1. 1 Survey of Admission Control of Supporting VoIP Services in IEEE 802.11e QoS-enabled WLAN R93725003 卓德忠 R93725010 鍾佳芳 2 Jan. 2006

2. 2 Outline Introduction Introduction Admission Control Admission Control Parameterized EDCA Parameterized EDCA Conclusions Conclusions

3. 3 Introduction Motivation Motivation Introduction of 802.11e features Introduction of 802.11e features Introduce the reference design of HCCA in 802.11e Introduce the reference design of HCCA in 802.11e

4. 4 Motivation 802.11e is an enhanced QoS support in WLANs 802.11e is an enhanced QoS support in WLANs The most promising framework among QoS enhancements of WLANs The most promising framework among QoS enhancements of WLANs The contention-based MAC access scheme is hard to provide quality of service (QoS) assurance for VoIP services. The contention-based MAC access scheme is hard to provide quality of service (QoS) assurance for VoIP services.

5. 5 Introduction of 802.11e IEEE 802.11 WG, “ Draft Supplement to Standard for Telecommunications and Information Exchange between Systems-LAN/MAN Specific Requirements — Part 12: Wireless MAC and PHY Specifications: MAC Enhancements for QoS, ” IEEE 802.11e/draft 12.0, Nov. 2004. Qiang Ni, ” Performance Analysis and Enhancements for IEEE 802.11e Wireless Networks, ” in IEEE Network, July/August 2005 Qiang Ni, ” Performance Analysis and Enhancements for IEEE 802.11e Wireless Networks, ” in IEEE Network, July/August 2005

6. 6 Introduction of 802.11e A new MAC layer function called the hybrid coordination function (HCF) is proposed. HCF uses a contention-based channel access method, also called enhanced distributed channel access (EDCA) Polling-based HCF-controlled channel access (HCCA) method Transmission opportunity (TXOP) refers to a time duration during which a QSTA is allowed to transmit a burst of data frames EDCA-TXOP HCCA-TXOP

7. 7 MAC Architecture for QoS L.W Lim, R. Malik, P.Y. Tan, C. Apichaichalermwongse, K. Ando, Y. Harada, “ A QoS scheduler for IEEE 802.11e WLANs ”, First IEEE Consumer Communications and Networking Conference, 2004. Jan 2004, pp.199 – 204

8. 8 HCCA Features 1. 1. Different traffic classes called traffic streams (TSs) are introduced in HCCA. 2. 2. QSTA is not allowed to transmit a packet if the frame transmission cannot finish before the next beacon 3. 3. TXOP Limit is used to bound the transmission time of a polled QSTA. In order to initiate a TS connection, a QSTA sends a traffic specification (TSPEC) to the QAP. A TSPEC describes the QoS requirements of a TS Mean Data Rate, Nominal MSDU Size Maximum Service Interval or Delay Bound

9. 9 Reference scheduling algorithm in 802.11e The schedule for an admitted stream is calculated in three steps. 1. 1. Calculation of the Scheduled Service Interval (SI). 2. 2. Calculation of TXOP duration for a given SI 3. 3. Admission control scheme 1. 1. Service Interval (SI) calculates the minimum of all Maximum Service Intervals for all admitted streams. Let this minimum be "m". chooses a number lower than "m" that is a submultiple of the beacon interval. Ex. MSI 1 =15ms, MSI 2 =20ms, beacon interval = 100 =>SI = 10ms

10. 10 Reference scheduling algorithm in 802.11e 2. Calculation of TXOP duration Mean Data Rate (ρ) Mean Data Rate (ρ) Nominal MSDU Size (L i ) from the negotiated TSPEC Nominal MSDU Size (L i ) from the negotiated TSPEC Scheduled Service Interval (SI) calculated in the first step, Scheduled Service Interval (SI) calculated in the first step, N i : the number of MSDUs that arrived at the Mean Data Rate during the SI

