H. Hayden, Voice Flow Control in Integrated Packet Networks, Master's thesis, MIT Dept. of EECS, Cambridge MA, 1981.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the following sections, we design a distributed protocol to achieve the GMM policy and prove its convergence. 3 A Distributed Protocol There have been extensive prior efforts on the design of distributed algorithms to achieve the classical maxrain rate allocation. Early algorithms by Hayden [6], Jaffe [8] and Gafni [5] required synchronization of all nodes for each iteration. Mosely s work in [11] was the first asynchronous algorithm. Unfortunately, this algorithm could not offer satisfactory convergence performance. Later, Ramakrishnan et al. proposed to use a single bit to ....

H. P. Hayden, "Voice Flow Control in Integrated Packet Networks," M.S. Thesis, Dept. of Elec. Eng. and Comp. Sci., MIT, Cambridge, MA, June 1981.

.... Queue s s s FCFS Queue Output Packets Figure 1: Fairness Queue 1 INTRODUCTION Current datagram networks are vulnerable to congestive collapse when offered load approaches network capacity [BG87] Although several endto end congestion avoidance algorithms have been proposed [BG87, Hay81, JR87, RFS90, Jac88], none of them have been shown to perform optimally in today s high speed, high bandwidth delay product networks [DKS89, BG87] This has led some researchers to conclude that gateways must participate in congestion avoidance [Nag87] To this end, it has been proposed that gateways use a fair ....

H. Hayden. Voice flow control in integrated packet networks. Technical Report LIDS-TH-1152, MIT Laboratory for Information and Decision Systems, 1981.

....several simultaneous sessions with different destinations. However, the analysis of dynamic traffic scenarios poses some challenges that need to be faced beforehand. The main problem is the identification of an appropriate performance measure for fairness. The Max Min Fairness definition [Ber87] [Hay81] [Jaf81] is universally used as an optimal fairness measure in static scenarios. Therefore, in the first part of this paper we extend this definition to dynamic scenarios and introduce an algorithm, hereafter the Dynamic Max Min algorithm, for the computation of Max Min fair rates in dynamic ....

H. Hayden. Voice flow control in integrated packet networks. MIT Laboratory for Information and Decision Systems - LIDS Report TH-1152, 1981.

....to changes in network state. The third avoids this trade off by use of direct rate control of sessions; however, it is subject to making inappropriate adjustments of the session rates because it lacks the information needed to estimate the effects of its adjustments. More sophisticated algorithms [6, 7, 8] use knowledge of the number of sessions flowing through each node to avoid these problems. The dictionary hash technique presented in this paper provides this knowledge in an effective and timely manner. This paper presents the dictionary hash algorithm and compares it to similar algorithms used ....

H. Hayden. Voice flow control in integrated packet networks. Technical Report LIDSTH -1152, MIT Laboratory for Information and Decision Systems, 1981.

....we increase the link control parameter until the capacity is fully utilised. Similarly, given a link with total flow through it exceeding the link capacity, we decrease the link control parameter to maintain feasibility. One such additive increase decrease algorithm was presented by Hayden in [11]. Some simple modifications to it were made by Mosely in [15] in order to prevent the link control parameters from increasing infinitely in non bottleneck links. Hayden s Algorithm has been adapted to our framework in the following algorithm. Algorithm 3.1 Initialisation: k = 0; 8l 2 L Do ....

H.P. Hayden, "Voice flow control in integrated packet networks", Master's thesis, MIT, Cambridge, October 1981.

....simultaneous sessions. The design of distributed algorithms for max min fair rate allocation has received much attention in existing literature. Distributed max min fair sharing algorithms were discussed in earlier literature in the context of speech transmission with variable rate coding (Hayden [12], Mosely [19] Hayden s algorithm consisted of a simple additive successive approximation update at each node. Mosely [19] extended Hayden s algorithm to accomodate asynchronous updates and network delays. A similar class of algorithms (ERICA [14] UT [10] have been proposed in recent literature ....

....The max min fair allocation is obtained by solving the following equation for C N X s=1 max( s ; 0 Since C P N s=1 s , the above equation has a unique solution . The session rates are then given by: for 1 s N , r s = max( s ; An iteration proposed by Hayden in [12] for solving this problem is (k 1) k) 1 N C N X s=1 max( s ; k) and it is easily shown that (k) 8 C w(k ) h( k h C h( k C w(k ) Figure 2: When Hayden s algorithm is applied to a link with random capacity, the computed link control parameter also ....

