Multiuser capacity for heterogeneous QoS constraints in uncorrelated Rayleigh channels

In this work, we derive analytical expressions for the maximum attainable user rate by accomplishing a probabilistic delay constraint in an uncorrelated Rayleigh channel. These rates are derived for two representative scheduling schemes, round robin (RR) and best channel (BC) allocation, but presented tools can be applied to any scheduling policy. On the basis of the effective bandwidth theory, the result follows the same procedure as, where the capacity with probabilistic delay constraint C_{D t ,¿} was defined as the maximum source rate that may be supported in a single-user wireless system with a probability ¿ of exceeding certain delay bound D^t. The extension to a multiuser system and the evaluation of the user rates show to what extent the throughput maximization of BC allocation is attained at the expense of fairness, whereas in the RR scheduling both the differences among users and the total system capacity are smaller. Simulations are finally performed in order to validate our results.