11. 11 Reference scheduling algorithm in 802.11e Parameters Parameters Nominal MSDU Size (L i ) from the negotiated TSPEC Nominal MSDU Size (L i ) from the negotiated TSPEC Min Physical Transmission Rate (R), Min Physical Transmission Rate (R), Maximum allowable MSDU size (M) Maximum allowable MSDU size (M) Overheads in time units (O): IFSs, ACKs, and CF-Polls Overheads in time units (O): IFSs, ACKs, and CF-Polls 3. Admission Control

12. 12 Admission Control AC for CBR traffic AC for CBR traffic AC for VBR traffic AC for VBR traffic

13. 13 Survey of Admission Control Deyun Gao, Jianfei Cai and King Ngi Ngan, “ Admission Control in IEEE 802.11e Wireless LANs, ” in IEEE Network, July/August 2005 Deyun Gao, Jianfei Cai and King Ngi Ngan, “ Admission Control in IEEE 802.11e Wireless LANs, ” in IEEE Network, July/August 2005 Admission Control for CBR Traffic Admission Control for CBR Traffic physical-rate-based admission control PRBAC physical-rate-based admission control PRBAC Admission Control for VBR Traffic Admission Control for VBR Traffic Effective TXOP duration Effective TXOP duration Variable Service Interval Variable Service Interval

14. 14 Admission Control for CBR Traffic Gao, D.; Cai, J.; Zhang, L., “ Physical rate based admission control for HCCA in IEEE 802.11e WLANs ”, Advanced Information Networking and Applications, 2005. AINA 2005 Gao, D.; Cai, J.; Zhang, L., “ Physical rate based admission control for HCCA in IEEE 802.11e WLANs ”, Advanced Information Networking and Applications, 2005. AINA 2005 physical-rate-based admission control (PRBAC) physical-rate-based admission control (PRBAC) long-term average physical rates for admission control long-term average physical rates for admission control instantaneous physical rates to distribute TXOPs instantaneous physical rates to distribute TXOPs

15. 15 Admission Control for CBR Traffic

16. 16 The maximum numbers of VoIP traffic stream Woo-Yong Choi, “ A Centralized MAC-Level Admission Control Algorithm for Traffic Stream Services in IEEE 802.11eWireless LANs ”, International Journal of Electronics and Communications, 2004 Woo-Yong Choi, “ A Centralized MAC-Level Admission Control Algorithm for Traffic Stream Services in IEEE 802.11eWireless LANs ”, International Journal of Electronics and Communications, 2004 obtain the maximum numbers of VoIP traffic streams that can be admitted to IEEE 802.11a/e, IEEE 802.11b/e and IEEE 802.11g/e wireless LANs for various delay requirements.

17. 17 Arrival Pattern Di: the constant inter-arrival of burst Li: burst size, [0, maximum burst size] Pi: the length of the burst period,

18. 18 Arrival Pattern (cond) Li: burst size, [0, maximum burst size] PRi: the peak data rate MRi: the mean data rate Pi: the length of the burst period, Di: the constant inter-arrival of burst

19. 19 Max Queue Size B: the maximum queue state Si: constant service rate Pi: the length of the burst period PRi: the peak data rate PRi-SiSi ^

20. 20 Max delay Ti : the maximum delay provide the traffic stream with the constant service rate, Si (bits/second). ==> Admission Control Admission Control ΣSi < available service rate (AR) ΣSi < available service rate (AR)

21. 21 Numerical examples Burst length Pi = 1.5 sec Burst length Pi = 1.5 sec Burst inter-arrival time Di = 1 sec Burst inter-arrival time Di = 1 sec IMBE codec:4.8Kbps IMBE codec:4.8Kbps User payload of VoIP MPDU is 88 bits User payload of VoIP MPDU is 88 bits Number of MPDU = 4.8*1.5/88 = 82 Number of MPDU = 4.8*1.5/88 = 82 Burst size Li Burst size Li = 4.8*1.5 + 82*(UDP, IP and MAC) = 7200 + 82*(16 + 224 + 240) = 46560 bits

