K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. In INFOCOM'90, June 1990. 85] A. E. Baratz, J. P. Gray, P. E. Green Jr., J. M. Jaffe, and D. P. Pozefsky. Sna networks of small systems. IEEE Journal on Selected Areas in Com., SAC-3(3):416--426, May 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....traffic are known to be other than deterministic. Technical Subject Area: Modeling and Simulation Techniques 1 N. Prithviraj A Simulation Study Of Token Generation Policies 2 1 Introduction The leaky bucket is a recently proposed cell level access control mechanism for high speed networks [1, 5, 8]. Combined with call level admission controls, cell level controls such as the leaky bucket form a key part of the congestion control framework [2, 3] in high speed integrated services networks, such as B ISDN ATM [7] The leaky bucket traffic shaper serves two main purposes. First, it reduces ....

....The token pool size at the shaper determines the maximum number of back to back cells that can enter the network. Most of the existing research on the leaky bucket mechanism has concentrated on sizing the input buffer and token pool to reduce the burstiness of traffic entering the network [1, 5], and to balance the tradeoff between delay and loss at the shaper [6] However, the token generation rate and the token generation policy are also important parameters. The conventional leaky bucket mechanism uses a deterministic policy for token generation: if the token generation rate is R, ....

K. Bala, I. Cidon, and K. Sohraby, "Congestion Control for High Speed Packet Switched Networks", Proceedings of INFOCOMM

....traffic specification from the traffic source. This traffic description is used to determine if the new connection s QOS demands can be met without violating the QOS demands of the preestablished connections. After connection establishment, the network imposes a usage parameter control (UPC) [1, 2, 6, 8, 16, 18, 27, 30, 31, 33] to make sure that the admitted connections do not deviate from their traffic specifications. Statistical multiplexing in ATM networks aims to increase the network bandwidth utilization (and lower the call blocking rate) by taking advantage of the bursty nature of traffic. Most connections will ....

Krishna Bala, Israel Cidon, and Khosrow Sohraby. Congestion control for high speed packet switched networks. In Proc. INFOCOM '90, pages 520--526. IEEE, 1990.

....bandwidth of network devices [34] and it becomes more pronounced and frequent as networks become more heterogeneous. Early debate in the 1980s centered around a few issues, namely, whether congestion control should take the form of a prior reservation or a walk in; whether it should be rate based [63, 2, 64, 60, 66, 5] or window based; whether the control should be done in the routers [43, 13] or at the end hosts [31, 33, 38, 55] and whether an open loop or a close loop mechanism would suffice. In [35] Jain presents an objective comparison of the alternatives and argues that a complete congestion management ....

BALA, K., CIDON, I., AND SCHRABY, K. Congestion control for high-speed packet switched networks. In Proceedings of Infocom (1990), pp. 520--526.

....to the high cost associated with retransmission in case of packet loss. In the context of Broadband Integrated Services Digital Network (B ISDN) with the Asynchronous Transfer Mode (ATM) transport mechanism, numerous congestion control protocols have been proposed that address the above problems [2, 3, 7, 10, 14, 21, 27, 30]. What is lacking in many of the approaches is a global viewpoint of the congestion problem as it pertains to large scale, networked systems, where it is not clear how the system as a whole will behave given the complex interactions among circuit switched packet switched traffic flows, sharing ....

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. In Proc. IEEE INFOCOM '90, pages 520--526, 1990.

....studying ATM. See Table 2 for a comparison of network layer rate control versus transport layer rate control. A well known open loop rate control algorithm is the leaky bucket, in which network access is controlled by a quota of tokens which accumulate at a fixed rate, up to the bucket capacity [2][32] Konstantopoulos et al. 30] prove that the classic leaky bucket controller provides optimal cell loss performance under finite buffer constraints. An enhancement to the leaky bucket is a marking algorithm, which allows traffic with insufficient credit to still enter the network with a ....

....provides optimal cell loss performance under finite buffer constraints. An enhancement to the leaky bucket is a marking algorithm, which allows traffic with insufficient credit to still enter the network with a violation tag, at the risk of being dropped first if it later encounters congestion [2][10] 41] Berger [4] examines an implementation in which traffic arriving at a queue with no tokens is dropped. Rubin and Lin [41] analyze a credit manager algorithm with feedback at the usernetwork access point, using a zero delay assumption Another popular network layer flow control algorithm is ....

Bala, I. Cidon, K. Sohraby, "Congestion Control for High Speed Packet Switched Networks," IEEE INFOCOM 1990, pp. 520-526.

