These frameworks employ a single kind of scheduler for all types of WiMAX traffic.
Simulations have been used to evaluate the QoS performance of WiMAX. DRR has been selected as a DL scheduler and WRR has been chosen as an UL scheduler. Similarly, both WRR and DRR have been selected, independently, to be employed at SSs. This chapter does not consider the application of a wide variety of other schedulers for the various traffic classes.
Two sets of simulation experiments are performed: first EDF for all traffic classes and second WFQ for all traffic classes. It has been reported in these studies that, if the total traffic load is under 100%, both schedulers can satisfy traffic class performance requirements. However, when one traffic class starts consuming more than its fair share of bandwidth, EDF favors streams with more crucial deadlines, whereas WFQ strives for fairness and punishes the flow that is unfair. It has been concluded that using either EDF or WFQ for all traffic classes is not optimal.
Based on the fair queueing model proposes a hierarchical scheduling model for WiMAX traffic classes. It specifies three types of scheduling servers, viz., hard-QoS, soft-QoS, and best effort servers. UGS traffic is mapped into the hard-QoS server and rtPS traffic into the soft-QoS server; whereas nrtPS can be mapped into either. All servers implement the WF2Q algorithm.
A token bucket based call admission control and an UL packet scheduler employing EDF. The focus of their algorithm is to meet the bandwidth and delay requirements of rtPS flows. It employs the arrival service curve and a two-dimensional rtPS database. However, this algorithm does not consider the ertPS traffic class. A token rate estimation model, for Poisson traffic fed to both an infinite and a finite queue, has been developed by employing well-known Markov chain models. The integration of admission control and scheduling has been left as future work. Thus, this approach needs to be investigated further before it can be adopted for use in WiMAX. It is also known that a potential problem with token bucket-based schemes involves allowing large bursts of traffic, for which a leaky bucket can be arranged after the token bucket