22. 22 Numerical examples Peak data rate PRi = Li/Pi = 31Kbps Peak data rate PRi = Li/Pi = 31Kbps Mead data rate = PRi*1.5/(1.5+1) = 18.6Kbps Mead data rate = PRi*1.5/(1.5+1) = 18.6Kbps Actual available service rate R Actual available service rate R R=11.34Mbps(a, g), 2.2Mbps(b) R=11.34Mbps(a, g), 2.2Mbps(b)

23. 23 Numerical examples 5 times About 35 VoIP pairs

24. 24 Admission control for VBR traffic W.F. Fan, D.Y. Gao, D. H.K. Tsang and B. Bensaou, "Admission Control for Variable Bit Rate traffic in IEEE 802.11e WLANs, ” to be appewed in The Joint Conference of 10th Asia-Pacific Conference on Communications and 5th International Symposium on Multi-Dimensional Mobile Communications, Aug. 2004 W.F. Fan, D.Y. Gao, D. H.K. Tsang and B. Bensaou, "Admission Control for Variable Bit Rate traffic in IEEE 802.11e WLANs, ” to be appewed in The Joint Conference of 10th Asia-Pacific Conference on Communications and 5th International Symposium on Multi-Dimensional Mobile Communications, Aug. 2004 introducing Effective TXOP duration ( introducing Effective TXOP duration (the necessary TXOPs which can statistically guarantee that the packet loss probability is less than a threshold ) guarantee the packet loss rate guarantee the packet loss rate

25. 25 Admission control for VBR traffic W. F. Fan;Tsang, D.H.K.; Bensaou, B., “ Admission Control for Variable Bit Rate traffic using variable Service Interval in IEEE 802.1 le WLANs ”, ICCCN 2004. Proceedings W. F. Fan;Tsang, D.H.K.; Bensaou, B., “ Admission Control for Variable Bit Rate traffic using variable Service Interval in IEEE 802.1 le WLANs ”, ICCCN 2004. Proceedings using Variable Service Interval using Variable Service Interval avoid over-guarantee on packet delay (which with large delay bound) avoid over-guarantee on packet delay (which with large delay bound) guarantee the packet loss rate guarantee the packet loss rate The larger the service interval, the less TXOP durations required The larger the service interval, the less TXOP durations required

26. 26 Parameterized EDCA Comparison of HCCA and EDCA Comparison of HCCA and EDCA Admission Control algorithm Admission Control algorithm Resource Allocation algorithm Resource Allocation algorithm Performance Evaluation Performance Evaluation

27. 27 Parameterized EDCA Chun-Ting Chou, Sai Shankar N and Kang G. Shin, “ Achieving Per-Stream QoS with Distributed Airtime Allocation and Admission Control in IEEE 802.11e Wireless LANs, “ INFOCOM 2005 Chun-Ting Chou, Sai Shankar N and Kang G. Shin, “ Achieving Per-Stream QoS with Distributed Airtime Allocation and Admission Control in IEEE 802.11e Wireless LANs, “ INFOCOM 2005 Distributed Control of Airtime Usage in Multi-rate Wireless LANs Chun-Ting Chou, Kang G. Shin and Sai Shankar, “ Distributed Control of Airtime Usage in Multi-rate Wireless LANs, ” under review of the IEEE/ACM Transactions on Networking

28. 28 Comparison of HCCA and EDCA Challenges of HCCA Challenges of HCCA The HC needs to re-compute the service schedule whenever a new traffic stream is added to, or deleted from a WLAN The HC needs to re-compute the service schedule whenever a new traffic stream is added to, or deleted from a WLAN When two WLANs using HCCA operate on the same channel, it requires additional coordination between them When two WLANs using HCCA operate on the same channel, it requires additional coordination between them 802.11k 802.11k