H.P. Hayden, "Voice flow control in integrated packet networks," Master's thesis, MIT, Cambridge, October 1981.

....control mechanism is adversely affected. In order to avoid large transients in the cases with large round trip times a low starting gain is used in the initial phase of a control cycle. Early work on MMF rate control in packet networks was done in the context of packet voice sessions; see [15], 14] 25] The basic framework is the one described in [14] The design of explicit rate control algorithms for elastic sessions, in the ATM ABR service context, has received much attention in the literature in the last five to six years. In [7] there is a comprehensive survey of the issues, ....

H.P. Hayden, "Voice flow control in integrated packet networks," M.S. thesis, Dept. of Electrical Engg. and Computer Sci. Massachussetts Inst. Tech. Cambridge, MA, June 1981.

.... example showing that the max min vector is fairly unstable, namely that a change of ffi is the rate of some session may change the allocation of another session by Omega Gamma ffi Delta 2 n 2 ) 1 Introduction The ATM Forum on Traffic Management has recently adopted rate based flow control [Cha94, Jaf81, Hay81, Mos84, GB84, Gaf82, BG87] as the basis for flow control in its networks for ABR (Available Bit Rate) traffic (see [Rob94] for example) The major arguments for rate based flow control, is the simplicity and modest hardware requirements per virtual circuit, compared with those of the credit based scheme. Conceptually, the ....

....computation essentially determines how the network resources (link 1 capacities) are divided among the different virtual circuits. The basic principle guiding this computation is the desire to share the link capacities in a fair way among the different sessions. The max min fairness criterion [BG87, Jaf81, GB84, Gaf82, Hay81, Mos84, Cha94] is widely accepted as the theoretical criteria to guide this computation. The network is considered to be in a state of max min fairness if it is impossible to infinitesimally increase the rate of any session without decreasing the rate of sessions whose rate is equal or smaller. Thus, the ....

[Article contains additional citation context not shown here]

H. Hayden. Voice flow control in integrated packet networks. Master's thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering, Cambridge, MA, 1981.

....kleinber cs.cornell.edu. Supported in part by a David and Lucile Packard Foundation Fellowship, an ONR Young Investigator Award, and NSF Faculty Early Career Development Award CCR 9701399. 1 each individual. In the networking literature, this approach underlies the notion of maxmin fairness [1, 11, 12], a widely used mechanism which produces an allocation for a fixed set of routes that is lexicographically maximum: the minimum bandwidth assigned to any connection is maximized; subject to this, the minimum bandwidth assigned to any of the remaining connections is maximized; and so forth. Such a ....

....to a well studied distribution of wealth: it is an instance of the Pareto distribution [18] one of the simplest types of allocations to follow an inverse power law. 2 Overview of Results Bandwidth Allocation. We consider the problem that underlies the Hayden Ja#e model of max min fairness [1, 11, 12]. We are given a graph G = V, E) and paths P 1 , P n in this graph, representing users transmitting data. We must assign a rate x i to each path P i so that the following capacity condition is satisfied: if P i 1 , P i t all pass through a common 4 edge, then # t j=1 x i j ....

H. Hayden, Voice Flow Control in Integrated Packet Networks, Master's thesis, MIT Dept. of EECS, Cambridge MA, 1981.

....of feedback. Upon receipt of the feedback signal the source adjusts its estimate of the allowed transmission rate according to some rule. These algorithms essentially differ in the particular choices of link controls and the type of feedback provided to the source by the network. References [7] [17], 19] describe distributed algorithms of this type. However, these algorithms required synchronization, which is difficult to achieve. Mosley in [31] suggested an asynchronous algorithm for distributed calculation of maxmin fair rates. However, the algorithm convergence time was rather long and ....

....more rapid convergence, and accounting for the bandwidth used by the feedback traffic. 2 Global Calculation of Optimal Flow Rates This section presents an analytical development of the global maxmin computation, and extends the basic idea of global computation of maxmin rates found in [2] 31] [17], 35] to account for the bandwidth consumed by feedback traffic, under the assumption that the feedback rate is proportional to the data rate in the forward direction. Suppose there are f forward and b feedback flows traversing a given link. Let k be the ratio of feedback to forward data rates. ....

