data, smtp, etc.), we create a database for each TCP application (in our case, in S+), and store in it sampled conversations. Each sample consists of successive packet arrival times of an observed connection in a measured trace. To generate a trace with a desired trac mix of TCP applications, we sample conversations from the database with the appropriate frequency, and for simplicity, make the inter-conversation arrival times Poisson distributed. The latter is not completely true in observed trac, but it does not appear to make much dierence in packet-level trac statistics and queue statistics. An example of a real trac trace from an NSFNET CNSS and the corresponding trace-sampled trac are shown in Figure 1. A Poisson model is shown for contrast. The data has been aggregated to a level where the Poisson model `attens out' by the law of large numbers. We observe that the match between real and sampled trac is fairly good in most statistics except possibly in the small peaks and valleys in their auto-correlation functions that do not coincide at the same lag values. To see the performance accuracy of trace-sampled trac, Figure 2 shows the 95-percentile of queue length distributions from the three traces. We observe that real and trace-sampled trac are fairly close to each other. The accuracy of sampling of course depends on the representativeness of the database of application signatures. For example, if we have an extremely loud conversation in a trac trace with no parallel in the database (or even if we have it, if it is not sampled), the results will be less accurate. The method of trace-sampling appears to be general enough to apply to other network environments. One needs a way to identify addressable entities that generate trac. For example, for a connection-oriented network, one needs to store the packet arrival timestamps and associate them with a connection identier. For connectionless networks, one may use the host-pair identity and a time-out to create pseudo connections and store timestamps accordingly. For multi-cast communications, one may use a tree routed at the source to generate packet arrivals for a source. (Of course, routing and destination group selection problems will still need to be addressed.) What needs to be accurately modeled depends on the problem being studied. For example, trace-sampled trac is already congestion-controlled, and is therefore, not appropriate for studying congestion control tcplib style statistics [10] is more appropriate in this case. Hence, for the same trac data, multiple library databases are likely to be appropriate, and will be investigated. We expect to draw on our resource management work to determine the issues that are likely to be of interest for a trac library. 2.2 Parameterized trac models We consider next an example of a seasonal ARIMA based parametric model for generating sourcelevel trac, appropriate for studying media access protocols currently being considered for upstream channel access in community networks (by the IEEE standards body). The method was also found to apply to a fairly large class of aggregated Internet trac data. The attached paper [4] gives models and parameter estimates for trac over two campus FDDI rings, an Ethernet, several NSFNET external interfaces, and aggregate trac from popular TCP protocols. Thus far, we have found it to not apply only to aggregate FTP-data trac, for which the burstiness is signicantly larger than these models can capture. 2
3 Real seconds Trace-Sampled seconds Poisson seconds Series: real Rescaled Adjusted Range plot slope=1.0 slope= slope= log10(d) H = Series: sampled Rescaled Adjusted Range plot slope=1.0 slope= slope= log10(d) H = Series: pois Rescaled Adjusted Range plot slope=1.0 slope= slope= log10(d) H = log10(r/s) log10(r/s) log10(r/s) Series : real Series : sampled Series : pois Figure 1: Comparison of real trac with trace-sampled trac from the NSFNET. A Poisson generated stream is shown to provide contrast. The plots show the arrival timeseries, the R/S analysis estimates, the histogram of the arrival processes, and their autocorrelation functions (in that order). The arrival time-series is aggregated over 2-seconds, yet they are bursty (as compared to the Poisson model). 3
4 95% Queue Length Dist Real Trace-Sampled Poisson Utilization Figure 2: 95-percentile of queue length distribution versus mean utilization for real trac, tracesampled trac, and Poisson trac. Observe that the y-axis is in log scale. Therefore, capacity planning using a Poisson trac will seriously under-estimate buer-overow probabilities. The trac data we use in this example is from the GT FDDI backbone. The level of heterogeneity in sources is seen from a 100-sec interval the number of packets transmitted by the top ten sources were as follows. Source-id # Packets in (0; 100] sec Source-id # Packets in (0; 100] sec When the number of packets is aggregated over periods of 0:01-seconds, trac from Source 2 shows a period at lag 4: See it's auto-correlation function in Figure 3. An enlarged plot (not shown) with a larger number of lags on the x-axis shows another potentially period at lag 20: If this series is denoted fy t ; (t = 1; 2; )g; the dierenced sequence fy t = Y t? Y t?4 g shows a correlation at lag 4 and a smaller one at lag 20: y t can therefore, be potentially modeled as an Auto-Regressive Moving-Average (ARMA) process. Two potential models suggest themselves (see [5]): y t = a t? 4 a t?4 ; or (1) y t = a t? 4 a t?4? 20 a t?20 ; (2) where fa t g is a sequence of uncorrelated random variables and 0 s are parameters that need to be i estimated. Once these are estimated, the series fy t g can be generated using (1) or (2), and the distribution of a t : The latter is derived from the data as well see [4] (attached paper). Next fy t g may be generated using Y t = Y t?4 + y t : Some numerical problems arise with such a summation, similar to noise propagation in numerical integration. A detailed look at the problem of generating non-negative integer sequences and a solution are presented in [4]. The end-result is that for Y t > 0; 4
5 Series : Y(t) Series : Y(t) - Y(t-4) Figure 3: Left: auto-correlation function of original sequence fy t g shows a potential period at lag 4: Right: auto-correlation function of dierenced sequence fy t? Y t?4 g damps out for most lags (except lags 4 and 20), and may potentially admit to an Auto-Regressive Moving-Average (ARMA) model quantile of queue 0.95 quantile of queue length mean load points=generated, lines=trace Figure 4: 95-percentile of queue length distribution versus mean utilization for seasonal ARIMA model of Source 2. Points correspond to the generated sequence. The solid line corresponds to the real trace. it is truncated with a probability u and rounded with probability (1?u); where u is derived from the data by minimizing a weighted objective function that attempts to match the correlation structure of the generated sequence with that of the real trac data. In practice we have found it to be in the range 0:1? 0:3: If Y t < 0; however, it is set to zero (boundary condition). For our example trac source, parameter estimates were found to be ^ 4 = 0: in Eq. (1), and (^ 4 ; ^ 20 ) = (0:967855; 0:012218) in Eq. (2). u was 0:15 in both cases. The performance of the second model and that of the trace data are shown in Figure 4. It was found that Eq. (2) produced more accurate results than Eq. (1). At mean utilization levels of 0:9 or higher, however, even Eq. (2) is not satisfactory. More study is needed to determine accurate models in this regime. Based on the absolute values for the measured trac, however, systems running at this load level would not give acceptable performance. To complete the example, Figure 5 shows a section of the trac trace and its auto-correlation function, and Figure 6 shows the corresponding plots for a synthetic non-negative integer sequence using a seasonal ARIMA model with degree of dierencing equal to 4: While no agreement can be 100%, the generation does not appear to be too inaccurate. A discrepancy at lag 20 still exists, and needs more work. For the theory behind generation (and noise suppression), please see [4]. The appendix in that paper also gives a list of seasonal ARIMA models for aggregate data trac from 5
