An event-driven simulator is developed to evaluate the performance of the proposed CAC and MAC enhancements. Unless otherwise stated, the following assumptions are made in our simulation. (1)We set TXOP limit to be zero for four ACs, which means that a QSTA can only transmit one packet each successful contention. (2)We assume an error-free IEEE 802.11b/e environment in which there are one QAP and several QSTAs. No RTS/CTS is used. (3)The beacon interval(BI) is set to 500ms.
(4)Each QSTA, except the one to generating background traffic, has a VoIP ses-sion(AC VO) using G.726 as the codec with PI equal to 20ms. (5)We assume there are two static QSTAs generating background traffic(AC BK) by poisson process of rate 10 kbps. (6)We set the initial physical rate of all stations equal to 11 Mbps.
(7)AC VO has the CWmin equals to 7, CWmax equals to 15, and IFSN equals to 2; AC BK has the CWmin equals to 31, CWmax equals to 1023, and IFSN equals to 7.
A. Influence of Admission Control
In this scenario, we want to show the importance of admission control. The admission control here includes not only the admission control of call entrance but also the regulation for entered real-time traffic. The regulation determines whether a traffic stream can contend the medium or not; a TS can contest the medium for sending packet only when TxUsedDn[AC VO] (or TxUsedUp[AC VO]) is smaller than TxAdDn[AC VO] (or TxAdUp[AC VO]).
In addition to one QAP and two QSTAs generating background traffic in the
Number of VoIP Calls
Figure 5.1: Scenario A: The number of calls as time goes on.
original network, we will add one QSTA to the network every two seconds if possible (call arrival rate is 0.5 call/sec). We assume the queue size limitation is 50 to each AC and all the network resource is giving to the AC VO. The following we will compare the network performance when either the admission control mechanism is used or not.
Fig. 5.1 shows us that the network can at most increase the VoIP calls to 17 when the CAC mechanism is used (because of the call entrance control). Although the calls number can unboundedly increase when there is no admission control, the collision rate will rapidly increase when the call number is more than 17 after 34th second(Fig. 5.4). Because of the high collision rate, we can observe that the net-work goodput without the CAC will less than the one with CAC after 34th second in Fig. 5.2. Before the 34th second, the goodput with the CAC mechanism is better than the no-CAC one because of the regulation and control message overhead. Nev-ertheless, this CAC regulation is helpful to control the collision probability(Fig. 5.4) and resource sharing. For example, Fig. 5.3 shows us that the throughput of back-ground traffic is better with the CAC mechanism because the resource sharing is more fair.
With these simulation results, we conclude that the admission control is necessary especially when supporting VoIP service.
B. Influence of Host Mobility
Mobility is a nature character in WLANs. This character will affect the
trans-Total Goodput of VoIP Sessions
Figure 5.2: Scenario A: Total goodput of the WLANs either the CAC mechanism is applied.
Figure 5.3: Scenario A: Throughput of background traffic in the WLANs either the CAC mechanism is applied.
Average Collisions per-STA
0 50 100 150 200 250
1 11 21 31 41 51 61 71 81 91 Time(Sec)
No.ofCollisionsper-Sec
AC noAC
Figure 5.4: Scenario A: Average collisions number per-QSTA per-second either the CAC mechanism is applied.
1Mbits/s 2Mbits/s 5.5Mbiits/s 11Mbits/s
1/4 1/4 1/4
1/4 1/4 1/4
1/2 1/2
3/4 3/4
Figure 5.5: movement.
mission rate of stations, and will even cause the handoff problem.
In this scenario, we assume all QSTAs move within a QAP. Thus, we use the following pattern to simulate the movement: every second, each non-AP QSTA has 50% probability to change it’s physical rate. If the decision of changing rate is taken, to increase or decrease the transmission rate has the same probability.
Fig. 5.5 presents the transition diagram of this imitate movement.
What we want to learn in this experiment are the influence of mobility and if our mechanism proposed in 3.2 will increase the performance or not. Therefore, we eval-uate the average goodput (Fig. 5.6) and collisions (Fig. 5.7) per-QSTA per-second in this environment. In the Fig. 5.6, we can learn that the mobility function will decrease the network throughput(AC M) and using our third enhancement mecha-nism proposed in 3.2 will alleviate the goodput decreasing (AC ME). Fig. 5.7 shows that collisions number is proportion to goodput and QSTAs number. The reason is that the collision probability is inverse proportion to the transmission opportunity.
The goodput degradation is large in our simulation when mobility function is added. This is because we use a high mobile condition. Therefore, the degradation won’t so serious if the network is more stable. In conclusion, the mobile condition will affect the network goodput, and our mechanism will moderate this degradation.
