Author : Andreas, Laudria; Hafidudin, Hafidudin; Bogi A.K., Nyoman;
In an ATM-based network, Call Admission Control (CAC) is required to reserve resocurces in advance. In case of a multimedia calls, each of its substream (i.e., video, audio, and data) has its own distinct Quality of Services (QoS) requirements (e.q., cell loss ratio, delay, jitter, etc). The network attempts to deliver the required QoS by allocating an appropriate amount of resources (e.q., Bandwidth (BW), buffers). In static allocation approach, the negotiated QoS requirements constitute a certain QoS level that remains ficed during the call. Accordingly, the corresponding allocated resources also remain unchaged. This paper will present an adaptive allocation of resources algorithm based on Genetic Algorithm (GA). In contrast to the static approach, each call declares a range of an acceptable QoS levels, instead just a single one. As the availability of resources in network varies, the algorithm selects the best possible QoS level that each call can obtain. This is done, however, under the constraint of achieving maximum utilization of the resources while simultaneously distributing them fairly among the calls and also to lower the blocking probability of the network. The degradation is limited to a minimum value predefined in User-Defined Profile (UDP). From the user perspective, the perception of the QoS degradation is very graceful and happens during overload periods. The network services, on the other hand, are greatly enhanced due to the fact that the call blocking probability is significantly decreased, while BW utilization increased.Simulation results demonstrate that the proposed algorithm performs well in terms of increasing the number of admitted calls while utilizing the available BW fairly and effectively.
Keyword : ATM, Genetic Algorithm, bandwidth, QoS
Sumber : http://repository.petra.ac.id/64/