6 Section of trace data t #Packets Series : Trace Figure 5: Section of trace data (Source 2 in Section II-D), and its. Section of generated sequence t #Packets Series : Generated Figure 6: Section of corresponding generated sequence and its. FDDI, Ethernet, NSFNET External Interfaces and popular TCP protocols. 3 Related Work Properties of network trac have been the subject of considerable recent interest, see for example [8, 7, 9, 18, 22, 29, 30]. Our current approach is to model dierenced series as ARMA processes. Dierencing makes a non-stationary series stationary when there is a drift in the data. However, not all classes of non-stationarities admit a dierencing strategy. Some example datasets are given in [4]. These show quite a few non-random periodicities that are dicult to capture with seasonal models. In [4], we were able to model Bellcore's Ethernet trac data as seasonal ARIMA processes with a non-random period of approximately 80msec. This trac data has also been shown [22, 35] to possess the properties of a second-order self-similar process which is dened as a widesense stationary process (constant, time-independent, mean, variance and auto-covariance function) possessing long-memory (sum of the auto-correlations diverge), and further, the auto-correlation function is the same for all time scales over which the data is aggregated. The self-similar/longmemory model has subsequently led to insights into the burstiness properties of individual TCP connections that constitute aggregate trac [18, 29] based on processes that are known to 6
7 be second-order self-similar. For instance, the distribution of connection lifetimes and number of packets transmitted are both heavy-tailed, the latter being close to a Pareto distribution: F (x) = 1? x? ; with 1 < < 2; (which leads to a nite mean and innite variance; in practice the estimated variance can never be innite for nite datasets, but they are large). On the other hand, the models that we study are adequate and appropriate for the purposes of generating and forecasting. However, no model is true reality, and its usefulness lies in one's ability to ask and obtain answers to questions of interest from the model. Alternative models and paradigms allow for answering dierent questions on the same real phenomenon. The theoretical assumptions for the two models are dierent. Second-order self-similar models are dened for the family processes that are wide-sense stationary. The correlations for these processes extend into far lags, and by denition, the sum of their autocorrelations diverge. ARMA models for the dierenced sequences, on the other hand, result in the original sequence to be nonstationary. In practice, whether the mean, the variance and the auto-covariance are all stationary or non-stationary is often dicult to determine. Both classes of models appear to be useful, and although, theoretically, stationarity and non-stationarity are exclusive of one another, we believe that they ought to be considered inclusively in modeling repertoire (based on the results obtained from the two classes on the same datasets). In other related work on Internet trac, Clay et al studied ow-parameterization and trac characteristics of the T1 NSFNET backbone [7, 9]. Danzig et al [10] developed a trac library for TCP conversations based on empirical burst distributions they studied in [8]. Klivansky et al studied the impact of dierent factors on potential long-range dependence in aggregate TCP trac over NSFNET core switches, see [18]. Paxson developed statistical distributions for TCP burst distributions, see [30]. Paxson and Floyd reported signicant dependence among conversation arrivals for specic TCP protocols (e.g., for FTP-data and NNTP conversations), and independence among others (e.g., for TELNET and FTP-control conversations). They also showed that Danzig et al's TCP trac library produces similar variance-time plots as measured TCP trac, see [29]. Willinger et al [35] explained the presence of self-similarity in trac auto-correlation structure at multiple time-scales through a heavy-tailed distribution of source on-periods. Trac traces on variable bit-rate video have been collected and studied by several research groups, for example, [6, 12, 13, 19, 20, 28, 33, 34]. Some of the earlier studies (e.g., [33, 34]) have provided interesting data. The studies in [19, 20] model VBR video trac as TES processes [15]. The study in [20] investigates a hierarchically encoded video stream with two priorities. The studies in [6, 12, 13, 28] provide the important insight that there exists a slowly decaying auto-correlation function in many VBR video applications. TES processes [15] provide an alternate parametric method for short-memory trac with arbitrary marginal distributions. ARMA and TES models should be considered as non-exclusive alternatives for model forms. The above trac studies are an excellent resource to the networking community. However, there is still a need for a larger and broader base of applications for realistic evaluation of protocols 7
8 for quality of service guarantees. Their signatures need to be understood and used in studying resource management (e.g., in predictive controls, admission control, policing). Regarding control algorithms, networking algorithms that we are aware of do not really attempt to exploit the correlation structure in the underlying trac. Example studies are [14, 23, 24, 25, 27, 31, 32]. The studies in [23, 32] develop a controller of the class discussed under sender-initiated controls, assuming a known ARMA model. However, since trac data is not used, no attempt is made to match models with measured trac signatures. Our experience is that (i) accurate trac models are a pre-requisite for eective prediction, and apparently small model inaccuracies can lead to signicant departure in performance, and (ii) the precise denition of the problem depends on where the control is initiated (sender-initiated versus network or receiver-initiated, because of dierent information available, and dierent objectives) [1, 26]. 