Hayden, H. 1981. "Voice Flow Control in Integrated Packet Networks', (report LIDS-TH-1152). Cambridge, MA: MIT Laboratory for Information and Decisions Systems.

.... bu er over ow has been one of the main causes of process failures in distributed systems (see, e.g. 31] The function of ow control algorithms is to alleviate throughput degradation, unfairness, deadlocks, and failures by preventing such situations from arising (see, e.g. 3, Chapter 6] or [4, 10, 15, 17, 22, 24, 25, 32]) In particular, rate based ow control algorithms adjust transmission rates of di erent sessions in an end to end manner, with the objective to optimize network utilization while still maintaining fairness between di erent sessions (see, e.g. 1, 3, 7, 16, 19, 20, 24] The ratebased approach ....

.... Transfer Mode (ATM) Forum on Trac Management has adopted rate based ow control as the prime mechanism for ow control of Available Bit Rate (ABR) trac in its networks (see, e.g. 5, 12, 30, 36, 38, 41] A widely accepted fairness criterion for rate based ow control is max min fairness [1, 3, 7, 8, 9, 19, 20, 21, 22, 24, 27], requiring that it be impossible for any session to receive a larger rate on the account of a session with a smaller or equal rate. Call max min rates those achieving max min fairness. Any rate based ow control algorithm can be classi ed into one of two broad classes, conservative and ....

[Article contains additional citation context not shown here]

H. Hayden, \Voice Flow Control in Integrated Packet Networks," Technical Report MIT/LIDS/TH/TR-601, Laboratory for Information and Decision Systems, Massachusetts Institute of Technology, 1981.

....resources. In this work, we take a first step in this direction by assembling a theory of max min fair, rate based flow control sensitive to priorities assigned to distributed applications, as a significant extension of the classical theory of max min fair, rate based flow control (see, e.g. [2, 7, 20, 21, 22]) to the setting with priorities. In a nutshell, our theory advocates a new fairness condition we call priority max min fairness, which accounts for priorities, the study of a class of algorithms, termed priority bottleneck algorithms, that are found to converge to priority maxmin fairness, and ....

....respects the capacity constraint for each individual link. One of the most challenging aspects of modern networking is treating all sessions fairly when it is necessary to turn some of the bursty traffic away from the network. The most widely accepted fairness condition is max min fairness [1, 2, 7, 15, 16, 17, 18, 20, 21], which, roughly speaking, requires that it be impossible to infinitesimally increase the rate of any application without decreasing the rate of some other with a smaller rate. However, this condition is inadequate in case different applications have acquired varying priorities on the basis of ....

[Article contains additional citation context not shown here]

H. Hayden, "Voice Flow Control in Integrated Packet Networks," Report LIDS/TH/1152, Laboratory for Information and Decision Systems, Massachusetts Institute of Technology, 1981.

....may converge faster. However, in rare configurations we show that the resulting rates may be substantially different than the rates of the max min vector (this difference is not necessarily bad) 1 Introduction The ATM Forum on Traffic Management has recently adopted rate based flow control [Cha94, Jaf81, Hay81, Mos84, GB84, Gaf82, BG87] as the basis for flow control in its networks (see [Rob94] for example) The major arguments for rate based flow control, is the simplicity and modest hardware requirements of per virtual circuit, compared with those of the credit based scheme. Conceptually, the rate base flow control adjusts ....

....computation essentially determines how the network resources (link capacities) will be divided among the different virtual circuits. The basic principle guiding this computation is the desire to share the link capacities in a fair way among the different sessions. The max min fairness criterion [BG87, Jaf81, GB84, Gaf82, Hay81, Mos84, Cha94] is widely accepted as the guideline for this computation. The network is considered to be in a state of max min fairness if it is impossible to infinitesimally increase the rate of any session without decreasing the rate of sessions whose rate is equal or smaller. Thus, the fairness is that each ....

[Article contains additional citation context not shown here]

H. Hayden. Voice flow control in integrated packet networks. Master's thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering, Cambridge, MA, 1981.

