• Documents
  • Authors
  • Tables
  • Log in
  • Sign up
  • MetaCart
  • DMCA
  • Donate

CiteSeerX logo

Advanced Search Include Citations
Advanced Search Include Citations

Fluid approximations for a processor-sharing queue. Queueing Systems Theory Appl (1997)

by H CHEN, O KELLA, G WEISS
Add To MetaCart

Tools

Sorted by:
Results 1 - 6 of 6

The Fluid Limit of a Heavily Loaded Processor Sharing Queue

by H. Christian Gromoll, Amber L. Puha, Den Dolech , 2001
"... ..."
Abstract - Cited by 34 (6 self) - Add to MetaCart
Abstract not found
(Show Context)

Citation Context

...ion type of property for the delays incurred by customers, Grishechkin [18] who obtains asymptotics of the steady state distribution as the traffic intensity approaches one, and Chen, Kella and Weiss =-=[9]-=- who consider a fluid approximation for the queue length process. In particular, there is no heavy traffic diffusion approximation for the queue length process of a GI/GI/1 processor sharing queue. (N...

Stochastic Models for Throughput Analysis of Randomly Arriving Elastic Flows in the Internet

by Arzad A. Kherani, Anurag Kumar - IN THE INTERN IN PROCEEDINGS OF IEEE INFOCOM , 2002
"... This paper is ahnut analytical mndels fnr calculating the aver-age handwidth shares nhtained hy TCP cnntrnlled finite file transfers that arrive randnmly and share a single (bottleneck) link. Owing tn the complex nature nf the TCP cnngestinn cnntrnl algnrithm, a single mndel dries nnt wnrk well fnr ..."
Abstract - Cited by 25 (3 self) - Add to MetaCart
This paper is ahnut analytical mndels fnr calculating the aver-age handwidth shares nhtained hy TCP cnntrnlled finite file transfers that arrive randnmly and share a single (bottleneck) link. Owing tn the complex nature nf the TCP cnngestinn cnntrnl algnrithm, a single mndel dries nnt wnrk well fnr all cnmhinatinns nf netwnrk parameters (i.e., mean file size, link capacity, and prnpagatinn delay). We prnpnse twn mndels, de-velnp their analyses, and identify the reginns nf their applicability. One mndel is nhtained frnm a detailed analysis nf TCP’s AIMD adaptive win-dnw mechanism; the analysis accnunts fnr sessinn arrivals and departures, and finite link buffers. It is essentially a Prncessnr Sharing (PS) mndel with time varying service rate; hence we call it TCP-PS. The nther mndel is a simple mndificatinn nf the PS mndel that accnunts fnr large prnpagatinn delays; we call this mndel Rate Limited-PS (RL-PS). The TCP-PS mndel analysis accnmmndates a general file size distrihutinn hy apprnximating it with a mixture nf expnnentials. The RL-PS mndel can he used fnr general file size distrihutinns. We shnw that the TCP-PS mndel cnnverges tn the standard PS mndel as the prnpagatinn delay apprnaches zern. We alsn nhserve that the PS mndel prnvides very pnnr estimates nf thrnughput unless the prnpagatinn delay is very small. We nhserve that the key parameters affecting the thrnughput are the handwidth delay prnduct (BDP), file size distrihutinn, the link buffer and the traffic intensity. Several numerical cnmparisnns between analytical and simulatinn re-sults are prnvided. We nhserve that the TCP-PS mndel is accurate when the BDP is small cnmpared tn the mean file size, and the RL-PS mndel wnrks well when the BDP is large cnmpared tn the mean file size. I
(Show Context)

Citation Context

... cumbersome [20]. This makes � analytically intractable. However, it is easily seen that for the PS model, � �s�. See also other related performance measures that have been used in [1], [16], =-=[2] and [17]. 0-7803-7476-2/02/$17.00 (c) 2002 IEEE. If all the ongoing sessions -=-get an equal share of the total rate Ê Ù (i.e., Ö� Ù �Ö Ù ) Ê Ø Ö Ù �Ù � � Ð�Ñ Ê Ø� Ø Á�Æ Ù � ��Ù � Ð�ÑØ� Ê Ø Ø Ö Ù �Ù Ê Ø Ð�ÑØ� Ø ...