4 Predictive Control 4.1 Sender-initiated forecasting For trac with a slowly decaying auto-correlation function, static channel and buer allocations are likely to result in low utilizations; a dynamic allocation strategy is necessary. Studies have shown that multiplexing gains are also dependent on the coecient of variation of the number of arrivals in a xed interval [2, 13], although precise queueing results on this are not yet available. Therefore, sender-initiated forecasting for adaptive bandwidth reservation are likely to be more eective at end-points where the coecient of variation is typically larger (e.g., through adaptive leaky bucket token rates). In order to achieve a forecast, it is necessary at time n to forecast and reserve a portion of channel and switch buers for a future time-point n + j given the history until time n: In a recent paper [1], A. Adas (Ph.D. student working with AM) studies the performance of several prediction algorithms applied to six one-half-hour MPEG-I traces (made available in the public domain by Oliver Rose). One such algorithm, the Normalized Least Mean Square (NLMS) computes weights of the prediction lter adaptively, so (i) it does not require advanced knowledge of the auto-correlation structure of the arrival process, and (ii) it adapts to changing levels due to scene changes in the video stream without knowing anything of the contents. The relative buer-sizes with xed and adaptive reservations are shown in Figure 7. Note that the y-axis scales are markedly dierent. A signicant reduction in the queue length process is achieved with adaptive reservation. See also Tables 1 and 2. The forecast ^Xn+j can be chosen so as to minimize the mean squared error dened as E[( ^X n+j? X n+j ) 2 ]. However, this can lead to unacceptable performance, especially for heavytailed distributions. We investigate methods for forecasting the uth percentile (at the higher tails) of the distribution of X n+j given the past history (X 1 ; ; X n ) in [4]. This amounts to obtaining an accurate predictive distribution of X n+j and obtaining the appropriate percentiles from the predictive distribution. As an illustration of how the forecast compares with true values, consider an example trace of aggregate SMTP trace over the Georgia Tech campus FDDI backbone. We look at 15 minute segments of the data separately. Dierent percentile values used in the forecast, and corresponding observed under-forecasts are shown in Table 3. See [4] for details. Online updating of the parameter values that are used in the forecast function is important, and need to be investigated. 8
9 U Sequence 50-percentile 75-percentile 99.9-percentile 0:9 Jurassic Park Star Wars Gold Finger Terminator Talk Show Soccer :8 Jurassic Park Star Wars Gold Finger Terminator Talk Show Soccer Table 1: Queue length percentiles for dierent mean utilizations, U; when bandwidth reserved is xed. U Sequence 50-percentile 75-percentile 99.9-percentile one twelve one twelve one twelve 0:9 Jurassic Park Star Wars Gold Finger Terminator Talk Show Soccer :8 Jurassic Park Star Wars Gold Finger Terminator Talk Show Soccer Table 2: Queue length percentiles for dierent mean utilizations, U; when bandwidth reserved is dynamic. The column `one' refers to using a one-frame-ahead prediction. The column `twelve' refers to a 12-frame prediction. In general, the further away the forecast is, less its accuracy. Data Period % Under-forecasts (u) 0:5 0:8 0:9 0:95 0:99 First 15-min 41:7 19:1 11:3 5:3 1:8 Second 15-min 40:8 13:3 5:1 1:9 0:1 Third 15-min 43:7 22:8 13:6 8:0 1:6 Fourth 15-min 45:0 18:4 9:4 5:0 0:6 Table 3: Percent of times the forecasting procedure under-forecasted using the one step ahead forecast for an SMTP trace data. u is the percentile used to forecast as given in [4]. The results are based on 1000 out-of-sample forecasts. 9
10 Jurassic Park Kbits Kbits (a) (b) Jurassic Park Figure 7: Queue length process for the Jurassic Park movie (a) Fixed reservation (b) Dynamic reservation using 1-step NLMS prediction. 4.2 Network-initiated or receiver-initiated control In [26], hop-by-hop ow control of aggregate upstream trac is studied in the context of current Internet trac models. For each multiplexing point, control decision is centralized at the downstream side of transmission. Two optimization models over a nite time horizon are considered. One assumes a time-series model for the arrival process at a down-stream station; the other assumes a Brownian motion model for it. For time-series based models, equations relating (i) current and previous assigned-rates to each of the upstream switches, and (ii) corresponding departures from these switches over a time horizon, are investigated. A number of such transfer-function models are necessary, and their precise formulation is dictated by the desired optimization formulation. Model parameter estimates for example Internet trac traces are developed. The modeling method admits feedback delay. However, prediction accuracy decays with the distance between the current time and the time for which a prediction is desired. The transfer-function based approach has been applied in other engineering domains. Box et al [5] give an example of controlling the rate of reaction in a chemical plant by controlling an oxygen feed for which a transfer function between oxygen feed-rate and reaction rate is rst developed. This study needed to control one variable (reaction rate), so the model form was simple and elegant. The need to control multiple entities simultaneously requires multiple transfer functions, and makes the networking problem more challenging. This method may also have potential in receiver-initiated control, and needs to be investigated. 5 Concluding Remarks It is important to measure, characterize, and create trac libraries from emerging applications from dierent networks, each providing a window into a class of new signatures. These need to be put into a database of application signatures suitable for studying and validating resource management algorithms. Such a trac library and associated statistical models are of practical interest in the management of broadband networks. In conjunction with signature databases generated by colleagues in the community, our trac 10
11 signature library can serve as the network equivalent of standardized benchmarks that are currently available in the Database and Processor worlds. At some point in future, these could be used to study (and perhaps enable) quality of service across multiple network providers. Based on current knowledge of trac and the need to provide dependable quality of service to multi-media applications, a multi-disciplinary eort between Networking and Statistics appears to be a necessity. Our research and that of colleagues in the networking community, is addressed to meet this need. Acknowledgment This is joint work with A. Adas (GT), S. Basu (Departmental of Statistical Sciences, Southern Methodist University), and S. M. Klivansky (GT). References [1] A. Adas, \Supporting real time VBR video using dynamic reservation based on linear prediction," Proc. IEEE INFOCOM, San Francisco, March 1996 (to appear). [2] Adas, A. and A. Mukherjee, \On resource management and QoS guarantees for long range dependent trac," Proc. IEEE INFOCOM, pp , Boston, April Extended version: reports/index.94.html/git-cc [3] Adas, A. and A. Mukherjee, \Delay-jitter bound and statistical loss bound for heterogeneous correlated trac architecture and equivalent bandwidth," College of Computing Technical Report GIT-CC-96-6, Georgia Inst. of Technology. (In preparation for IEEE Journal on Selected Areas in Communications, special issue on supporting heterogeneous VBR video, due date: April 15, 1996.) reports/index.96.html/git-cc [4] Basu, S., A. Mukherjee and S. M. Klivansky, \Time-series models of Internet trac," (in review), IEEE Transactions on Networking. Preliminary version to appear in Proc. IEEE IN- FOCOM, San Francisco, March 1996 (to appear). Also College of Computing Tech Report GIT- CC-95-27, July ( reports/index.94.html/git-cc ) [5] Box, G.E.P, G.M. Jenkins, and G.C. Reinsel, Time series analysis: forecasting and control, Prentice Hall, [6] Beran, J., R. Sherman, M.S. Taqqu and W. Willinger, \Variable-bit-rate video trac and long-range dependence," accepted for publication in IEEE Trans. Networking, [7] Clay, K. C., H. W. Braun, and G. C. Polyzos, \A parameterizable methodology for Internet trac proling," IEEE Journal on Selected Areas in Communications. [8] Caceres, R., P.B.Danzig, S.Jamin and D.J.Mitzel, \Characteristics of wide-area TCP/IP conversations," Proc of ACM Sigcomm, pp , Zurich, Sept [9] Clay, K. C., G. C. Polyzos and H. W. Braun, \Trac characteristics of the T1 NSFNet backbone," Proc. IEEE Infocom, San Francisco, [10] Danzig P. B., S. Jamin, R. Caceres, D.J. Mitzel, and D. Estrin, \An Empirical Workload Model for Driving Wide-Area TCP/IP Network Simulations", Journal of Internetworking: Practice and Experience, 3; No. 1, March