....access control such as buffer insertion or slotted ring, fairness is a major problem. The main result of this paper is a new local distributed scheduling algorithm to regulate access to the network, such that the competing session are allocated, in steady state, approximate Max Min fair rates [Hay81, Jaf81, BG87] and experience minimum transmission delay. In contrast, we also show that exact Max Min rates (rather than approximate ones) would have required global state information. Global information based protocols do not assure minimum delay, are centralized by nature (even if nodes act locally, they do ....

H. Hayden. Voice flow control in integrated packet networks. MIT Laboratory for Information and Decision Systems - LIDS Report TH-1152, 1981.

....Many different satisfaction functions have been suggested. Another fairness approach, called max min flow control or bottleneck flow control, is used in various forms in [7: Section 3] 9: Chapter 4] 10: Chapter 3] and [19] 25] Only the simplest version of this objective, viz. Hayden s [9], will be defined here. To satisfy the max min flow criterion, the smallest session rate in the network must be as large as possible. Subject to this constraint, the second smallest session rate must be as large as possible, etc. Given a network with its link capacities and a set of sessions with ....

....C(x, h, s, t) is defined to equal C (x , l , s , t ) For s t , define C (x , l , s , t ) and C (x , h , s , t ) to be zero. 3. FAIRNESS CRITERION This section describes the max min flow criterion, our definition of throughput fairness. The version presented here was developed by Hayden [9]. Similar versions were proposed independently by Jaffe [20, 21] and Luss and Smith [25] The criterion is described here as it applies to the system model presented in Section 2. In particular, it is assumed that all links have unit capacity (in each direction) and that the sessions and their ....

[Article contains additional citation context not shown here]

H. P. Hayden, "Voice Flow Control in Integrated Packet Networks," Report LIDS-TH-1152, Lab. for Info. and Decision Sys., Mass. Inst. of Technology, Cambridge, Mass., Oct. 1981.

....at the corresponding nodes; in case this happens, packets arriving at nodes with no available buffer space will have to be discarded and later retransmitted, thereby wasting communication resources. The function of flow control algorithms is to prevent such situations from arising (see, e.g. [3, 7, 9, 12, 13, 16]) In particular, rate based flow control algorithms adjust transmission rates of different sessions in an end to end manner, with the objective to optimize network utilization while maintaining fairness between different sessions (see, e.g. 1, 3, 6, 8, 10, 11, 13] The rate based approach has ....

.... approaches [15] Indeed, the ATM Forum on Traffic Management has adopted rate based flow control as the mechanism for flow control of Available Bit Rate (ABR) traffic in its networks (see, e.g. 4, 5, 18] A widely accepted fairness criterion for rate based flow control is max min fairness [1, 3, 6, 10, 11, 12, 13], requiring that, in a state of max min fairness, it be impossible to infinitesimally increase the rate of any session without decreasing the rate of a session whose rate is equal or smaller. All existing rate based flow control algorithms fall into one of two broad classes, conservative and ....

H. Hayden, "Voice Flow Control in Integrated Packet Networks," Technical Report MIT/LIDS/TH/TR-601, Laboratory for Information and Decision Systems, Massachusetts Institute of Technology, 1981.

....ffi Delta 2 n 2 ) but by no more than O(ffi Delta 2 n ) This implies that the max min fairness criteria may provide a bad estimate of how far a set of flow allocations is from the optimal allocation. 1 Introduction The ATM Forum on Traffic Management has adopted rate based flow control [Cha94, Jaf81, Hay81, Mos84, GB84, Gaf82, BG87] as the basis for flow control in its networks for ABR (Available Bit Rate) traffic (see [BF95] for example) The major arguments for rate based flow control, is the simplicity and modest hardware requirements per virtual circuit, compared with those of the credit based scheme [KM95] ....

....This computation essentially determines how the network resources (link capacities) are divided among the different virtual circuits. The basic principle guiding this computation is the desire to share the link capacities in a fair way among the different sessions. The max min fairness criterion [BG87, Jaf81, GB84, Gaf82, Hay81, Mos84, Cha94] is widely accepted as the theoretical criteria to guide this computation. A set of sessions is said to be in the state of max min fairness if it is impossible to infinitesimally increase the rate of any session without decreasing the rate of some sessions whose rate is equal or smaller. It can be ....

