A call admission control scheme in ATM networks

Overview of measurement-based connection admission control methods in ATM networks

“…Several CAC methods calculate the CLR, use REM, and are traffic-descriptor-based [12][13][14][15][16][17][18][19]. Murase et a1.…”

“…pointed out that the CLRs observed for individual connections differ from the aggregate CLR [12]. Miyao developed a fast algorithm for calculating the aggregate CLR under heterogeneous traffic conditions by using the Chernoff bound [13]. Saito developed a CLR upper-bound formula [14], and Shioda and Saito gave the mathematical proof for it [14,19].…”

“…This situation has improved with (1) several lightweight CLR calculation algorithms [13,15,16,18] that can make an admis sion decision within a constant time (i. e. , not proportional to the number of connections) and (2) • Figure 10. Effect of peak rate and burstiness factor on bandwidth efficiency using TG93.…”

Evaluation of overflow probabilities in resource management

“…A number of other upper bounds can be found in literature ([71 [81 [91 [lo]). All of these bounds are based on a Chemoffqpe bound [6] which states that for any random variable Y: Instead of implementing the numerical method for solving a complex equation, we tried to compare our approximation method by using the same workload that was used in [7]. The workload consisted of a mix of two types of connections, one with a pi of 0.5 and a wi of 0.05, and the other one consisted of connections with a p i of 0.1 and a wi of 0.05.…”

“…Then, in Section IV, we compare it to the approximation method proposed in [7], which obtains the best possible Chemoff bound that can be used. We will present the algorithms in the context of real-time communication, that is, evaluate the probability of overflow when some number of real-time connections are established at a node.…”