12 [11] A. Erramilli, O. Narayan and W. Willinger, \Experimental queuing analysis with long-range dependent trac," Preprint, September 28, [12] Garrett, M., \Contributions toward real-time services on packet-switched networks," Ph.D. Thesis, Columbia University, [13] Garrett, M., and W. Willinger, \Analysis, modeling and generation of self-similar VBR video trac," Proc. ACM Sigcomm, pp , London, [14] Jacobson, V., \Congestion avoidance and control," Proc. of the ACM Sigcomm, pp , Aug [15] D.L. Jagerman and B. Melamed, \The Transition and Autocorrelation Structure of TES Processes Part I: General Theory", Stochastic Models 8, , [16] Jelenkovic, P. R., and A. A. Lazar, \On the Dependence of the Queue Tail Distribution on Multiple Time Scales of ATM Multiplexors," Proceedings of the 1995 Conference on Information Sciences and Systems, The Johns Hopkins University, Baltimore, MA, March 22-24, [17] Klivansky, S. M., and A. Mukherjee, \Disclosure on trac-models for performance analysis of a class of wide-area network problems." Provisional patent led with the US patent oce under the title \Trace-Sampling," serial number 60/005,112, July [18] Klivansky S., A. Mukherjee and C. Song, \On long-range dependence in NSFNET trac," 7th IEEE LAN/MAN Conference, Marathon, March Also, College of Computing Tech Report GIT-CC-94-61, December ( reports/index.94.html/git-cc ) [19] Lazar, A. A., G. Pacici and D. E. Pendarakis, \Modeling Video Sources for Real Time Scheduling," Multimedia Systems, 1, 6, , April [20] Lee, D. S., B. Melamed, A. R. Reibman and B. Sengupta, \TES modeling for analysis of a video multiplexor," Performance Evaluation, 16, 21-34, [21] Livny, M., B. Melamed and A.K. Tsiolis, \The impact of auto-correlation on queueing systems," Management Science, pp , March [22] Leland, W.E., M.S. Taqqu, W. Willinger and D.V. Wilson, \On the self-similar nature of Ethernet trac (extended version)," IEEE Trans. Networking, 2, (1), 1-15, February, [23] Mishra, P.P., and H. Kanakia \A hop-by-hop rate-based congestion control scheme," Proc. ACM Sigcomm, Baltimore, pp , August [24] Mitra, D. and J. Seery, \Dynamic adaptive windows for high speed data networks: theory and simulations," Proc. ACM Sigcomm, pp 30-37, Philadelphia, September [25] Mitra, D., \Asymptotically optimal design of congestion control for high speed data networks," IEEE Trans. Commun., 40, (2), , February [26] Mukherjee, A., and B. Sengupta, \Modeling considerations in the rate assignment of heterogeneous aggregate trac streams," (in preparation). 12
13 [27] Mukherjee, A. and J.C. Strikwerda, \Analysis of Dynamic Congestion Control Protocols A Fokker-Planck Approximation," Journal of High Speed Networks, 3, (1), 1994, (Previously appeared in Proc. ACM Sigcomm, Zurich, pp , September 1991.) [28] Pancha P., and M. El Zarki, \Variable bit rate video transmission\ IEEE Commun. Mag., Vol.32, No.5, pp , May [29] Paxson, V. and S. Floyd, \Wide-area trac: the failure of Poisson modeling," IEEE/ACM Trans. on Networking, 3, (3), , June [30] Paxson, V., \Empirically-derived analytical models of wide-area TCP connections: extended report," IEEE/ACM Transactions on Networking, 2, (4), , August [31] Ramakrishnan, K.K., and R. Jain, \A Binary Feedback Scheme for Congestion Avoidance in Computer Networks with a Connectionless Network Layer," ACM Trans. Computer Systems, vol. 8, pp , May [32] Ramamurthy, G., and B. Sengupta, \A predictive congestion control policy for high-speed wide-area networks," Proc. IEEE INFOCOM, San Francisco, [33] Sen, P., B. Maglaris, N.E. Rikli and D. Anastassiou, \Models for packet switching of variablebit-rate video sources," IEEE Journal on Selected Areas in Communications, pp , [34] Verbiest, W., and Pinnoo, L., \A Variable Bit Rate Video Codec for Asynchronous Transfer Mode Networks," IEEE Journal on Selected Areas in Communications, Vol. 7, pp , June [35] W. Willinger, M.S. Taqqu, R. Sherman, and D. Wilson, \Self-similarity through highvariability: statistical analysis of Ethernet LAN trac at the source level," Proc. ACM Sigcomm,, pp , Boston, Aug
UCLA Computer Science. Department. Off Campus Gateway. Department. UCLA FDDI Backbone. Servers. Traffic Measurement Connection
Contradictory Relationship between Hurst Parameter and Queueing Performance Ronn Ritke, Xiaoyan Hong and Mario Gerla UCLA { Computer Science Department, 45 Hilgard Ave., Los Angeles, CA 924 ritke@cs.ucla.edu,
More informationVisualization of Internet Traffic Features
Visualization of Internet Traffic Features Jiraporn Pongsiri, Mital Parikh, Miroslova Raspopovic and Kavitha Chandra Center for Advanced Computation and Telecommunications University of Massachusetts Lowell,
More informationINTERNET OVER DIGITAL VIDEO BROADCAST: PERFORMANCE ISSUES
INTERNET OVER DIGITAL VIDEO BROADCAST: PERFORMANCE ISSUES Hakan Yılmaz TÜBİTAK Marmara Research Center Information Technologies Research Institute Kocaeli, Turkey hy@btae.mam.gov.tr Bülent Sankur Boğaziçi
More information9. D. Tse, R. Gallager, and J. Tsitsiklis. Statistical multiplexing of multiple timescale
9. D. Tse, R. Gallager, and J. Tsitsiklis. Statistical multiplexing of multiple timescale markov streams. Preprint. 10. J.S. Turner. Managing Bandwidth in ATM Networks with Bursty Trac. IEEE Network Magazine,
More informationActive Queue Management for Self-Similar Network Traffic
Active Queue Management for Self-Similar Network Traffic Farnaz Amin*, Kiarash Mizanain**, and Ghasem Mirjalily*** * Electrical Engineering and computer science Department, Yazd University, farnaz.amin@stu.yazduni.ac.ir
More informationMPEG VIDEO TRAFFIC MODELS: SEQUENTIALLY MODULATED SELF-SIMILAR PROCESSES
MPEG VIDEO TRAFFIC MODELS: SEQUENTIALLY MODULATED SELF-SIMILAR PROCESSES Hai Liu, Nirwan Ansari, and Yun Q. Shi New Jersey Center for Wireless Telecommunications Department of Electrical and Computer Engineering
More informationMulticast Transport Protocol Analysis: Self-Similar Sources *
Multicast Transport Protocol Analysis: Self-Similar Sources * Mine Çağlar 1 Öznur Özkasap 2 1 Koç University, Department of Mathematics, Istanbul, Turkey 2 Koç University, Department of Computer Engineering,
More informationperform well on paths including satellite links. It is important to verify how the two ATM data services perform on satellite links. TCP is the most p
Performance of TCP/IP Using ATM ABR and UBR Services over Satellite Networks 1 Shiv Kalyanaraman, Raj Jain, Rohit Goyal, Sonia Fahmy Department of Computer and Information Science The Ohio State University
More informationCharacterization and Modeling of MPEG Video. Trac on Multiple Timescales. Nemo Semret. May 5, Abstract
Characterization and Modeling of MPEG Video Trac on Multiple Timescales Nemo Semret May 5, 995 Abstract We analyze the statistical properties of MPEG video trac based on a collection of data from real
More informationIP Traffic Characterization for Planning and Control
IP Traffic Characterization for Planning and Control V. Bolotin, J. Coombs-Reyes 1, D. Heyman, Y. Levy and D. Liu AT&T Labs, 200 Laurel Avenue, Middletown, NJ 07748, U.S.A. IP traffic modeling and engineering
More informationNetwork Model for Delay-Sensitive Traffic
Traffic Scheduling Network Model for Delay-Sensitive Traffic Source Switch Switch Destination Flow Shaper Policer (optional) Scheduler + optional shaper Policer (optional) Scheduler + optional shaper cfla.
