J.M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29(7), 1981.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....must be non negative. In order to be optimal, a rate allocation must utilize network resources as much as possible without obstructing the fairness goal. The criterion most widely used to achieve fairness in rate allocation is the max min fairness criterion, presented in various versions in [8, 10, 16] and other places. Max min fairness attempts to allocate equal rates to all active sessions. However, this is often impossible because of link capacity limitations. According to max min fairness, a feasible rate allocation is fair if the rate r s allocated to a session s cannot be increased ....

....allocate a rate of 0 to some of the sessions. The third definition guarantees a positive allocation to all sessions. Thus, it is fair in the sense that it allows network access to all sessions (but without performance guarantees) The bottleneck or max min fair flow control scheme presented in [16] concentrates on fair and optimal allocation of rates. In order to avoid formation of infinite delay, a fraction of the link capacity is always kept unallocated. For each link if r = c Gamma is the unallocated bandwidth, then the maximum bandwidth that is allowed to be allocated to a session on ....

[Article contains additional citation context not shown here]

J. M. Jaffe. Bottleneck flow control. IEEE Transactions on Communications, COM29: 954--962, April 1981.

....router. When a packet arrives at a router, it is first classified into flow by the classifier. The packet is then put into its corresponding subqueue for transmission. The scheduler picks packets from subqueues for transmission in a Round Robin fashion. FQ WFQ can achieve perfect max min fairness [38], and has been widely used (for comparison purposes) to evaluate the performance of other fair algorithms proposed in recent years (for example in [9] 17] 42] 51] and [65] 3.1.2 CSFQ Core Stateless Fair Queue (CSFQ) is a simple probabilistic packet dropping algorithm proposed by Stoica ....

....the maximum queueing delay at a router) as recommended in [65] 4.3.1. 2 Estimation of Fairshare Apart from packet arrival rate, fairshare (of the bottleneck link) is another variable maintained by UPTA (in order to compute packet loss rate) UPTA estimates fairshare based on max min criterion [38]. We use following notations: C: Output link capacity. N : Total number of flows. n: Collection of all flows (n = 1, 2, N ) S: Number of unconstrained flows (# i for i = 1, S) M : Number of flows currently in U intervals. m: Collection of U flows (m = ....

[Article contains additional citation context not shown here]

J. M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7):954--962, July 1981.

....call it WFQ ) As the MPEG 2 video is encoded at a bit rate of 1.5 Mbps, packet arrival rate of the video application (at router R 1 ) is about 1.76 Mbps, after taking TS header (4 bytes) UDP header (8 bytes) and IP header (20 bytes) into account. Based on max min fair allocation criterion [8], we calculate packet arrival rate, packet departure rate, and packet loss rate for the video flow, as shown in Table 1. From Table 1 we can see that the packet loss rate (of the video flow) seen by routers R 1 R 7 is less than the loss threshold of 12 . Therefore, the video flow is considered ....

J. M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7):954--962, July 1981. 28

....o#ered rates C user max(i) and C . In other words, we need to basically divide this extra capacity C ded unused among all the needy groups in a fair manner. However, fair sharing of extra resources is not a trivial issue and was addressed by others for di#erent network situations [ZC93] Jaf81] Wd89] WSF82] Some proposals [Jaf81] are in favor of sharing the bottleneck capacity equally among users independent of their requirements and others [ZC93] Wd89] advocate to penalize users causing overloads. While we do share the resources among VPN connections in each group, equal sharing ....

..... In other words, we need to basically divide this extra capacity C ded unused among all the needy groups in a fair manner. However, fair sharing of extra resources is not a trivial issue and was addressed by others for di#erent network situations [ZC93] Jaf81] Wd89] WSF82] Some proposals [Jaf81] are in favor of sharing the bottleneck capacity equally among users independent of their requirements and others [ZC93] Wd89] advocate to penalize users causing overloads. While we do share the resources among VPN connections in each group, equal sharing of unused dedicated capacity will not ....

J.M. Ja#e. Bottleneck Flow Control. IEEE Transactions on Communications, 29(7):954 -- 962, 1981.

....fairness definitions are often used to reduce the set of desired allocations. A weighted max min fair allocation is one where each agent receives their fair share with respect to their weight (priority or wealth) and who they compete against. This definition was proposed for computer networks [14], and is the fairness sought by many bandwidth allocation mechanisms [2, 12] The following indicates that the allocation of the economy in equilibrium is weighted max min fair, where the wealth of each user is their weight. Definition 2.4. Weighted max min fair: An allocation of resources ....

