Call Admission and Resource

Many audio and video \play-back" applications require that all of their packets arrive within their play-back time. Furthermore, the applications with the hard real-time requirements also require guaranteed service from the network. These applications are covered under the \Guaranteed Quality of Service" speciications by the IETF. It guarantees that the packets will arrive within the guaranteed delivery time and the packets will not be discarded due to queue overrows, provided the ow traac stays within its speciied traac parameters. This paper gives an eecient and distributed algorithm based on the cost function to divide the end-to-end guaranteed QoS requirements into local QoS requirements which are then mapped into local resource requirements. The algorithm operates in three phases. In the rst phase, the nodes in the selected route give the information about their utilization level and the parameters associated with the scheduling algorithm used by them. The receiving node, based upon the information provided by the nodes in the route and its end-to-end delay requirements, takes the call admission decision. Once the decision to admit the call is taken, the nodes in the route do the resource allocation to meet the delay bound. The sender node now calculates the slack produced if any. The slack is then distributed to the nodes in the third phase. The results obtained are then shown to be applicable to the resource allocation for multicast connection establishment.