....Once a new connection is admitted by the connection admission control, a usage parameter control function or source policing function is required to ensure that traffic submitted into the network does not exceed the parameters defined for that connection. Most of the methods consider the CLP bit [3,4,5]. In fact, ATM cells have an explicit cell loss priority bit (CLP bit) in the header so two different priority classes can be distinguished. The current ATM standard does not provide any requirements for utilization of this bit. Two possibilities should be considered when using this CLP bit: On ....

K. Bala, I. Cidon and K. Sohraby, "Congestion Control for High Speed Packet Switched Networks", IEEE INFOCOM 1990, pp520-526.

....this study we assume that service guarantees are deterministic, i.e. no guarantees are ever violated. Any network that offers service guarantees, especially guarantees on the maximum delay, requires admission control tests for new connection and policing mechanisms for existing connections [1, 4, 8, 9]. Admission control tests determine if the network has sufficient resources to accommodate a new connection without degrading the service of existing connections. The admission tests for a new connection are based on a characterization of the expected traffic for the new connection. If accepting ....

K. Bala, I. Cidon, and K. Sohraby. Congestion Control for High Speed Packet Switched Networks. In Proc. IEEE Infocom '90, pages 520--526, May 1990.

....better at ensuring smoothness because a leaky bucket may release all its tokens at once during one burst. This can be prevented by using a spacer function in the leaky bucket scheme [46] To use the bandwidth more efficiently, when the network load is light, marking may be used in an ATM network [3] [46] Cells violating the allocated bandwidth of a call, are marked red before being sent. Since bandwidth may be available in the network, marking of cells allows a call to exceed its allocated bit rate if it does not adversely affect other calls. If, at some point along its path, a marked cell ....

....data service. The Bus Control Device (BCD) handles the incoming and outgoing data of the IRB bus at the interfaces (Fig. 35) It is implemented in a PAL22V10. For incoming data, it checks the SAP address and the empty full (E F) flag to see if the interface should 1. the interface with address SAP[3 0] Empty Full 1 bit IRB port addr. 3 bit local address 12 bit IRB data 64 bit N bit 1 bit E FSAP [0 3] LCI[0 11] OWN SAP TO SLOT NR[0 11] 0,0 0(LSB) 1 16 17 20 21 32 33,34 4 5 [0 3] bit 3. Dynamic synchronous Transfer Mode An overview 42 read the data. It then issues a read ....

[Article contains additional citation context not shown here]

K. Bala, I. Cidon, K. Sohraby, "Congestion Control for High Speed Packet Switched Networks", IEEE INFOCOM, pp. 520-526, 1990.

....load balancing throughput enforcement Usage Parameter Control and Traffic Shaping Figure 3. 1: Congestion Control Approaches [3] 17 Several traffic shaping mechanisms have been proposed, such as the jumping window [8] the exponentially weighted moving average (EWMA) 34] and the leaky bucket [2, 14, 24, 35]. The basis for all these schemes is the desire to enforce the rate at which traffic cells are released into the network. The jumping window approach limits the maximum number of cells (N) that can be accepted from a source within a given time interval or window. The first N cells arriving in a ....

K. Bala, I. Cidon, and K. Sohraby, "Congestion Control for High Speed Packet Switched Networks", Proceedings of INFOCOMM 1990, San Francisco, pp. 520525, June 1990.

....for unmarked cells [24] However, buffer management mechanisms that largely prevent marked cells from interfering with unmarked cells exist. One such mechanism is realized by utilizing a second token pool for marked cells in order to limit the rate at which marked cells are transmitted [25]. Another disadvantage of marking is that it may interfere with user specifications of service requirements and priorities. The user has a better understanding of which cells are more important in delivery, whereas the marking system determines priority regardless of user level priority. This ....

K. Bala, I. Cidon and K. Sohraby, "Congestion Control for High Speed Packet Switched Networks, " Proc. IEEE INFOCOM '90, pp. 520526.

....guarantees to connections must tightly control the use of network resources by limiting both the number of connections as well as the traffic transmitted by each connection. The control of network traffic is performed by relying on admission control functions and policing functions as follows [1, 3, 7, 10, 11]: 1. Admission Control: When a network client requests the establishment of a connection, it submits a specification of its maximum traffic together with the desired service guarantees. Admission control functions in the network verify if guarantees can be given without violating any previously ....

K. Bala, I. Cidon, and K. Sohraby. Congestion Control for High Speed Packet Switched Networks. In Proc. IEEE Infocom '90, pages 520--526, May 1990.