More informationFB(9,3) Figure 1(a). A 4-by-4 Benes network. Figure 1(b). An FB(4, 2) network. Figure 2. An FB(27, 3) network
Congestion-free Routing of Streaming Multimedia Content in BMIN-based Parallel Systems Harish Sethu Department of Electrical and Computer Engineering Drexel University Philadelphia, PA 19104, USA sethu@ece.drexel.edu
More informationSIMULATING CDPD NETWORKS USING OPNET
SIMULATING CDPD NETWORKS USING OPNET Michael Jiang, Stephen Hardy, and Ljiljana Trajkovic * School of Engineering Science Simon Fraser University Vancouver, British Columbia Canada V5A 1S6 {zjiang, steve,
More informationTHE TCP specification that specifies the first original
1 Median Filtering Simulation of Bursty Traffic Auc Fai Chan, John Leis Faculty of Engineering and Surveying University of Southern Queensland Toowoomba Queensland 4350 Abstract The estimation of Retransmission
More informationMeasurement and Analysis of Traffic in a Hybrid Satellite-Terrestrial Network
Measurement and Analysis of Traffic in a Hybrid Satellite-Terrestrial Network Qing (Kenny) Shao and Ljiljana Trajkovic {qshao, ljilja}@cs.sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/cnl
More informationPerformance Characteristics of a Packet-Based Leaky-Bucket Algorithm for ATM Networks
Performance Characteristics of a Packet-Based Leaky-Bucket Algorithm for ATM Networks Toshihisa OZAWA Department of Business Administration, Komazawa University 1-23-1 Komazawa, Setagaya-ku, Tokyo 154-8525,
More informationModeling of End-to-End Available Bandwidth in Wide Area Network
2008 International Symposium on Parallel and Distributed Processing with Applications Modeling of End-to-End Available Bandwidth in Wide Area Network Wanida Putthividhya Dept. of Computer Science Thammasat
More informationA Novel Two-Step MPEG Traffic Modeling Algorithm Based on a GBAR Process
A Novel Two-Step MPEG Traffic Modeling Algorithm Based on a GBAR Process Yevgeni Koucheryavy, Dmitri Moltchanov, Jarmo Harju Institute of Communication Engineering, Tampere University of Technology, P.O.
More informationA SIMULATION STUDY OF THE IMPACT OF SWITCHING SYSTEMS ON SELF-SIMILAR PROPERTIES OF TRAFFIC. Yunkai Zhou and Harish Sethu
Proceedings of the IEEE Workshop on Statistical Signal and Array Processing Pocono Manor, Pennsylvania, USA, August 14 16, 2000 A SIMULATION STUDY OF THE IMPACT OF SWITCHING SYSTEMS ON SELF-SIMILAR PROPERTIES
More informationBuffer Management for Self-Similar Network Traffic
Buffer Management for Self-Similar Network Traffic Faranz Amin Electrical Engineering and computer science Department Yazd University Yazd, Iran farnaz.amin@stu.yazd.ac.ir Kiarash Mizanian Electrical Engineering
More informationAdaptive Methods for Distributed Video Presentation. Oregon Graduate Institute of Science and Technology. fcrispin, scen, walpole,
Adaptive Methods for Distributed Video Presentation Crispin Cowan, Shanwei Cen, Jonathan Walpole, and Calton Pu Department of Computer Science and Engineering Oregon Graduate Institute of Science and Technology
More informationDelay Analysis of Fair Queueing Algorithms with the. Stochastic Comparison Approach. Nihal Pekergin
Delay Analysis of Fair Queueing Algorithms with the Stochastic Comparison Approach Nihal Pekergin PRi SM, Universite de Versailles-St-Quentin 45 av des Etats Unis, 78 035 FRANCE CERMSEM, Universite de
More informationA New Optical Burst Switching Protocol for Supporting. Quality of Service. State University of New York at Bualo. Bualo, New York ABSTRACT
A New Optical Burst Switching Protocol for Supporting Quality of Service Myungsik Yoo y and Chunming Qiao z y Department of Electrical Engineering z Department of Computer Science and Engineering State
More informationCredit-Based Fair Queueing (CBFQ) K. T. Chan, B. Bensaou and D.H.K. Tsang. Department of Electrical & Electronic Engineering
Credit-Based Fair Queueing (CBFQ) K. T. Chan, B. Bensaou and D.H.K. Tsang Department of Electrical & Electronic Engineering Hong Kong University of Science & Technology Clear Water Bay, Kowloon, Hong Kong
More informationUCLA Computer Science. Department. Off Campus Gateway. Department. UCLA FDDI Backbone. Servers. Traffic Measurement Host
Contradictory Relationship between Hurst Parameter and Queueing Performance (extended version) Ronn Ritke y, Xiaoyan Hong and Mario Gerla UCLA { Computer Science Department, 45 Hilgard Ave., Los Angeles,
More informationNetwork Bandwidth Utilization Prediction Based on Observed SNMP Data
160 TUTA/IOE/PCU Journal of the Institute of Engineering, 2017, 13(1): 160-168 TUTA/IOE/PCU Printed in Nepal Network Bandwidth Utilization Prediction Based on Observed SNMP Data Nandalal Rana 1, Krishna
More informationMeasurement and Analysis of Traffic in a Hybrid Satellite-Terrestrial Network
Measurement and Analysis of Traffic in a Hybrid Satellite-Terrestrial Network Qing (Kenny) Shao and Ljiljana Trajkovic {qshao, ljilja}@cs.sfu.ca Communication Networks Laboratory http://www.ensc.sfu.ca/cnl
More informationSIMULATING CDPD NETWORKS USING OPNET
Michael Jiang Stephen Hardy Ljiljana Trajkovic SIMULATING CDPD NETWORKS USING OPNET TM Communication Networks Laboratory School of Engineering Science Simon Fraser University Road Map Introduction Simulation
More informationComparison of Shaping and Buffering for Video Transmission
Comparison of Shaping and Buffering for Video Transmission György Dán and Viktória Fodor Royal Institute of Technology, Department of Microelectronics and Information Technology P.O.Box Electrum 229, SE-16440
More informationReal-Time Protocol (RTP)
Real-Time Protocol (RTP) Provides standard packet format for real-time application Typically runs over UDP Specifies header fields below Payload Type: 7 bits, providing 128 possible different types of
More informationSimulation of an ATM{FDDI Gateway. Milind M. Buddhikot Sanjay Kapoor Gurudatta M. Parulkar
Simulation of an ATM{FDDI Gateway Milind M. Buddhikot Sanjay Kapoor Gurudatta M. Parulkar milind@dworkin.wustl.edu kapoor@dworkin.wustl.edu guru@flora.wustl.edu (314) 935-4203 (314) 935 4203 (314) 935-4621
More informationA PRACTICAL APPROACH FOR MULTIMEDIA TRAFFIC MODELING
A PRACTICAL APPROACH FOR MULTIMEDIA TRAFFIC MODELING Timothy D. Neame,l Moshe Zukerman 1 and Ronald G. Addie2 1 Department of Electrical and 2 Department of Mathematics Electronic Engineering, and Computer
More informationRecent measurements based on long empirical traces have revealed that many important
1 DTMW: A New Congestion Control Scheme for Long-Range Dependent Trac C. Huang a, I. Lambadaris a, M. Devetsikiotis a,p. W. Glynn b, and A. R. Kaye a a Department of Systems & Computer Engineering, Carleton
More informationOn Checkpoint Latency. Nitin H. Vaidya. In the past, a large number of researchers have analyzed. the checkpointing and rollback recovery scheme
On Checkpoint Latency Nitin H. Vaidya Department of Computer Science Texas A&M University College Station, TX 77843-3112 E-mail: vaidya@cs.tamu.edu Web: http://www.cs.tamu.edu/faculty/vaidya/ Abstract
More informationPerformance Analysis and Trac Behavior of the Xphone. Videoconferencing Application System on an Ethernet. B. K. Ryu and H. E.
