Uplink Scheduling with Quality of Service in IEEE 802.16 Networks

1. Uplink Scheduling with Quality of Service in IEEE 802.16 Networks Juliana Freitag and Nelson L. S. da Fonseca State University of Campinas, Sao Paulo, Brazil IEEE Global Telecommunications Conference, 2007. GLOBECOM '07. Mei-Jhen Chen

2. Outline • Introduction • A scheduling Mechanism • Simulation Experiments • Conclusions

3. Introduction • To support a wide variety of multimedia applications, the IEEE 802.16 standard defines four types of service flows, each with different QoS requirements. • Each connection between the SS and the BS is associated to one service flow. • UGS (Unsolicited Grant Service) • rtPS (real time Polling Service) • nrtPS (non-real time Polling Service) • BE (Best Effort)

4. Introduction (cont.) • A signaling mechanism for information exchange between the base station (BS) and subscriber stations (SSs) was defined. • allow the SSs to request bandwidth to the BS • Bandwidth allocation is provided on demand • When an SS has backlogged data, it sends a bandwidth request to the BS. • The BS allocates time slots to the SS. • Each frame is divided in two parts • the downlink subframe and the uplink subframe

5. Introduction (cont.) • different scheduling mechanisms have been proposed • not all of them comply with the IEEE 802.16 standard • Authors introduce a BS uplink scheduling algorithm which allocates bandwidth to the SSs based on the QoS requirements of the connections. • The proposed policy is fully standard-compliant and it can be easily implemented in the BSs.

6. A Scheduling Mechanism • According to the IEEE 802.16 standard • the BS uplink scheduler provides grants (time slots) at periodic intervals to the UGS flows to send data. • Periodic grants are also given to rtPS and to nrtPS flows to request bandwidth. • the uplink scheduler must guarantee that the delay and the bandwidth requirements of rtPS and nrtPS flows are met. • The BS executes the uplink scheduler during each frame, and it broadcasts the schedule in the UL-MAP message in the downlink subframe.

7. A Scheduling Mechanism high priority queue intermediate queue low priority queue

8. A Scheduling Mechanism

9. A Scheduling Mechanism -CheckDeadline current_time = 8ms Frame_duration = 5ms high priority queue UGS request rtPS request nrtPS request BE request intermediate queue Server UL-MAP 3 2 1 3 2 1 deadline = 18ms 13ms 20ms frame[i] : 2 1 2 low priority queue

10. A Scheduling Mechanism -CheckMinimumBandwidth current_time = 8ms Frame_duration = 5ms high priority queue UGS request rtPS request nrtPS request BE request 2 intermediate queue Server UL-MAP 3 2 1 3 1 BWmin : 10 13 11 20 24 granted_BW_tmp : 15 8 10 15 14 backlogged_tmp : 16 17 13 12 15 low priority queue Priority[i] = backlogged_tmp[CID] – (granted_BW_tmp[CID]-BWmin[CID]) 0 12 18 17 25

11. Simulation Experiments -assumption • NS-2 • 250x250 meter area • a BS：located at the center • the SSs：uniformly distributed • The frame duration：5 ms • the capacity of the channel：40 Mbps

12. Simulation Experiments -assumption • Each SS has only one traffic flow. • four types of traffic • Voice • “on/off”： a mean of 1.2 s and 1.8 s • During “on” periods, packets of 66 bytes are generated every 20 ms • Video • real MPEG traces • FTP • a hybrid Lognormal/Pareto distribution • an area of 0.88 with a mean of 7247 bytes • the tail is modeled with a mean of 10558 bytes • WEB • exponential distribution with a mean of 512 KBytes.

13. Simulation Experiments -assumption • UGS • The interval between data grants is 20 ms since the BS • rtPS • The interval is 20 ms • the delay requirement is 100 ms • nrtPS • the interval of the nrtPS service 1 s. • minimum bandwidth requirement of 200Kbps • BE • not have any QoS requirement

14. Simulation Experiments -experiment 1 • Whether the BS is able to allocate bandwidth to connections in the same service level in a fair way, regardless the number of connections in the network. • one BS • the number of SSs varies between 10 and 30 • Each SS has one nrtPS connection that generates FTP traffic with rate of 600 Kbps.

15. Simulation Experiments -experiment 1 Fig. 2. Throughput of each SS

16. Simulation Experiments -experiment 2 • whether or not the increase of the UGS traffic load degrades the QoS level of services with lower priority. • one BS and 81 SSs • 6 rtPS connections • 20 nrtPS connections • 20 BE connections • the number of active UGS connections varies from 15 to 35.

17. Simulation Experiments -experiment 2 Fig. 3. Delay of UGS and rtPS connections

18. Simulation Experiments -experiment 2 Fig. 4. Throughput of nrtPS and BE connections

19. Simulation Experiments -experiment 3 • the impact of the load increase of the rtPS service on the performance of other service classes. • one BS and 62 SSs • 15 UGS connections • 20 nrtPS connections • 20 BE connections • the number of active rtPS connections varies from 1 to 7

20. Simulation Experiments -experiment 3 Fig. 5. Delay of UGS and rtPS connections

21. Simulation Experiments -experiment 3 Fig. 6. Throughput of nrtPS and BE connections

22. Simulation Experiments -experiment 4 • whether the increase of the BE traffic load influences or not the QoS level of services which has higher priority. • one BS and 70 SSs. • 15 UGS connections • 5 rtPS connections • 15 nrtPS connections • the number of active BE connections varies from 10 to 35.

23. Simulation Experiments -experiment 4 Fig. 7. Delay of UGS and rtPS connections

24. Simulation Experiments -experiment 4 Fig. 8. Throughput of nrtPS and BE connections

25. Conclusions • An uplink scheduling mechanism for IEEE 802.16 networks was introduced. • The proposed solution supports the four service levels specified by the standard and considers their QoS requirements for scheduling decisions. • The complexity of the proposed mechanism is O(k + rlogr) k : number of slots in the uplink subframe r : the number of rtPS and nrtPS bandwidth requests in the intermediate queue

26. Thank You!