| Call admission control (CAC) extends the capabilities of Quality of service (QoS) tools which protect voice traffic from the negative effects of other voice traffic. It does not allow oversubscription of a Voice over Internet Protocol (VoIP) network. To achieve better performance for efficient call admission control, various dynamic routings are being proposed. In the dynamic routing mechanism, the condition of the network is learned by observing the network condition via the probe packets and according to the defined threshold, routes are chosen dynamically. In such schemes, various combination of route selection is used such as two routes are used where one is fixed and other is random or two random routes are chosen and after observation one is chosen if it passes the test. Few schemes use a route history table along with the two random routes. But all have some issues like it selects random routes (not considering the number of hops), does not process memorization before admission threshold test, it calculates all selected paths regardless of the fact that they are selected or not, thereby wasting central processing unit (CPU) time and since these uses two routes so obviously the call admission probability is less.; In this thesis work, a new dynamic routing scheme is proposed which considers a routing history table with endpoint admission control increasing the call admission probability, makes call establishment time faster and it saves valuable CPU resources. The proposed scheme considers a combination of three routes with routing history table---one is the direct route and the other two are selected randomly from all available routes and the routing history table is used to memorize the rejected calls. CAC tests like Admission Threshold were performed on the selected routes. Various parameters such as delay, packet loss, jitter, latency etc from the probe packets are used to carry out the tests. Performance of the proposed scheme with respect to other dynamic routing schemes is studied using a mathematical/analytical model. Also, effect of arrival rate probe packets on utilization, busy period, waiting period, acceptance probability of calls, probe packets, and the number of successful calls was also studied. |
