S. Akhtar, "'Congestion Control In A Fast Packet Switching Network', Master's thesis, Washington University, 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....traffic descriptor is the set of traffic parameters in the source traffic descriptor, the cell delayvariation tolerance and the conformance definition. The conformance definition is used to decide which cells are conforming in the connection. A typical conformance definition is the leaky bucket [1]or Generic Cell Rate Algorithm ( GCRA ) although many such algorithms may be used in tandem. The CAC will use the connection traffic descriptor to allocate resources and to derive parameters for the UPC. Any connection traffic descriptor must be enforceable by the UPC. Even though a cell is found ....

S. Akhtar, "'Congestion Control In A Fast Packet Switching Network', Master's thesis, Washington University, 1987.

....with a certain probability greater than a given value. 14 between different classes. 1.3.7 Other works There are a number of other papers that have addressed the problem of resource allocation in ATM networks. A solution similar to the class related rule uses the concept of virtual bandwidth [Akhtar 87] Resource allocation at multiple levels, namely the connection level, the burst level, and the cell level have also been proposed [Hui 88] A simple admission control rule for admitting bursty traffic is mentioned in [Esaki 90] and [Todorova 90b] In the realm of packet switching networks, the ....

S. Akhtar, "Congestion Control in a Fast Packet-Switching Network", Master's Thesis, Washinngton University, December 1987.

....hand though, seldom will a channel only carry the traffic of a single service type. Therefore, in order to determine the equivalent bandwidth to be allocated to each pair w 2 W to carry the mixture of different traffic services offered to the pair, we shall use the linear approximation proposed in [7]. By this approximation, each connection of a service type s 2 S, sharing a channel with connections of other types, will require the same bandwidth needed as if it were sharing the channel with connections of its own type s. Thus, the equivalent bandwidth required by a service s connection shall ....

Shahid Akhtar, "Congestion control in a fast packet switching network", Master's thesis, Washington University, December 1987.

....the acceptance criterion is the existence of link bandwidth equal to the equivalent bandwidth. The equivalent bandwidth is defined as the bandwidth required to satisfy the requested QoS, given a particular buffer size and set of active sources. The use of equivalent bandwidth is exemplified in [1, 17, 22]. In [17, 22] simulation is used to obtain the equivalent bandwidth for the case of homogeneous on off sources with a given mean rate, peak rate and mean burst length. The equivalent bandwidth method is effective and simple for the case of homogeneous sources. For heterogeneous sources, however, ....

....the precise bandwidth needed for each source. Approximate methods must be used to estimate bandwidth requirements. Many of the approximate methods try to use the analysis for the homogeneous case to approximate the non homogeneous case. One of the simplest, called the linear approximation method [1], assumes that the bandwidth needed for a source is the same as the bandwidth that the source would require if the whole available bandwidth were occupied by homogeneous sources of the same type as the source under consideration. It has been shown by simulation that the linear approximation method ....

[Article contains additional citation context not shown here]

S. Akhtar "Congestion Control in a Fast Packet Switching Network", Master Thesis, Washington University in St. Louis, Dec. 1987.

....and the distribution of the autocorrelation function. Burstiness can also be represented by the squared coefficient of variation of the interarrival times, which is the ratio of the variance of the cell interarrival times to the square of the average value of the cell interarrival times. Akhtar [2] defines burstiness in the IPP model as = ff fi) where is the bit rate in the active state and 1=fi and 1=ff are the mean holding times in the active and idle states, respectively. Similar to the case of the definition of burstiness, there is no commonly accepted definition for smoothness ....

S. Akhtar, "Congestion Control in a Fast Packet Switching Network", M.S. Thesis, Washington University, St. Louis, Missouri, 1987.

....in transit between Leland Window Based Congestion Management 2 one source and its destination. As an alternative approach to access control, rate based windows [3] leaky buckets or input throttles ) have recently received much attention in the industrial [4] 5] 6] 7] and academic [8] [9] 10] research communities. A leaky bucket forces a traffic source not to exceed some specified average rate of traffic input (over some interval) and some specified burst length (maximum sequence of consecutive cells at the peak input rate) Within those constraints, however, the accepted ....

Shahid Akhtar, Congestion Control in a Fast Packet Switching Network, Masters Thesis, Washington University (December, 1987).

....burst length (i.e. the mean duration of the time interval during which the traffic source transmits at the peak rate) 5] 3. Burst factor defined as the average number of bits accumulated in a buffer during a burst, namely, Peak bit rate 0 Average service bit rate) 2 Average burst duration [6] 4. Cell jitter ratio defined as the variance to mean ratio of the cell interarrival times, namely, Var [Cell interarrival times] E [Cell interarrival times] 7] 5. The squared coefficient of variation for the interarrival times, namely, Var [Cell interarrival times] E 2 [Cell interarrival ....

....much buffer space since each short burst can be easily handled in a small buffer space, while cell stream B may require a large amount of buffer space corresponding to its larger average burst length. The example in Figure 1 suggests that one key consideration is the burst length. It is shown in [1, 5, 6, 10] that the burst length significantly affects network performance; the longer the burst length, the worse the network performance becomes. Some number of papers have investigated the effects of burstiness. A few general conclusions can be made about its effects. 1. Bursty traffic should only be ....

S. Akhtar, "Congestion Control in a Fast Packet Switching Network," Master's Thesis, Washington University, December 1987.

....sharing identical loss requirements. They are based on the following approach for a given loss requirement, determine the maximum number of sources, n max , that can share a link so that they all satisfy their requirements. Then c = r=n max . This can be obtained either through analysis, [1], or simulation, 16] Observe that the test for call admission can be stated in the following equivalent form, is n 1 n max Here n denotes the number of sessions currently using the link. Two types Number of class 2 sessions Number of class 1 sessions Feasible region Figure 7: Feasible ....

S. Akhtar. Congestion control in a fast packet switching network. Master's thesis, Washington University, St. Louis, Missouri, 1987.

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