1. A Joint Channel and Queue-Aware Scheduling for IEEE 802.16 Wireless Metropolitan Area Networks Mehri Mehrjoo, Xuemin Shen, Kshirassgar Naik WCNC 2007

2. Agenda Introduction Opportunistic stable queue (OSQ) scheduling scheme Simulation results Conclusions 2

3. Introduction(1/3) IEEE 802.16 WMANs have gained considerable attractions Provides wireless broadband services over a large coverage area WMANs will be deployed in a broad set of applications Last-mile and backhaul network 3

4. Introduction(2/3) Transmission scheduling process is critical The promising variety of applications and the volume of traffic to be carried by each node Scheduling scheme is an open issue No scheduling scheme has been standardized or commercialized for IEEE 802.16 WMANs 4

5. Introduction(3/3) Goal Provides a scheduling scheme for the DL of the IEEE 802.16 WMANS Channel-aware Diverse wireless channel status Proportional fair scheduling scheme Queue-aware The unbounded delay may cause overflow for queue Queue length is finite 5

6. Opportunistic stable queue (OSQ) scheduling scheme(1/10) Stable ratio for a queue In order to maintain the queues absolutely stable, the scheduler should keep the ratio greater or equal to one for a queue 6

7. Opportunistic stable queue (OSQ) scheduling scheme Maintaining absolute stability may not be always possible,especially bursty traffic source A bursty traffic source lets a large volume of traffic into network over short time interval Non-real-time applications 7

8. Opportunistic stable queue (OSQ) scheduling scheme(2/10) For average arrival rate The scheduler deploys an estimation of the arrival traffic that filters the fast variations For average departure rate The scheduler provides an estimation of the current available rate and previously transmitted rate to each user 8

9. Opportunistic stable queue (OSQ) scheduling scheme(3/10) Solutions EWMA(Exponentially Weighted Moving Average) technique A weighing average technique that assigns smaller weights to old data 9

10. Opportunistic stable queue (OSQ) scheduling scheme(4/10) Partial stability for queue j R j (n): The EWMA departure rate of queue j,which is measured of the departure history from queue j 10

11. Opportunistic stable queue (OSQ) scheduling scheme(5/10) Partial stability for queue j W j (n): The EWMA arrival rate of queue j,which is measured of the arrival history to queue j 11

12. Opportunistic stable queue (OSQ) scheduling scheme(6/10) Stable Queue(SQ) scheduling The lower is the ratio for user j,the higher is the probability of transmission to that user 12

13. Opportunistic stable queue (OSQ) scheduling scheme(7/10) Assumption The variations of the channel of each SS are mutually independent from the others in a multiuser environment Some SSs may experience deep fading while others may not Optimal scheduling policy Transmit to a group of SSs with the best channel quality at maximum achievable rates 13

14. Opportunistic stable queue (OSQ) scheduling scheme(8/10) Joint consideration of channel status and queue status mitigate the problem Scheduling cost function 14 Achievable rate of SS j in the scheduling interval from n to n+1 Normalized value of for queue j

15. Opportunistic stable queue (OSQ) scheduling scheme(9/10) OSQ scheduling scheme in IEEE 802.16 Assumption The channels are slowly varying The estimated channel gain values are valid for the next frame transmissions The estimated channel gain values by each SSs are reported back to BS before every DL subframe begins 15

16. Opportunistic stable queue (OSQ) scheduling scheme(10/10) OSQ scheduling scheme in IEEE 802.16 The BS updates the vector of W j before each Downlink subframe transmission The BS updates the vector of R j after each Downlink subframe transmission The scheduler sorts transmissions to SSs based on the descending order of 16

17. Simulation results(1/3) Frame specification parameters 17

18. Simulation results(2/3) 18 OPP: Pure Opportunistic MLWDF: Modified Largest weightd delay first

19. Simulation results(3/3) 19 OPP: Pure Opportunistic MLWDF: Modified Largest weightd delay first

20. Conclusions This paper proposed a scheduling scheme Channel-aware Queue-aware Simulation results demonstrate that the scheduler can significantly control the queue length maintain a good level of throughput for the network 20