V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generat ing self-similar network trac. Computer Communication Review, 27:5-18, October 1997.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....trace data if the trace model is to be used for all resources. All the simulation results in this paper use the extraneous job model for servers (with Poisson inter arrival time) and the trace model for network links. We took advantage of Paxson s high quality self similar network traffic traces [29] for all network simulations. This method makes it possible to generate a variety of realistic network traffic with a relatively small set of parameters as input. We did not have access to a wide variety of high quality traces for modeling server load. Therefore, we opted for the extraneous job ....

V. Paxson. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Traffic. Computer Communication Review, 27(5):5--18, 1997. 7

....be done by means a simple transformation if the process is Gaussian. So, we have decided to use FGN time series in our simulations because, in addition to its simplicity to exhibit LRD, it is a Gaussian process. In order to generate FGN, we will make use of the fast approximated method by Paxson [26], based on FFTs. 3 FDDI MAC protocol Fiber Distributed Data Interface (FDDI) is a token ring network that uses optic fiber as transmission medium. The Medium Access Control (MAC) protocol of FDDI is a timed token protocol and includes provisions for synchronous traffic and multiple asynchronous ....

....simulate is a symmetrical network with 10 nodes, where each node sends data only to the next node in the ring. The traffic patterns used by every node are identical. In each one, frame sizes are taken from a file that contains samples generated, independently of the rest, with the Paxson s method [26] using H = 0.9 (the estimated value of H for the resulting series was into the range [0.87 0.92] The marginal distribution of these samples is changed to get uniform and lognormal distributions. In order to vary the total offered load to the network, we vary only the mean interarrival time. To ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Computer Communication Review, 27(5):5--18, October 1997. Available at ftp://ftp.ee.lbl.gov/papers/fast-approx-selfsim.ps.

....model for network dimensioning and control. First, the FBT has been shown to model accurately the Internet traffic coming from the multiplex of a large number of users [5] Secondly, a variety of methods exist to generate synthetic traces of FBTs, that are useful for simulation purposes [6][7]. Third, the single server single queue model has been extensively analyzed and there is a closed analytical expression for the queue distribution tail, which is given by: c m)2Hq 2 2H P(Qegt; q) e 2K( 2am (1) Where c is the server capacity and and K(H) H(1 H) The above formula ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Computer Communications Review, 27(5), 1997.

....model for network dimensioning and control. First, the FBT has been shown to model accurately the Internet trac coming from the multiplex of a large number of users [5] Secondly, a variety of methods exist to generate synthetic traces of FBTs, that are useful for simulation purposes [6][7]. Third, the single server single queue model has been extensively analyzed and there is a closed analytical expression for the queue distribution tail, which is given by: P (Q length q) e (c m) 2H q 2 2H 2K(H) 2 am (1) Where c is the server capacity and and K(H) H (1 H) 1 H The ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network trac. Computer Communications Review, 27(5), 1997.

....an adjustable amount of self similarity (as represented by the Hurst parameters) allows testing against a traffic type that does not have the same multiplexing characteristics as the 2 state Markov model, yet can be easily compared against simulations. Such an implementation of a Hursty source is [29]. Direct video list conversion This technique is intended to incorporate the off line generation of video traces into the physical generator itself (Section 2.3) This would be desirable as it allows the much smaller sized lists of video frame 66 sizes to be kept in memory and converted into ....

Vern Paxson. Fast, approximate synthesis of fractional gaussian noise for generating seldsimilar network traffic. ACM SIGCOMM Computer Communications Review, 25(5):5--18, Oct 1997.

....non stationarity of the self similar component. Matching the behavior of a real trace will hence require some adjustments in the synthetic trace non stationarity in order to keep close to the average evolution of the real process to be matched. Other methods available in the literature like [Pax97] LL00] or [RL95] for example, allow to generate a better (matching a particular value of H) self similar process than ours, but we leave this aspect for further study. 3D LD for seasonality and self similar components 5 10 15 20 25 Time (unit = 110 min) 0 2 4 6 8 10 12 14 16 Octave ....

V. Paxson. Fast, approximate synthesis of fractional Gaussian noise for generating self-similar network traffic. Computer Communication Review, vol. 27, pages 5--18, October 1997.

....length. The time complexity for RMD is only O(n) where n is the length of the FBM process. Note that RMD generates a somewhat inaccurate self similar process and that the resulting Hurst parameter may be slightly smaller than the target value. Other methods such as the fast Fourier Transform [31] can be implemented and used to replace this module in Gismo. VBR Marginal Distribution: To model the high variability of streaming media bit rates, we use a heavy tailed marginal distribution to characterize the bit rate. A heavy tailed distribution is one whose upper tail declines like a power ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generat ing self-similar network trac. Computer Communication Review, 27:5-18, October 1997.

....onto the network between S1 and D1, and S2 and D2, resulting in two congested links competing for bandwidth with the application. The competing traffic was generated using fractional Gaussian noise, a method described by Paxson for representing realistic aggregate traffic encountered on networks [10]. The average rate of competing traffic on each link was chosen between 0Mbps and 100Mbps (link capacity) and changed an average of every 10 minutes. 4.2 Experimental results Over 65,000 observations were taken during the experiments. To determine the accuracy of the two prediction techniques, ....

