Blocking Probability Estimation for General Traffic Under Incomplete Information

Abstruct- A simple, robust, explicite upper bound is derived on the blocking probability for general multirate traffic, dropping a number of traditional assumptions, such as Poisson arrivals, while still maintaining optimally tight exponent of the estimation. The new approach also makes it possible to estimate blocking probability under incomplete information. Furthermore, it remains valid in situations when the individual call bandwidth demands aggregate in complex, nonlinear ways, e.g., in case of compressible flows, priority classes or processing constraints. We show that the bound is easily applicable for fast, robust link dimensioning. Moreover, it is very well fitted for embedding into more sophisticated network optimization problems, due to its convexity properties.