29. 29 Comparison of HCCA and EDCA (cont ’ d) Challenges of EDCA Challenges of EDCA A quantitative control of stations ’ medium occupancy cannot be achieved via the current EDCA A quantitative control of stations ’ medium occupancy cannot be achieved via the current EDCA The link adaptation allows stations to vary their PHY transmission rate based on the link condition makes the airtime usage control even harder The link adaptation allows stations to vary their PHY transmission rate based on the link condition makes the airtime usage control even harder

30. 30 Admission Control Algorithm Guaranteed Rate (g) -----------------Appendix A Guaranteed Rate (g) -----------------Appendix A C is the channel capacity C is the channel capacity Dependent on PHY rates Dependent on PHY rates We must consider multi-rate 802.11 environment We must consider multi-rate 802.11 environment airtime ratio(r i,j ) airtime ratio(r i,j )

31. 31 Admission Control Algorithm New condition New condition Overall conditions Overall conditions EA: Efficient Airtime Ratio EA: Efficient Airtime Ratio

32. 32 Allocation of Airtime EDCA EDCA Control the TXOP Limit of each stations Control the TXOP Limit of each stations Same EDCA parameters Same EDCA parameters Control the frequency of station ’ s access to the wireless medium Control the frequency of station ’ s access to the wireless medium Same TXOP (access duration) Same TXOP (access duration)

33. 33 Controlling the TXOP Limit TXOP TXOP N i : The number of data frames per one access N i : The number of data frames per one access Transmission time : Transmission time : NiNi NiNi NiNi

34. 34 Controlling the TXOP Limit (cont ’ d) Example Example L i = 600, 600, 1200, 1200 bytes L i = 600, 600, 1200, 1200 bytes R i = 48, 48, 48, 24Mbps R i = 48, 48, 48, 24Mbps T i =100, 100, 200, 400(T M )  sec T i =100, 100, 200, 400(T M )  sec r i = 0.1, 0.2, 0.2, 0.1(r M ) r i = 0.1, 0.2, 0.2, 0.1(r M ) N i = 4, 8, 4, 1 N i = 4, 8, 4, 1

35. 35 Controlling the Access Frequency TXOP limit TXOP limit Access frequency approximation[24] Access frequency approximation[24] [24] Chun-Ting Chou, Kang G. Shin and Sai Shankar, “ Distributed Control of Airtime Usage in Multi-rate Wireless LANs, ” under review of the IEEE/ACM Transactions on Networking

36. 36 Controlling the Access Frequency (cont ’ d) Example Example L i = 600, 600, 1200, 1200 bytes L i = 600, 600, 1200, 1200 bytes R i = 48, 48, 48, 24Mbps R i = 48, 48, 48, 24Mbps T i =100, 100, 200, 400(T M )  sec T i =100, 100, 200, 400(T M )  sec N i = 4, 4, 2, 1 N i = 4, 4, 2, 1 r i = 0.1, 0.2, 0.2, 0.1 r i = 0.1, 0.2, 0.2, 0.1

37. 37 System Efficiency t =35 s 37 Mbps N=16

38. 38 Time-varying Transmission Rates 54Mbps → 24Mbps

39. 39 Conclusion HCCA HCCA High system efficiency (higher EA) High system efficiency (higher EA) Contention free Contention free with admission control mechanisms, the delay and packet loss rate of VoIP / other multimedia streams are guaranteed. with admission control mechanisms, the delay and packet loss rate of VoIP / other multimedia streams are guaranteed. However, the influence of Jitter is not discussed here. However, the influence of Jitter is not discussed here.