....message and its reception at the destination, the less valuable it is in contributing to convergence. Congestion control in heterogenous, high speed networks is a difficult problem aggravated by the high propagation delay bandwidth product and the different characteristics of multi media traffic [BCS90, DMR91, Eck92, Par93]. One of the advantages of asynchronous iterative algorithms is that they do not require reliable message transmission. Therefore, they need not have access to reserved, connection oriented channels advocated for delay sensitive ATM traffic [WAMG92] Instead they can use the variable bit rate ....

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. In Proc. IEEE INFOCOM '90, pages 520--526, 1990.

....However, not all possible combinations may be acceptable. A switch that can process 1 packet per millisecond may not be able to handle 5 packets arriving back to back every 5 milliseconds (see Figure 2) Therfore, all rate based schemes, including the popular leaky bucket [2] and its variations [3,4,5,6], require specifying the burst size and the interburst interval. Analytical models of rate based controls generally ignore the two parameters and instead model the arrival process with a single parameter = n=T . Second, it is generally not understood that rate based controls are hop by hop ....

K. Bala, I. Cidon, and K. Sohraby, "Congestion Control for High Speed Packet Switched Networks," Proc. INFOCOM'90.

....into the network can be found in [11] It has been shown that the LB and the EWMA are the most promising mechanisms to cope with short term fluctuations and hence suited for policing bursty traffic. Several improvements of the LB has been proposed for increasing utilization in an ATM environment [3, 5, 15]. Traffic enforcement schemes police the source streams to check that their characteristics conform to the declared values throughout the life of the connection. The various schemes have been studied from the point of view of their capability to smooth the burstiness in the source. Traffic ....

....LBP output traffic. These characteristics motivated the development of the scheme proposed in this paper. A brief description of EWMA, a window based policer is also given for comparison with the proposed scheme. 3. 1 Leaky Bucket Scheme Leaky Bucket [18] and its variant schemes are described in [3, 5, 11, 15]. In a generalized model of the leaky bucket shown in Figure 1, tokens are generated at a fixed rate as long as the token buffer of size b is not full. When a packet arrives from the source, it is released into the network only if there is at least one token in the token buffer. This scheme ....

Krishna Bala, Israel Cidon, and K.Sohraby. Congestion Control for High Speed Packet Switched Networks. In IEEE INFOCOM, pages 520--526, 1990.

....discarding is cell violationtagging [2, 3] The violating cells are identified and tagged. The tagged cells are selectively discarded later in the network as a last resort to alleviate congestion in the backbone network. Schemes to regulate both the untagged and tagged traffic have been proposed [1, 3]. However, due to the bursty and unpredictable nature of broadband traffic, even with violation tagging, the pure preventive scheme alone is not sufficient to satisfy the two conflicting goals: the QoS guarantee and the network efficiency. Without any control on tagged traffic, a violating user ....

....conflicting goals: the QoS guarantee and the network efficiency. Without any control on tagged traffic, a violating user can potentially dominate the usage of the available resources and cause congestion in the network. On the other hand, exercising stringent rate control on tagged traffic, as in [1], can potentially reduce network efficiency. A compromise between the efficiency and the QoS guarantee can be difficult to achieve. Reactive control schemes, on the other hand, allow the network to experience some congestion and take corrective action to relieve congestion. A typical reactive ....

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. INFOCOM '90, pages 344--349, 1990.

.... (in the sense mentioned above) i.e. for matching the source speed to the processing speed of the destination in a point to point scenario (e.g. in the data link layer) The applicability of window based flow control schemes to high speed networks is addressed in a number of references [Kle92, BCS90]. The problems associated with this approach mostly stem from the large normalized propagation delays across the network which render the feedback information obsolete 10 The number of packets that can be outstanding in the network at a time. 11 i.e. data link, network, and transport. and ....

....leaky bucket is a preventive technique, operates in the transport layer (buckets are allocated on a per connection basis) is not based on feedback, and admits packet loss. Various forms of the leaky bucket scheme have been proposed in the literature. One variation of this technique [BCS90] deals with two categories of packets which are marked by the source as green or red. Green packets are transmitted at the rate negotiated during the call setup. Red packets represent the rate in excess of the contract and are handled differently at the intermediate nodes, according to the ....