Performance Analysis and Trac Behavior of the Xphone Videoconferencing Application System on an Ethernet B. K. Ryu and H. E. Meadows Department of Electrical Engineering and Center for Telecommunications
More informationExtensions to RTP to support Mobile Networking: Brown, Singh 2 within the cell. In our proposed architecture [3], we add a third level to this hierarc
Extensions to RTP to support Mobile Networking Kevin Brown Suresh Singh Department of Computer Science Department of Computer Science University of South Carolina Department of South Carolina Columbia,
More informationIP Traffic Prediction and Equivalent Bandwidth for DAMA TDMA Protocols
IP Traffic and Equivalent Bandwidth for DAMA TDMA Protocols J. Aracil, D. Morato, E. Magaña, M. Izal Universidad Pública de Navarra, 316 Pamplona, SPAIN email:javier.aracil@unavarra.es Abstract The use
More informationSIMULATION OF PACKET DATA NETWORKS USING OPNET
SIMULATION OF PACKET DATA NETWORKS USING OPNET Nazy Alborz, Maryam Keyvani, Milan Nikolic, and Ljiljana Trajkovic * School of Engineering Science Simon Fraser University Vancouver, British Columbia, Canada
More informationAbstract Studying network protocols and distributed applications in real networks can be dicult due to the need for complex topologies, hard to nd phy
ONE: The Ohio Network Emulator Mark Allman, Adam Caldwell, Shawn Ostermann mallman@lerc.nasa.gov, adam@eni.net ostermann@cs.ohiou.edu School of Electrical Engineering and Computer Science Ohio University
More informationLoss Performance Analysis for Heterogeneous
IEEE/ACM TRANSACTIONS ON NETWORKING, VOL. 10, NO. 1, FEBRUARY 2002 125 Loss Performance Analysis for Heterogeneous ON OFF Sources With Application to Connection Admission Control Guoqiang Mao, Student
More informationAdvanced Internet Technologies
Advanced Internet Technologies Chapter 3 Performance Modeling Dr.-Ing. Falko Dressler Chair for Computer Networks & Internet Wilhelm-Schickard-Institute for Computer Science University of Tübingen http://net.informatik.uni-tuebingen.de/
More informationRESOURCE DIMENSIONING IN A GENERAL
RESOURCE DIMENSIONING IN A GENERAL PACKET-SWITCHED NETWORK ENVIRONMENT S J E Roon and W T Penzhorn Department of Electrical, Electronic and Computer Engineering, University of Pretoria, 0002 Pretoria,
More informationConnection Admission Control for Hard Real-Time Communication in ATM Networks
MITSUBISHI ELECTRIC RESEARCH LABORATORIES http://www.merl.com Connection Admission Control for Hard Real-Time Communication in ATM Networks Qin Zheng, Tetsuya Yokotani, Tatsuki Ichihashi, Yasunoni Nemoto
More informationSIMULATION OF PACKET DATA NETWORKS USING OPNET
Nazy Alborz Maryam Keyvani Milan Nikolic Ljiljana Trajkovic SIMULATION OF PACKET DATA NETWORKS USING OPNET TM Communication Networks Laboratory School of Engineering Science Simon Fraser University Road
More informationOn the Relationship of Server Disk Workloads and Client File Requests
On the Relationship of Server Workloads and Client File Requests John R. Heath Department of Computer Science University of Southern Maine Portland, Maine 43 Stephen A.R. Houser University Computing Technologies
More informationA Study onthe Cause of Long- Range Dependence Observed in Empirical TCP Traffic Traces
The University of Kansas Technical Report A Study onthe Cause of Long- Range Dependence Observed in Empirical TCP Traffic Traces Georgios Lazarou and Victor Frost ITTC-FY2000-TR-10980-28 July 1999 Project
More informationReplicate It! Scalable Content Delivery: Why? Scalable Content Delivery: How? Scalable Content Delivery: How? Scalable Content Delivery: What?
Accelerating Internet Streaming Media Delivery using Azer Bestavros and Shudong Jin Boston University http://www.cs.bu.edu/groups/wing Scalable Content Delivery: Why? Need to manage resource usage as demand
More informationCHARACTERIZING THE END-TO-END BEHAVIOR OF THE INTERNET: MEASUREMENTS, ANALYSIS, AND APPLICATIONS Jean-Chrysostome Bolot INRIA B. P. 93 06902 Sophia-Antipolis Cedex France bolot@sophia.inria.fr ABSTRACT
More informationSelf Similar Network Traffic present by Carl Minton. Definition of Self Similarity
Self Similar Network Traffic present by Carl Minton Agenda Definition of self similarity Quantifying self similarity Self similarity of network traffic Implications for network performance Pointers for
More information136 Proceedings of the rd International Teletraffic Congress (ITC 2011)
Gaussian Approximation of CDN Call Level Traffic Andrzej Bak and Piotr Gajowniczek Institute of Telecommunications Warsaw University of Technology Nowowiejska 5/9, -665 Warsaw, Poland Email: bak@tele.pw.edu.pl
More informationMaximizing the Number of Users in an Interactive Video-on-Demand System
IEEE TRANSACTIONS ON BROADCASTING, VOL. 48, NO. 4, DECEMBER 2002 281 Maximizing the Number of Users in an Interactive Video-on-Demand System Spiridon Bakiras, Member, IEEE and Victor O. K. Li, Fellow,
More informationCongestion Propagation among Routers in the Internet
Congestion Propagation among Routers in the Internet Kouhei Sugiyama, Hiroyuki Ohsaki and Makoto Imase Graduate School of Information Science and Technology, Osaka University -, Yamadaoka, Suita, Osaka,
More information\Classical" RSVP and IP over ATM. Steven Berson. April 10, Abstract
\Classical" RSVP and IP over ATM Steven Berson USC Information Sciences Institute April 10, 1996 Abstract Integrated Services in the Internet is rapidly becoming a reality. Meanwhile, ATM technology is
More informationCharacterizing Internet Load as a Non-regular Multiplex of TCP Streams
Characterizing Internet Load as a Non-regular Multiplex of TCP Streams J. Aracil, D. Morató Dpto. Automática y Computación Universidad Pública de Navarra {javier.aracil,daniel.morato}@unavarra.es http://www.tlm.unavarra.es
More informationINFORMATION DYNAMICS APPLIED TO LINK-STATE ROUTING 1
InfoDyn Routing, October 25, 2 INFORMATION DYNAMICS APPLIED TO LINK-STATE ROUTING Abstract Hyeonsang Eom, Ashok K. Agrawala, Sam H. Noh, A. Udaya Shankar ({hseom, agrawala, noh, shankar}@cs.umd.edu) Computer
More informationSource 1. Destination 1. Bottleneck Link. Destination 2. Source 2. Destination N. Source N
WORST CASE BUFFER REQUIREMENTS FOR TCP OVER ABR a B. Vandalore, S. Kalyanaraman b, R. Jain, R. Goyal, S. Fahmy Dept. of Computer and Information Science, The Ohio State University, 2015 Neil Ave, Columbus,
More informationReferences. [7] J. G. Gruber. Delay related issues in integrated voice and data networks. In IEEE Trans. on Commun., pages , June 1981.
