A new approach to connection admission control is used to optimize the weighted sum of blocking across traffic classes, where weights signify the relative importance between classes. An optimal upper limit policy is developed from an asymptotic approximation for large networks. The result is a very simple algorithm that could be implemented on standard network hardware. This upper limit policy and weighted sum of blocking metric is then used to perform the first validation of the need for prioritized traffic by demonstrating that the use of an upper limit policy is superior to traditional approaches of adding extra capacity or partitioning capacity, both in terms of the amount of resources required and in sensitivity to load variations.
To recognize which connections are more important, an architecture of geographically distributed ticket servers is proposed that issues importance tickets in the current context of an emergency or natural disaster. User agents interact with ticket servers using an agent communication language. The ticket server architecture is shown to provide high user satisfaction through negotiation flexibility and adaptation to the dynamic context in which the network operates. The ticket server architecture is also shown to not have prohibitive implementation complexity, nor does it cause excessive additional connection setup delay. A hybrid performance analysis approach is used to simulate negotiation interactions with the ticket server using a system prototype, then analytical queuing network models are used to determine server processing rate requirements and connection setup delays.
Prioritized resource allocation is effective because it uses resources effectively and responds well to the dynamics of the network and the context in which it operates. Important activities are given preferred access to resources so that the important needs of society can be addressed when disasters or other special needs arise.