[Article contains additional citation context not shown here]

K. Bala, I. Cidon, K. Sohraby, "Congestion Control for High Speed Packet Switched Networks ", INFOCOM'90, 520--526.

....iterative algorithms, and in subsequent sections we will concentrate on the asynchronous case. Congestion control in heterogenous, high speed networks is a difficult problem aggravated by the high bandwidth delay product and quality of service (QoS) requirements of real time, multimedia traffic [1, 9, 10, 25]. One of the advantages of asynchronous iterative algorithms is that they do not require reliable message transmission. Hence, conditional on the severity of the packet loss rate, they may do without the access to reserved, high priority connections advocated for constant bit rate (CBR) and ....

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. In Proc. IEEE INFOCOM '90, pages 520--526, 1990.

....proposed in ( Tur86] and the credit manager scheme used in SMDS ( SLCG89] are examples of input rate regulation schemes. The PARIS rate control mechanism is a buffered version of previously suggested leaky bucket scheme with additional components (e.g. a spacer) as is described and analyzed in [BCS90, SLCG89, GG92]. In general, the scheme guarantees that the long term average rate does not exceed the pre specified rate of the connection. Over shorter periods, it permits bursts at a much higher rate which is constrained by the maximum speed of the communication links in the path. 3 Control cycle The PARIS ....

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high speed packet switched networks. In INFOCOM'90, June 1990. 85] A. E. Baratz, J. P. Gray, P. E. Green Jr., J. M. Jaffe, and D. P. Pozefsky. Sna networks of small systems. IEEE Journal on Selected Areas in Com., SAC-3(3):416--426, May 1985.

....as route computation, call admission, etc. 11, 14, 23] and is readily achieved by using the TREE routing mode. 3. 2 Delay and Loss Priority Another aspect of the heterogeneity of the transport architecture is the support for different delay and loss priority classes within the plaNET network [15, 20, 22, 35, 36]. plaNET currently defines three delay classes and two loss priority classes within each delay class. Delay classes are served using a nonpreemptive exhaustive service rule, i.e. a given delay class is only served when no packets from a higher priority class are waiting, while a threshold ....

....functions. 8.1.3 Access Control In the plaNET ORBIT network, the access control is distributed to the end points, and this traffic regulation function is typically performed in the adaptor attached to the ORBIT LAN (see Figure 9. This access control consists of a generalized Leaky Bucket (LB) [12, 15, 23, 35, 36], which determines when the packet waiting to be transmitted can be scheduled to access ORBIT. This mechanism is implemented in software, which provides for both flexibility and the ability to handle a large number of connections. The software implementation, rather than emulating the traditional ....

K. Bala, I. Cidon and K. Sohraby, "Congestion Control for High-Speed Packet Switched Networks," Proc. INFOCOM'90, San Francisco, CA, 1990.

....difficult to control the aggregate volume of excess traffic offered at any time to each network link. It is, therefore, important to protect regular packets from potential low priority traffic surges. A number of space priority policies have been proposed and analyzed, in the literature, e.g. [10, 4]. The emphasis 1 of these works has, however, mainly been on determining acceptable load regions as a function of the loss requirements of each priority class. This paper differs from these earlier works as it focuses on understanding and defining when and under which assumptions, performance ....

....infinite length sample path, for which the reference system is lossless and the main system loses M Gamma1 N M of the green cells. Note that the result holds for both the strong and weak definitions of the protective property. 3. 2 The Limited Red Policy The limited red policy was suggested in [10] as another policy for protecting green traffic from excessive red traffic, while providing an acceptable level of service to red cells. This policy allows the concurrent buffering of up to L red cells in the system. In other words, an arriving red cell is accepted if and only if there is a free ....

[Article contains additional citation context not shown here]

K. Bala, I. Cidon, and K. Sohraby. Congestion control for high-speed packet switched networks. In Proc. INFOCOM'90, San Francisco, 1990.

No context found.

K. Bala, I. Cidon, K. Sohraby, "Congestion Control for High Speed Packet Switched Networks, " IEEE ICC'90 Conf.

No context found.

Bala, K., Cidon, I., and Sohraby, K., "Congestion Control for High Speed Packet Switched Networks ", INFOCOM '90, pp. 520-526, 1990.

No context found.

K. Bala, I. Cidon, and K. Sohraby. Congestion Control for High Speed Packet Switched Networks. In Proceedings of INFOCOM'90, pages 520--526, June 1990.

No context found.

Bala, K., Cidon, I., and Sohraby, K., "Congestion Control for High Speed Packet Switched Networks ", INFOCOM '90, pp. 520-526, 1990.

No context found.

Bala, K., Cidon, I., and Sohraby, K., "Congestion Control for High Speed Packet Switched Networks", INFOCOM '90, pp. 520-526, 1990.

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