have adopted the leaky bucket mechanism to satisfy the application required quality of service parameters. The basic performance metrics such as the delay, delay jitter, and system utilization are evaluated
More informationModule objectives. Integrated services. Support for real-time applications. Real-time flows and the current Internet protocols
Integrated services Reading: S. Keshav, An Engineering Approach to Computer Networking, chapters 6, 9 and 4 Module objectives Learn and understand about: Support for real-time applications: network-layer
More informationAdvanced Computer Networks
Advanced Computer Networks QoS in IP networks Prof. Andrzej Duda duda@imag.fr Contents QoS principles Traffic shaping leaky bucket token bucket Scheduling FIFO Fair queueing RED IntServ DiffServ http://duda.imag.fr
More information2 Overview of MPEG-2 over ATM In this section, we give a quick introduction to the MPEG-2 over ATM model and introduce some MPEG-2 terminology. For a
Performance of TCP over ABR with Long-Range Dependent VBR Background Trac over Terrestrial and Satellite ATM networks 1 Shivkumar Kalyanaraman 3, Bobby Vandalore, Raj Jain, Rohit Goyal, Sonia Fahmy The
More informationQuality Differentiation with Source Shaping and Forward Error Correction
Quality Differentiation with Source Shaping and Forward Error Correction György Dán and Viktória Fodor KTH, Royal Institute of Technology, Department of Microelectronics and Information Technology, {gyuri,viktoria}@imit.kth.se
More informationUNIT 2 TRANSPORT LAYER
Network, Transport and Application UNIT 2 TRANSPORT LAYER Structure Page No. 2.0 Introduction 34 2.1 Objective 34 2.2 Addressing 35 2.3 Reliable delivery 35 2.4 Flow control 38 2.5 Connection Management
More informationUtilizing Neural Networks to Reduce Packet Loss in Self-Similar Teletraffic Patterns
Utilizing Neural Networks to Reduce Packet Loss in Self-Similar Teletraffic Patterns Homayoun Yousefi zadeh; EECS Dept; UC, Irvine Edmond A. Jonckheere; EE-Systems Dept; USC John A. Silvester; EE-Systems
More informationAnalysis of Random Access Protocol under Bursty Traffic
Analysis of Random Access Protocol under Bursty Traffic Jianbo Gao and Izhak Rubin Electrical Engineering Department, University of California, Los Angeles, CA 90095 {jbgao, rubin}@ee.ucla.edu Abstract.
More informationMultimedia Networks. University of Virginia. Charlottesville, VA Cornwallis Road
To appear: ACM/Springer Multimedia Systems Journal. Trac Characterization Algorithms for VBR Video in Multimedia Networks Jorg Liebeherr Dallas E. Wrege y Department of Computer Science University of Virginia
More informationImproving TCP Performance over Wireless Networks using Loss Predictors
Improving TCP Performance over Wireless Networks using Loss Predictors Fabio Martignon Dipartimento Elettronica e Informazione Politecnico di Milano P.zza L. Da Vinci 32, 20133 Milano Email: martignon@elet.polimi.it
More informationQOS IN PACKET NETWORKS
QOS IN PACKET NETWORKS THE KLUWER INTERNATIONAL SERIES IN ENGINEERING AND COMPUTER SCIENCE QOS IN PACKET NETWORKS by Kun I. Park, Ph.D. The MITRE Corporation USA Springer ebook ISBN: 0-387-23390-3 Print
More informationWeek 7: Traffic Models and QoS
Week 7: Traffic Models and QoS Acknowledgement: Some slides are adapted from Computer Networking: A Top Down Approach Featuring the Internet, 2 nd edition, J.F Kurose and K.W. Ross All Rights Reserved,
More information3.3. Traffic models and teletraffic dimensioning
3.3. Traffic models and teletraffic dimensioning Sándor Molnár, author Béla Frajka: reviewer 3.3.1. Introduction The basic teletraffic principles, equations and an overview of the nature of network traffic
More informationOn Checkpoint Latency. Nitin H. Vaidya. Texas A&M University. Phone: (409) Technical Report
On Checkpoint Latency Nitin H. Vaidya Department of Computer Science Texas A&M University College Station, TX 77843-3112 E-mail: vaidya@cs.tamu.edu Phone: (409) 845-0512 FAX: (409) 847-8578 Technical Report
More informationPredicting bandwidth requirements of ATM and Ethernet trac. February 29, Abstract
Predicting bandwidth requirements of ATM and Ethernet trac Simon Crosby 1, Ian Leslie 1, Meriel Huggard 2;3, J.T. Lewis 2, Brian McGurk 2, and Raymond Russell 2 February 29, 1996 Abstract In this paper,
More informationRED behavior with different packet sizes
RED behavior with different packet sizes Stefaan De Cnodder, Omar Elloumi *, Kenny Pauwels Traffic and Routing Technologies project Alcatel Corporate Research Center, Francis Wellesplein, 1-18 Antwerp,
More informationMultimedia Applications Require Adaptive CPU Scheduling. Veronica Baiceanu, Crispin Cowan, Dylan McNamee, Calton Pu, and Jonathan Walpole
Multimedia Applications Require Adaptive CPU Scheduling Veronica Baiceanu, Crispin Cowan, Dylan McNamee, Calton Pu, and Jonathan Walpole Department of Computer Science and Engineering Oregon Graduate Institute
More informationA Large Scale Simulation Study: Impact of Unresponsive Malicious Flows
A Large Scale Simulation Study: Impact of Unresponsive Malicious Flows Yen-Hung Hu, Debra Tang, Hyeong-Ah Choi 3 Abstract Researches have unveiled that about % of current Internet traffic is contributed
More informationDynamic Multi-Path Communication for Video Trac. Hao-hua Chu, Klara Nahrstedt. Department of Computer Science. University of Illinois
Dynamic Multi-Path Communication for Video Trac Hao-hua Chu, Klara Nahrstedt Department of Computer Science University of Illinois h-chu3@cs.uiuc.edu, klara@cs.uiuc.edu Abstract Video-on-Demand applications
More informationComputer Networks 1 (Mạng Máy Tính 1) Lectured by: Dr. Phạm Trần Vũ
Computer Networks 1 (Mạng Máy Tính 1) Lectured by: Dr. Phạm Trần Vũ 1 Lecture 5: Network Layer (cont ) Reference: Chapter 5 - Computer Networks, Andrew S. Tanenbaum, 4th Edition, Prentice Hall, 2003. 2
More informationDelayed reservation decision in optical burst switching networks with optical buffers
Delayed reservation decision in optical burst switching networks with optical buffers G.M. Li *, Victor O.K. Li + *School of Information Engineering SHANDONG University at WEIHAI, China + Department of
More informationUncontrollable. High Priority. Users. Multiplexer. Server. Low Priority. Controllable. Users. Queue
Global Max-Min Fairness Guarantee for ABR Flow Control Qingyang Hu, David W. Petr Information and Telecommunication Technology Center Department of Electrical Engineering & Computer Science The University
More informationStability in ATM Networks. network.
Stability in ATM Networks. Chengzhi Li, Amitava Raha y, and Wei Zhao Abstract In this paper, we address the issues of stability in ATM networks. A network is stable if and only if all the packets have
More informationFrank Miller, George Apostolopoulos, and Satish Tripathi. University of Maryland. College Park, MD ffwmiller, georgeap,
Simple Input/Output Streaming in the Operating System Frank Miller, George Apostolopoulos, and Satish Tripathi Mobile Computing and Multimedia Laboratory Department of Computer Science University of Maryland
More informationSIMULATION BASED ANALYSIS OF THE INTERACTION OF END-TO-END AND HOP-BY-HOP FLOW CONTROL SCHEMES IN PACKET SWITCHING LANS
SIMULATION BASED ANALYSIS OF THE INTERACTION OF END-TO-END AND HOP-BY-HOP FLOW CONTROL SCHEMES IN PACKET SWITCHING LANS J Wechta, A Eberlein, F Halsall and M Spratt Abstract To meet the networking requirements
More informationCharacterization of Video Trac 21. [14] H. Kanakia, P.P. Mishra, and A Reibman. An adaptive congestion control scheme for real-time packet
Characterization of Video Trac 21 [14] H. Kanakia, P.P. Mishra, and A Reibman. An adaptive congestion control scheme for real-time packet video transport. Computer Communication Review, 23:20{31, October
More information0 Source. Destination. Destination. 16 Destination Destination 3. (c) (b) (a)
In Proc. Infocom '95, th Annual Joint Conference of the IEEE Computer and Communications Societies, Boston, Mass. (April 995), pp. -9. A Model for Virtual Tree Bandwidth Allocation in ATM Networks Adarshpal
More informationEnd-to-End Mechanisms for QoS Support in Wireless Networks
End-to-End Mechanisms for QoS Support in Wireless Networks R VS Torsten Braun joint work with Matthias Scheidegger, Marco Studer, Ruy de Oliveira Computer Networks and Distributed Systems Institute of
More informationQuality of Service (QoS)
Quality of Service (QoS) A note on the use of these ppt slides: We re making these slides freely available to all (faculty, students, readers). They re in PowerPoint form so you can add, modify, and delete
More informationUnit 2 Packet Switching Networks - II
Unit 2 Packet Switching Networks - II Dijkstra Algorithm: Finding shortest path Algorithm for finding shortest paths N: set of nodes for which shortest path already found Initialization: (Start with source
More informationWebTraff: A GUI for Web Proxy Cache Workload Modeling and Analysis
WebTraff: A GUI for Web Proxy Cache Workload Modeling and Analysis Nayden Markatchev Carey Williamson Department of Computer Science University of Calgary E-mail: {nayden,carey}@cpsc.ucalgary.ca Abstract
More informationMohammad Hossein Manshaei 1393
Mohammad Hossein Manshaei manshaei@gmail.com 1393 Voice and Video over IP Slides derived from those available on the Web site of the book Computer Networking, by Kurose and Ross, PEARSON 2 Multimedia networking:
More informationA Measurement-Based CAC Strategy for ATM Networks
A Measurement-Based CAC Strategy for ATM Networks Sponsor: Sprint Corporation Kunyan Liu David W. Petr Technical Report TISL-11230-01 Telecommunications & Information Sciences Laboratory Department of
More informationVoice and Data Session Capacity over Local Area Networks
Voice and ata Session Capacity over Local Area Networks Theresa A. Fry, Ikhlaq Sidhu, Guido Schuster, and Jerry Mahler epartment of Electrical and Computer Engineering Northwestern University, Evanston,
More informationModeling Bit Rate Variations in MPEG Sources. University of Maryland at College Park. College Park, MD
Modeling Bit Rate Variations in MPEG Sources Marwan Krunz y and Satish K. Tripathi z y University of Maryland Institute for Advanced Computer Studies z Department of Computer Science University of Maryland
More informationNetwork Layer Enhancements
Network Layer Enhancements EECS 122: Lecture 14 Department of Electrical Engineering and Computer Sciences University of California Berkeley Today We have studied the network layer mechanisms that enable
More informationResource allocation in networks. Resource Allocation in Networks. Resource allocation
Resource allocation in networks Resource Allocation in Networks Very much like a resource allocation problem in operating systems How is it different? Resources and jobs are different Resources are buffers
More informationReduction of Periodic Broadcast Resource Requirements with Proxy Caching
Reduction of Periodic Broadcast Resource Requirements with Proxy Caching Ewa Kusmierek and David H.C. Du Digital Technology Center and Department of Computer Science and Engineering University of Minnesota
More informationToward a Time-Scale Based Framework for ABR Trac Management. Tamer Dag Ioannis Stavrakakis. 409 Dana Research Building, 360 Huntington Avenue
Toward a Time-Scale Based Framework for ABR Trac Management Tamer Dag Ioannis Stavrakakis Electrical and Computer Engineering Department 409 Dana Research Building, 360 Huntington Avenue Northeastern University,
More informationOn the relationship between file sizes, transport protocols, and self-similar network traffic
To appear in Proc. of the Fourth International Conference on Network Protocols (ICNP 96), October 996 On the relationship between file sizes, transport protocols, and self-similar network traffic Kihong
More informationEmpirical Models of TCP and UDP End User Network Traffic from Data Analysis
Empirical Models of TCP and UDP End User Network Traffic from NETI@home Data Analysis Charles R. Simpson, Jr., Dheeraj Reddy, George F. Riley School of Electrical and Computer Engineering Georgia Institute
More informationVideo Streaming Over the Internet
Video Streaming Over the Internet 1. Research Team Project Leader: Graduate Students: Prof. Leana Golubchik, Computer Science Department Bassem Abdouni, Adam W.-J. Lee 2. Statement of Project Goals Quality
More informationInternet Services & Protocols. Quality of Service Architecture
Department of Computer Science Institute for System Architecture, Chair for Computer Networks Internet Services & Protocols Quality of Service Architecture Dr.-Ing. Stephan Groß Room: INF 3099 E-Mail:
More informationAnalytic Performance Models for Bounded Queueing Systems
Analytic Performance Models for Bounded Queueing Systems Praveen Krishnamurthy Roger D. Chamberlain Praveen Krishnamurthy and Roger D. Chamberlain, Analytic Performance Models for Bounded Queueing Systems,
More information