....were developed for the Remos system [8] Recent network research has focused on modeling the self similarity in network traffic, and these models may lead to more realistic traffic than Poisson processes. We used fractional Gaussian noise to generate self similar traffic for our experiments [10]. More realistic wavelet models have been investigated more recently [13] but we have not yet rerun our experiments with the new models. 8 Conclusions Tools for discovering network performance and topology are an important part of the support infrastructure for distributed computing. This paper ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Technical Report LBL-36750, Lawrence Berkeley National Laboratory, April 1995.

....onto the network between S1 and D1, and S2 and D2, resulting in two congested links competing for bandwidth with the application. The competing traffic was generated using fractional Gaussian noise, a method described by Paxson for representing realistic aggregate traffic encountered on networks [10]. The average rate of competing traffic on each link was chosen between 0Mbps and 100Mbps (link capacity) and changed an average of every 10 minutes. 4.2. Experimental results Over 65,000 observations were taken during the experiments. To determine the accuracy of the two prediction techniques, ....

....were developed for the Remos system [8] Recent network research has focused on modeling the self similarity in network traffic, and these models may lead to more realistic traffic than Poisson processes. We used fractional Gaussian noise to generate self similar traffic for our experiments [10]. More realistic wavelet models have been investigated more recently [13] but we have not yet rerun our experiments with the new models. 8. Conclusions Tools for discovering network performance and topology are an important part of the support infrastructure for distributed computing. This ....

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Technical Report LBL-36750, Lawrence Berkeley National Laboratory, April 1995.

....for short sequences. They are usually inappropriate for generating long sequences because they require multiple passes along generated sequences. To overcome this, approximate methods for generation of self similar sequences in simulation studies of computer networks have been proposed [14] [22]. Contrary to Haar wavelets used in generators proposed in [1] and [12] our generator uses Daubechies wavelets (DW) Our evaluation of the method proposed for generating self similar sequences concentrates on two aspects: i) how accurately a self similar process can be generated, and (ii) how ....

....self similar with H =1 (# 2) if # (m) k # # k ,asm ##. The most frequently studied models of self similar tra#c belong either to the class of fractional autoregressive integrated moving average (F ARIMA) processes or to the class of fractional Gaussian noise processes; see [10] 15] [22]. F ARIMA(p, d, q) processes were introduced by Hosking [10] who showed that they are asymptotically self similar with Hurst parameter H = d 1 2 ,aslongas0 d 1 2 .For the second class, the FGN process is the incremental process Y k = X k X k 1 ,k # 0, where X k ....

[Article contains additional citation context not shown here]

Paxson, V. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Tra#c. ACM SIGCOMM, Computer Communication Review 27, 5 (1997), 5--18.

No context found.

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generat ing self-similar network trac. Computer Communication Review, 27:5-18, October 1997.

No context found.

Paxson V. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Comput Commun Rev Oct. 1997;27(5):5e18.

No context found.

Vern Paxson. Fast, approximate synthesis of fractional Gaussian noise for generating self-similar network tra#c. ACM SIGCOMM Computer communication review, 27(5):5--18, October 1997.

No context found.

V. Paxson. Fast, approximate synthesis of Fractional Gaussian Noise for generating self-similar network tra#c, Computer Communication Review, vol. 27, no. 5, pp. 5--18.

No context found.

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network tra#c. Computer Comm. Rev., 27:5--18, 1997.

No context found.

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network traffic. Computer Communications Review 27(5):5--18, October 1997.

No context found.

Vern Paxson. Fast, approximate synthesis of fractional Gaussian noise for generating self-similar network tra#c. Computer Communications Review, 27(5):5--18, 1997.

No context found.

V. Paxson. Fast, approximate synthesis of fractional gaussian noise for generating self-similar network tra#c. In Computer Communication Review, volume 27, pages 5--18. ACM SIGCOMM, October 1997.

No context found.

V. Paxson. Fast, approximate synthesis of fractional Gaussian noise for generating self-similar network traffic, Computer Communication Review, 27(5):5-18, Oct. 1997.

No context found.

PAXSON, V. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Traffic. Computer Communication Review 27, 5 (October 1997), 5--18.

No context found.

Vern Paxson. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Traffic. Computer Communications Review, 27(5):5--18, October 1997. (pp 38, 39)

No context found.

Paxson, V. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating SelfSimilar Network Traffic Computer Communication Review, Vol. 27, No. 5, October 1997, pp. 5-18.

No context found.

V. Paxson, \Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Trac". Computer Communications Review, V. 27 No. 5, pp.5-18, October 1997.

No context found.

V. Paxson. \Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-similar Network Trac," Computer Communication Review, vol. 27, no. 5, pp. 5-18, 1997.

No context found.

PAXSON, V. Fast, Approximate Synthesis of Fractional Gaussian Noise for Generating Self-Similar Network Traffic. ACM Computer Communication Review, Vol. 27, No. 5, pp. 5--18. October 1997.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC