A FAST ADAPTIVE PROPORTIONAL FAIRNESS ALGORITHM FOR A MIXTURE OF REAL-TIME AND NON-REAL-TIME DATA IN CDMA/HDR

摘要: Even if PF (Proportionally Fair) scheduler efficiently provides data services by exploiting multi-user diversity, it cannot guarantee real-time users QoS (Quality of Service) such as the delay bound or the minimum throughput. In order to solve this problem, the M-LWD and EXP rules were proposed on [1], which introduced a supplementary adaptation term into PFS. The improvement of the delay performance is maximized when the adaptation term immediately converges to its equilibrium state where each user’s service rate is equal to each arrival rate except non-real-time users. If not, the bad channel user’s data will be backlogged until the equilibrium state, which cause poor delay performance for a long time. A proposed algorithm makes the adaptation term more fast converge into its equilibrium state than MLWDF and EXP. Simulations show that the delay performance is considerably improved comparatively with M …