N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and larger [10, 11] Though fBm is an appropriate traffic model in some cases [3, 10, 11] it can only model real world traces with a rigid, restrictive correlation structure. Indeed, the importance of shortterm correlations has been convincingly demonstrated for queuing in finite length buffers [12 14], and so called critical time scales that affect queuing have been discovered [14 16] Wavelet based multiscale models provide generalizations of fBm with more flexible correlation structures [17 20] Using efficient multiscale tree structures, these models provide fast O(N) synthesis ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, vol. 7, pp. 267--280, 1997.

....and larger [10, 11] Though fBm is an appropriate traffic model in some cases [3, 10, 11] it can only model real world traces with a rigid, restrictive correlation structure. Indeed, the importance of shortterm correlations has been convincingly demonstrated for queuing in finite length buffers [12 14], and so called critical time scales that affect queuing have been discovered [14 16] Wavelet based multiscale models provide generalizations of fBm with more flexible correlation structures [17 20] Using efficient multiscale tree structures, these models provide fast O(N) synthesis ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, vol. 7, pp. 267--280, 1997.

.... Though fGn is an appropriate traffic model in some cases [11, 12] it cannot model realworld traces with a correlation structure which is poorly matched by the rigid, restrictive (2) Indeed, convincing evidence has been produced establishing the importance of shortterm correlations for buffering [13 15] and so called relevant time scales have been discovered [14, 16, 17] Generalizations of fBm fGn with a more flexible correlation structure than (2) can be synthesized almost effortlessly using the amazing decorrelating capability of the wavelet transform [18 21] There, independent Gaussian ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....of the fBm fGn models. Though fGn is an appropriate traffic model in some cases [11, 12] it can only model real world traces with the rigid, restrictive correlation structure (1) Indeed, convincing evidence has been produced establishing the importance of short term correlations for buffering [13 15], and so called relevant time scales have been discovered [14, 16, 17] Generalizations of fBm fGn with a more flexible correlation structure than (1) can be synthesized almost effortlessly using the powerful decorrelating capability of the wavelet transform [18 21] There, independent Gaussian ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....and are at best merely asymptotically self similar. In other words, the single parameter H is not sufficient to capture the complicated correlation structure of real network processes. Indeed, convincing evidence has been produced establishing the importance of shortterm correlations for buffering [16 18] and so called relevant time scales have been discovered [19] The wavelet domain independent Gaussian (WIG) model generalizes fBm fGn by allowing a more flexible scaling relation than (1) By matching both long and short term correlations, the WIG model more completely matches the correlation ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....services, can realize statistical multiplexing gains and support a larger number of flows from the same amount of resource. Various analytical and empirical studies have shown that this statistical multiplexing gain can be significant, even if the traffic exhibits long range dependence [18 21]. The scalability of a QoS architecture is a function of the amount of state maintained and exchanged between the various QoS components (e.g. reservation manager, admission controller, traffic shaper, classifier, scheduler. For example, the IETF intserv framework [22] provides per flow service ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, 7 (1997) 1-3, pp. 267-280

....are at best merely asymptotically self similar. In other words, the single parameter H is not sufficient to capture the complicated correlation structure of real network processes. Indeed, convincing evidence has been produced establishing the importance of short term correlations for buffering [16 18] and so called relevant time scales have been discovered [19] The wavelet domain independent Gaussian (WIG) model generalizes fBm fGn by allowing a more flexible scaling relation than (1) By matching both long and short term correlations, the WIG model more completely matches the correlation ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....of the fBm fGn models. Though fGn is an appropriate traffic model in some cases [11, 12] it can only model real world traces with the rigid, restrictive correlation structure (1) Indeed, convincing evidence has been produced establishing the importance of short term correlations for buffering [13 15], and so called relevant time scales have been discovered [14, 16, 17] Generalizations of fBm fGn with a more flexible correlation structure than (1) can be synthesized almost effortlessly using the powerful decorrelating capability of the wavelet transform [18 21] There, independent Gaussian ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....Another useful feature of (1) is that it holds for a wider class of traffic than the corresponding result for large b asymptotics. It does not require that the traffic be mixing and, thus, the estimation based on it, described in this paper, should be valid for long range dependent traffic [30, 31]. The main condition we require of the multiplex of arrivals processes is that the limit of the finite time cumulant generating function (CGF) per source exists for each time scale t as N goes to infinity: t ( lim N 1 1 N t log Ee P N n=1 A n t (2) The finite time CGFs are ....

N.G. Duffield (1997). Economies of scale for long-range dependent traffic in short buffers. Telecommunications Systems. to appear.

....Data traffic models are an invaluable asset to the network analyst. In network analysis, model parameters are used to capture and summarize important characteristics of data traffic. With simple models, the impact of various parameters on network performance can be studied through analytical means [6, 63 67]. In cases where theoretical analysis is intractable, models are routinely used to synthesize test data traces for simulation purposes [68] Here, computational efficiency of the synthesis becomes as important as the accuracy. We begin with some historical remarks. Although LRD models have long ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....Data traffic models are an invaluable asset to the network analyst. In network analysis, model parameters are used to capture and summarize important characteristics of data traffic. With simple models, the impact of various parameters on network performance can be studied through analytical means [6, 64 68]. In cases where theoretical analysis is intractable, models are routinely used to synthesize test data traces for simulation purposes [69] Here, computational efficiency of the synthesis becomes as important as the accuracy. We begin with some historical remarks. Although LRD models have long ....

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

....time, the most likely way for the queue length Q to exceed a given level b is for arrivals to build up over the previous interval whose duration (b) is asymptotically proportional to b for large b. See [2, 12] for some general results of this nature for sources obeying mixing conditions; see [15, 16] specifically for large superpositions) Now suppose that the arrival process takes a time t r to relax back to stationarity from its To be published in Performance Evaluation conditioned state. Then the probability that the queue exceeds a level b at time u 0 will be the same as for the ....

N.G. Duffield, Economies of scale for long-range dependent traffic in short buffers, Telecommunications Systems. (1996) to appear.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, to appear, 1998.

No context found.

N. Duffield, "Economies of scale for long-range dependent traffic in short buffers," Telecommunication Systems, vol. 7, pp. 267--280, 1997.

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