Stability of Data Networks: Stationary and Bursty Models

by Heng-Qing Ye, Jihong Ou, Xue-Ming Yuan - OPERATIONS RESEARCH , 2005
"... This paper studies stability of network models that capture macroscopic features of data communication networks including the Internet. The network model consists of a set of links and a set of possible routes which are fixed subsets of links. A connection is dynamically established along one of ..."
Abstract - Cited by 19 (3 self) - Add to MetaCart
This paper studies stability of network models that capture macroscopic features of data communication networks including the Internet. The network model consists of a set of links and a set of possible routes which are fixed subsets of links. A connection is dynamically established along one of the routes to transmit data as requested, and terminated after the transmission is over. The transmission bandwidth of a link is dynamically allocated, according to specific bandwidth allocation policy, to ongoing connections that traverse the link. A network model is said to be stable under a given bandwidth allocation policy if, roughly, the number of ongoing connections in the network will not blow up over time.

Invariant states and rates of convergence for a critical fluid model of a processor sharing queue

by Amber L. Puha, Ruth J. Williams - ANNALS OF APPLIED PROBABILITY , 2004
"... This paper contains an asymptotic analysis of a fluid model for a heavily loaded processor sharing queue. Specifically, we consider the behavior of solutions of critical fluid models as time approaches ∞. The main theorems of the paper provide sufficient conditions for a fluid model solution to conv ..."
Abstract - Cited by 12 (2 self) - Add to MetaCart
This paper contains an asymptotic analysis of a fluid model for a heavily loaded processor sharing queue. Specifically, we consider the behavior of solutions of critical fluid models as time approaches ∞. The main theorems of the paper provide sufficient conditions for a fluid model solution to converge to an invariant state and, under slightly more restrictive assumptions, provide a rate of convergence. These results are used in a related work by Gromoll for establishing a heavy traffic diffusion approximation for a processor sharing queue.
(Show Context)

Citation Context

... in the sense that ¯µξ (t) converges to the zero measure as t tends to ∞, but 〈χ, ¯µξ(t)〉 does not converge to 0. The result in Theorem 1.2 is more general than, but consistent with, Proposition 5 of =-=[4]-=-, which concerns the asymptotic behavior of a fluid approximation for the queue length of a heavily loaded processor sharing queue. Theorem 1.2 is proved in Section 4, using proof techniques similar t...

Flow-size based priorities in Internet routers 1 Differential equation models of flow-size based priorities in Internet routers

by Ashvin Lakshmikantha, R. Srikant, C. L. Beck
"... Abstract: Simulation studies have shown that providing priority to short-flows in Internet routers can dramatically reduce their mean delay while having little impact on the long-flows that carry the bulk of Internet traffic. In this paper, we present simple differential equation models (commonly re ..."
Abstract - Add to MetaCart
Abstract: Simulation studies have shown that providing priority to short-flows in Internet routers can dramatically reduce their mean delay while having little impact on the long-flows that carry the bulk of Internet traffic. In this paper, we present simple differential equation models (commonly referred to as fluid models) that can be used to analytically quantify this observation. Our model is at the connection-level where file arrivals and departures are described using differential equations, while ignoring the packet-level effects. We demonstrate that network performance can be enhanced significantly by giving priorities to short-flows without seriously affecting the throughput of long-flows.
(Show Context)

Citation Context

... the assumptions of Lemma 3, then as t → ∞, nsi (t) → n∗si for all i ∈ S and nli (t) → n∗ for all i ∈ L, li exponentially. The proof of this theorem relies on some results on processor sharing queues =-=[5]-=- and some results from non-linear systems theory. Therefore before we can prove the theorem, we have to digress and present some preliminaries which were first developed in [5]. For the sake of comple...

© 2006 INFORMS The Fluid Limit of an Overloaded Processor Sharing Queue

by Amber L. Puha, Alexander L. Stolyar, Ruth J. Williams , 2004
"... informs ® doi 10.1287/moor.1050.0181 ..."
Abstract - Add to MetaCart
informs ® doi 10.1287/moor.1050.0181
(Show Context)

Citation Context

...sure (cf. Jean-Marie and Robert [10, Proposition 4] and the last paragraph of §3.3.2 here). Fluid approximations for the queue length processes of processor sharing queues are proposed in Chen et al. =-=[2]-=-. As was noted in the introduction of Gromoll et al. [9], the justification given in [2] for those fluid approximations is valid, provided that all limit points are deterministic. Our approach provide...

Powered by: Apache Solr
  • About CiteSeerX
  • Submit and Index Documents
  • Privacy Policy
  • Help
  • Data
  • Source
  • Contact Us

Developed at and hosted by The College of Information Sciences and Technology

© 2007-2019 The Pennsylvania State University