[Article contains additional citation context not shown here]

H. Hayden. Voice flow control in integrated packet networks. Master's thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering, Cambridge, MA, 1981.

....ffi Delta 2 n 2 ) but by no more than O(ffi Delta 2 n ) This implies that the max min fairness criteria may provide a bad estimate of how far a set of flow allocations is from the optimal allocation. 1 Introduction The ATM Forum on Traffic Management has adopted rate based flow control [Cha94, Jaf81, Hay81, Mos84, GB84, Gaf82, BG87] as the basis for flow control in its networks for ABR (Available Bit Rate) traffic (see [BF95] for example) The major arguments for rate based flow control, is the simplicity and modest hardware requirements per virtual circuit, compared with those of the credit based scheme [KM95] ....

....This computation essentially determines how the network resources (link capacities) are divided among the different virtual circuits. The basic principle guiding this computation is the desire to share the link capacities in a fair way among the different sessions. The max min fairness criterion [BG87, Jaf81, GB84, Gaf82, Hay81, Mos84, Cha94] is widely accepted as the theoretical criteria to guide this computation. The network is considered to be in a state of max min fairness if it is impossible to infinitesimally increase the rate of any session without decreasing the rate of sessions whose rate is equal or smaller. Thus, the ....

[Article contains additional citation context not shown here]

H. Hayden. Voice flow control in integrated packet networks. Master's thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering, Cambridge, MA, 1981.

....of feedback. Upon receipt of the feedback signal the source adjusts its estimate of the allowed transmission rate according to some rule. These algorithms essentially differ in the particular choices of link controls and the type of feedback provided to the source by the network. References [6] [15], 17] describe distributed algorithms of this type. However, these algorithms required synchronization, which is difficult to achieve. Mosley in [23] suggested an asynchronous algorithm for distributed calculation of maxmin fair rates. The algorithm was shown to converge to maxmin optimal rates. ....

....global procedure to obtain maxmin optimal rates for a network with feedback traffic. 2.2 Finding Globally Optimal Rates In this section we give a way to find the stationary optimal vector j given global information about the network. The results here are quite similar to those given in [1] 23] [15], 26] However, this work considers a somewhat different model than the cited authors, since our model accounts for the bandwidth consumed by feedback flows. Note that it is not clear a priori whether it is legitimate to treat feedback sessions in the same way as independent forward sessions, ....

Hayden, H. 1981. "Voice Flow Control in Integrated Packet Networks', (report LIDS-TH-1152). Cambridge, MA: MIT Laboratory for Information and Decisions Systems.

....namely, fairly sharing the network bandwidth among different users, defining it rigorously in a network environment is not trivial. A widely acceptable measure of fairness is the A preliminary version of this paper was presented in Sigcomm 96 max min fairness [BG87, Jaf81, GB84, Gaf82, Hay81, Mos84, Cha94] A bandwidth allocation is max min fair if for every session, one cannot increase its bandwidth without decreasing the bandwidth of sessions of equal or lower bandwidth [BG87] This paper presents an efficient, fair and simple scheme for flow control, called Phantom, which is ....

....without multiplications or divisions. The idea of adding another imaginary session to every link is not new. Jaffe [Jaf81] was first to use the idea in order to study the convergence of flow control algorithms, however, that algorithm, as well as other related works along this lines [GB84, Hay81, Mos84] may not be effectively used in a distributed environment in which the traffic frequently changes. In [CH96] Choudhury and Hahne employ the same principle of using the spare resources to control the resource consumption as suggested by Jaffe and as we do. However, Choudhury and Hahne ....

H. Hayden. Voice flow control in integrated packet networks. Master's thesis, Massachusetts Institute of Technology, Dept. of Electrical Engineering, Cambridge, MA, 1981.

No context found.

H. Hayden, Voice Flow Control in Integrated Packet Networks, Master's thesis, MIT Dept. of EECS, Cambridge MA, 1981.

No context found.

H. Hayden. Voice flow control in integrated packet networks. Technical Report LIDS-TH-1152, MIT Laboratory for Information and Decision Systems, 1981.

No context found.

H. P.Hayden, #Voice FlowControl in Integrated Packet Networks," M.S. Thesis, Dept. of Elec. Eng. and Comp. Sci., MIT, Cambridge, MA, June 1981.

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