J. M. Ja#e. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29:954 -- 962, July 1981.

....the rate vector resulting from this schedule will usually be deemed to be unfair, as it does not serve the links (0, 1) and (0, 2) at all. Since our main interest is in good general purpose topologies, a link based definition of fairness is appropriate. We apply the concept of max min fairness [10] to scatternets, i.e. the rate vector has to maximize the rate of the links with the minimal allocation. The max min fair rate vector may be defined in analogy to [11] Definition 3 A rate vector r is max min fair if it is feasible and for each (u, v) L and feasible r for which r uv r uv ....

J. Jaffe, Bottleneck Flow Control, IEEE Transactions on Communications, 29(7), July 1981.

....choose a diminishing weight function according to its administrative policies. Although the max min rate constitute Nash and Stackelberg Equilibrium, it is not clear how users can estimate their max min fair rates. For this, a decentralized distributed scheme such as the one is proposed [15] is required. Moreover one needs to establish that such a distributed scheme will be stable and will indeed converge to the max min fair rates when DWS schedulers are deployed in the network. This is a topic under investigation. Our current paper does not address this issue in a theoretical ....

J. M. Jae. Bottleneck ow control. IEEE Transactions on Communications, 29(7):954-962, 1981.

....offered rates C user max(i) and C shared . In other words, we need to basically divide this extra capacity C ded unused among all the needy groups in a fair manner. However, fair sharing of extra resources is not a trivial issue and was addressed by others for different network situations [24] [14], 22] 23] Some proposals [14] are in favor of sharing the bottleneck capacity equally among users independent of their requirements and others [24] 22] advocate to penalize users causing overloads. While we do share the resources among VPN connections in each group, equal sharing of unused ....

....C shared . In other words, we need to basically divide this extra capacity C ded unused among all the needy groups in a fair manner. However, fair sharing of extra resources is not a trivial issue and was addressed by others for different network situations [24] 14] 22] 23] Some proposals [14] are in favor of sharing the bottleneck capacity equally among users independent of their requirements and others [24] 22] advocate to penalize users causing overloads. While we do share the resources among VPN connections in each group, equal sharing of unused dedicated capacity will not help ....

J.M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29(7):954 -- 962, 1981.

....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 contexts. ....

J. M. Jaffe. Bottleneck flow control. IEEE Transactions on Communications, COM-29(7):954962, July 1981.

....proportional fairness and general weighted fairness are not intended to cover all imaginable de nitions, but we believe that they represent the most common ones for communication networks. 3.1. 1 Maxmin Fairness The most popular fairness concept in computer networks is that of maxmin fairness[18]. This fairness de nition corresponds to the Rawlsian welfare function W (u 1 ; u n ) min(u 1 ; u n ) with the individualistic utility functions u i (x i ) x i 8i f1; ng, i.e. for all users i, the utility of the resource bundle allocated to user i is represented by the value of ....

J. Jae, \Bottleneck ow control," IEEE Transactions on Communications, vol. 7, no. 29, pp. 954-962, July 1980.

....is, the bandwidth granularity and the time scale over which adaptation occurs. Network resource allocation schemes, however, are more complex than the case of a single link because one flow s allocation can be affected by other flows sharing a portion of a multi hop route. Max min fairness [13] is the most widely used fairness criterion found in bandwidth allocation algorithms for networks. Here, the idea is to maximize the allocation of flows with the least allocation; that is, to allow a flow to increase its allocation provided that the increase does not subsequently cause a decrease ....

J. Jaffe, Bottleneck flow control, IEEE Transactions on Communications 29(7) (July 1981).

....link bandwidth will be shared equally by all flows (not being bottlenecked elsewhere) If other better information is available, Remos can use different sharing policies when estimating flow bandwidths. The basic sharing policy assumed by Remos corresponds to the max min fair share policy [17], which is the basis of ATM flow control for ABR traffic [19, 2] and is also used in other environments [15] 4.3 Logical network topology Remos supports queries about the network structure and topology in addition to queries about specific flows in the network. The reason we expose a network ....

Jaffe, J. M. Bottleneck flow control. IEEE Transactions on Communications 29, 7 (July 1981), 954--962.

