Evaluation of the New and Accepted Customers Blocking Probabilties in a Network of Resource Loss Systems

The paper considers a network of resource loss systems (ReLS) with random resource requirements and two types of nodes. Customers initially arrive to the first type of nodes, where they receive service for exponentially distributed time. The service of customers can be interrupted. In this case, they are rerouted to the second type of nodes, where they receive service for an exponentially distributed time. Once the service is completed, they return back to the original node and continue its service. Customers require a random volume of limited resources. If there are not enough of unoccupied resources upon the arrival of a customer, then it is considered lost. Similarly, if an accepted customer is rerouted to another node and finds that there are not enough of resources to meet its requirements, then it is also lost. In this paper, we provide an approach to analyze the stationary behavior of the considered system, as well as establish expressions for the new customer loss probability and the accepted customer loss probability. The developed model has a wide range of applications in performance evaluation of fifth generation (5G) New Radio (NR) access networks. To this aim, we investigate the response of the considered service system in detail by revealing critical dependencies and trade-offs between input system parameters and performance measures of interest.