Approximate blocking probabilities in loss models with independence and distribution assumptions relaxed

Effective approximations are developed for the blocking probability in a general stationary loss model, where key independence and exponential-distribution assumptions are relaxed, giving special attention to dependence among successive service times, not studied before. The new approximations exploit heavy-traffic limits for the steady-state number of busy servers in the associated infinite-server model with the same arrival and service processes. In addition, a new heavy-traffic approximation is developed for the long-run proportion of time that all servers are busy. These new approximations are then combined to develop new approximations for the separate blocking probabilities of individual arrival streams in multiclass loss models. Simulation experiments show that these approximations are effective.