.... [1, 10] The design of binary adjustment algorithms for conventional networks with First In First Out (FIFO) link scheduling has been motivated by three requirements: responsiveness to congestion notifications; smoothness of rate adjustments; and fairness of resource allocation across flows [11, 12, 16, 29]. For instance, the large load oscillations characteristic of TCP has led to the development of several adjustment algorithms that provide smoother congestion control for streaming media applications [27] To achieve the goal of smoothness, some solutions offer new settings for the parameters of ....

....t equals: L(t) n X k=1 l k (t) 1) load throughput feedback load throughput feedback load throughput feedback Other Flows Examined Flow 1 0 1 capacity Network fair share Figure 1: The network model. The network splits its capacity between flows according to the principle of maxmin fairness [8, 11]. A recursive procedure for computing this fair allocation is given in [21] The procedure assigns a throughput b k (t) to flow k based on the notion of fair share s(t) at time t. If the flow demands less than the fair share, its demand is fully satisfied. Otherwise, the flow receives the fair ....

J. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29:954 -- 962, July 1981.

....we introduce the notion of the bandwidth utility function of a network application, which is the basis of our new generalized fairness definition. A. Bandwidth max min fairness A commonly used fairness criterion in a rate based flow control system is bandwidth max min fairness, introduced in [9]. This fairness definition has been specified by the ATM Forum as a major goal of the flow control algorithm in ABR service [1] To give a mathematical definition of bandwidth max min allocation, we first define the notion of feasible bandwidth allocation. Definition 1. A feasible bandwidth ....

J. M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29:954--962, July 1981.

....the bottleneck link bandwidth is shared equally by all flows (that are not bottlenecked elsewhere) If other information is available, Remos can use different sharing policies when estimating flow bandwidths. The basic sharing policy assumed by Remos corresponds to the max min fair share policy [11]. Applications that use topology based queries are themselves responsible for taking the effects of both internal and external sharing into account. 3.3 Accuracy Applications ideally want information about the level of service they can expect to receive in the future, but most users today must ....

J. M. Jaffe. Bottleneck flow control. IEEE Transactions on Communications, 29(7):954--962, July 1981.

....is fairly insensitive to how higher priority sessions are routed, assuming sufficient link capacity is set aside for these sessions. For this reason, we focus on high bandwidth traffic. We will assume that best effort sessions share resources according to the max min fairness principle (see [15, 17] for details) 3.2 Algorithm overview Since we expect best effort traffic to continue to be the dominant traffic class in the Internet, we will focus on that case, so our design has two goals. First, it seeks to admit the same number of QoS sessions as would be admitted by a network dedicated to ....

J.M. Jaffe. Bottleneck flow control. IEEE Transactions on Communications, COM-29(7):954--962, July 1981. Correspondence.

No context found.

J. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7):954--962, 1980.

....characterizes how competing users should share the bottleneck resources and has been. In this section, we review and compare standard definitions of fairness and generalize them. 3. 1 Max min Fairness One of the most common fairness definitions is max min or bottleneck optimality criterion [13, 1, 8, 16, 2]. A feasible flow rate x is defined to be max min fair if any rate x i cannot be increased without decreasing some x j which is smaller than or equal to x i [1] Many researcher have developed algorithms achieving max min fair rates [1, 16, 2] But max min fair vector needs global information ....

J.M. Jaffe. Bottleneck flow control. IEEE Transaction on Communications, COM-29(7), July 1981.

....of the global performance of the network, is not only difficult to measure but is also difficult to define. Resource centered metrics are more easily obtained, and do not require synchronization. That resource centered metrics are appropriate ones for use in optimization of power is shown by [Jaf81]. It would be valuable for the goal of developing effective gateway congestion handling if Management Information Base (MIB) objects useful for evaluating gateway congestion were developed. The reflections on the control interval described above should be applied when network management ....

J. Jaffe, Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7), July, 1981.

No context found.

J.M. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29(7), 1981.

No context found.

J. Jaffe. Bottleneck flow control. IEEE Transactions on Communication, 29:954--962, 1981.

No context found.

J. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, 29:954 -- 962, July 1981.

No context found.

J. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7)(?)( 1980.

No context found.

J. Jaffe. Bottleneck flow control. IEEE Transactions on Communications, 7(29):954--962, July 1980.

No context found.

J. Jaffe. Bottleneck Flow Control. IEEE Transactions on Communications, COM-29(7):954--962, 1980.

First 50 documents  Next 50

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