Approximate performance analysis of resource allocation for real-time and non-real-time multiclass services in cellular systems

An approximate analytical formulation of the resource allocation problem for handling real- and nonreal-time multiclass services in a cellular system is presented. Real-time handoff connections that arrive when the channels are fully utilized are blocked while those for non-real-time handoff connections are queued. Each class has its own transmission bandwidth requirement. A complete sharing scheme is used for the resource sharing policy. All non-real-time handoff connections are queued in a buffer when there is no channel resource available. Guard channels are also considered in the model. Guard channels and a queue for non-real-time handoff connections help to achieve higher probabilities of successful handoff connections. The model is solved using a K-dimensional Markov chain. Connection and packet level quality of services (QoSs) are considered. The QoS performance metrics are connection blocking probability, system utilization and packet loss probability. Numerical results illustrate the handoff connection blocking probabilities drop by several orders by having queued non-real-time handoff connections and the total system utilization is only increased slightly. Numerical results illustrate that higher gain in system utilization is achieved through the joint optimization of connection/packet levels.