40. 40 Conclusion (cont ’ d) Parameterized EDCA Parameterized EDCA No control overhead in overlapping multiple LANs No control overhead in overlapping multiple LANs No adjustment for change No adjustment for change HCCA needs to re-compute schedule HCCA needs to re-compute schedule Easy adaptation of extra airtime for the change of station ’ s PHY rate Easy adaptation of extra airtime for the change of station ’ s PHY rate

41. 41 Reference IEEE 802.11 WG, “ Draft Supplement to Standard for Telecommunications and Information Exchange between Systems-LAN/MAN Specific Requirements — Part 12: Wireless MAC and PHY Specifications: MAC Enhancements for QoS, ” IEEE 802.11e/draft 12.0, Nov. 2004. Qiang Ni, ” Performance Analysis and Enhancements for IEEE 802.11e Wireless Networks, ” in IEEE Network, July/August 2005 Qiang Ni, ” Performance Analysis and Enhancements for IEEE 802.11e Wireless Networks, ” in IEEE Network, July/August 2005 Deyun Gao, Jianfei Cai and King Ngi Ngan, “ Admission Control in IEEE 802.11e Wireless LANs, ” in IEEE Network, July/August 2005 Deyun Gao, Jianfei Cai and King Ngi Ngan, “ Admission Control in IEEE 802.11e Wireless LANs, ” in IEEE Network, July/August 2005 Gao, D.; Cai, J.; Zhang, L., “ Physical rate based admission control for HCCA in IEEE 802.11e WLANs ”, Advanced Information Networking and Applications, 2005. AINA 2005 Gao, D.; Cai, J.; Zhang, L., “ Physical rate based admission control for HCCA in IEEE 802.11e WLANs ”, Advanced Information Networking and Applications, 2005. AINA 2005

42. 42 Reference W.F. Fan, D.Y. Gao, D. H.K. Tsang and B. Bensaou, "Admission Control for Variable Bit Rate traffic in IEEE 802.11e WLANs, ” to be appewed in The Joint Conference of 10th Asia-Pacific Conference on Communications and 5th International Symposium on Multi-Dimensional Mobile Communications, Aug. 2004 W.F. Fan, D.Y. Gao, D. H.K. Tsang and B. Bensaou, "Admission Control for Variable Bit Rate traffic in IEEE 802.11e WLANs, ” to be appewed in The Joint Conference of 10th Asia-Pacific Conference on Communications and 5th International Symposium on Multi-Dimensional Mobile Communications, Aug. 2004 W. F. Fan;Tsang, D.H.K.; Bensaou, B., “ Admission Control for Variable Bit Rate traffic using variable Service Interval in IEEE 802.1 le WLANs ”, ICCCN 2004. Proceedings W. F. Fan;Tsang, D.H.K.; Bensaou, B., “ Admission Control for Variable Bit Rate traffic using variable Service Interval in IEEE 802.1 le WLANs ”, ICCCN 2004. Proceedings Woo-Yong Choi, “ A Centralized MAC-Level Admission Control Algorithm for Traffic Stream Services in IEEE 802.11eWireless LANs ”, International Journal of Electronics and Communications, 2004 Woo-Yong Choi, “ A Centralized MAC-Level Admission Control Algorithm for Traffic Stream Services in IEEE 802.11eWireless LANs ”, International Journal of Electronics and Communications, 2004 Chun-Ting Chou, Sai Shankar N and Kang G. Shin, “ Achieving Per-Stream QoS with Distributed Airtime Allocation and Admission Control in IEEE 802.11e Wireless LANs, “ INFOCOM 2005 Chun-Ting Chou, Sai Shankar N and Kang G. Shin, “ Achieving Per-Stream QoS with Distributed Airtime Allocation and Admission Control in IEEE 802.11e Wireless LANs, “ INFOCOM 2005 Distributed Control of Airtime Usage in Multi-rate Wireless LANs Chun-Ting Chou, Kang G. Shin and Sai Shankar, “ Distributed Control of Airtime Usage in Multi-rate Wireless LANs, ” under review of the IEEE/ACM Transactions on Networking

43. 43 Appendix A. Guaranteed Rate Dual-token bucket filter Dual-token bucket filter Tokens arrive at Peak Data Rate Tokens arrive at Mean Data Rate MAC frame buffer Bucket size B = σ (1-ρ/P) σ Bits Time d mean rate : ρ peak rate : P Guaranteed rate : g Arrival curve : A(t) Arriving traffic stream Data frames drained at Guaranteed Rate error