C.Influence of Codec and Packetization Interval
In this part, we want to discuss about the influence of codec and packetization interval. Although there are many works probe into this issue ([7, 10]), we made a small experiment.
Fig. 5.8 shows the theoretical value of VoIP call numbers can be supported within an QAP computed using the MT computation equation mentioned in 3.1. In addition, Fig. 5.9 shows the simulated value of call numbers can be supported within an QAP when AC VO has queue size of 50 frames and dropping rate is limited by 0.02. With these information, we can figure out that the third mechanism in 3.2 may work well because that both the codec and the packetization interval may influence the number of phone calls within a QAP.
D. Goodput Evaluation of Redirecting Packets to other queues
In this experiment, we verify that how much improvement can be made by the second enhancement mentioned in chapter 4.
From Fig. 5.10 and Fig. 5.11, we learn that the network performance will im-proved with the mechanism that redirect VoIP packets to other queue. Compare
0
Figure 5.6: Scenario B: Average goodput per-QSTA per-second when WLAN has 5, 10, 15, 17 VoIP calls. AC in the figure presents the goodput under the condition which all QSTAs is static; AC M means the non-AP QSTA will move in the way we defined; AC ME shows the goodput improvement with our enhancement under the mobile environment .
Figure 5.7: Scenario B: Average collisions when WLAN has 5, 10, 15, 17 VoIP calls.
The figure shows the condition of collision within 100 seconds within 100 seconds.
Figure 5.8: The calculating number of VoIP calls an QAP can support with different conditions.
Figure 5.9: The number of VoIP calls an QAP can support when AC VO queues size is 50 frames and dropping rate is smaller than 0.02.
Average Goodput Comparison
Figure 5.10: Scenario D: In this figure, AC R means there is a mechanism of redi-recting packet to AC BE queue except original admission control. AC M R is the condition that using the redirect mechanism when mobility function is on.
this experiment(Fig. 5.10) and scenario B(Fig. 5.6) that using different method to improve the performance degradation, we find the gain of this MAC enhancement is much better. The reason may be that there is no AC BE traffic in our simulated environment. Cooperate these two mechanism can greatly alleviate the problem cased by the mobility nature in WLANs.
E. Enhancement to Downlink Traffic
We verify the final MAC enhancement in chapter 4 in this experiment. Therefore, when QAP receive the uplink traffic of VoIP, it will send back the downlink traffic of the same call if possible. Fig. 5.12 and Fig. 5.13 figure out the simulation results of goodput and collision condition. In substance, this mechanism will improve the goodput of VoIP traffics because it may decreasing the possible contention number.
In addition, the downlink traffic can be transmitted more smooth.
Total Goodput of 17 VoIP sessions
Figure 5.11: Scenario D: This figure shows the goodput variation within 100 seconds when there are 17 VoIP calls.
0
Figure 5.12: Scenario E: The average goodput enhancement per-QSTA per-second when we enhance the QAP to profit the downlink traffic.
Average collisions per-STA
0 20 40 60 80
1 2 3 4
No. of VoIP calls
No.ofcollisionsper-Sec
AC AC_EQAP
Figure 5.13: Scenario E: The average collision numbers per-QSTA per-second either of enhancing the QAP to profit the downlink traffic or not.
Chapter 6 Conclusions
In this paper we describe the novel protocol to enhance the performance of VoIP services by integrate the SIP call setup signaling and 802.11e QoS mechanism. IEEE 802.11e is in the final stage to become a standard, so we choice to make good use of it to solve the QoS problem in WLANs with minimum modification. We think the cross-layer protocol design to facilitate the process of applications will be a trend in future. In addition, we present some MAC enhancement to facilitate the VoIP service under WLANs. Moreover, the simulation results show that our adjustment during transmission and MAC enhancement will work well.
The future work may focus on several aspects: (1)solve the handoff problem;
(2)experiment in real environment; (3)extend the architecture to other call signaling system.
Bibliography
[1] http://www.cisco.com/en/US/products/hw/phones/ps379/ps5056/index.html.
Technical report.
[2] http://www.motorola.com/motoinfo/product/details/0,,48,00.html. Technical report.
[3] IEEE 802.11e/D8.0, Draft Amendment to Part 11: Wireless Medium Access Control(MAC) and Physical Layer(PHY) specifications: Medium Access Con-trol(MAC) Quality of Service(QoS) Enhancements. June 2003.
[4] F. Anjum, M. Elaoud, D. Famolari, A. Ghosh, R. Vaidyannthan, A. Dutta, and P. Ageawal. Voice Performance in WLAN Networks - An Experimental Study.
In IEEE Globecom’03, 2003.
[5] D. Conllins. Carrier Grade Voice over IP Second Edition. McGraw-Hill Com-panies, Inc, 2003.
[6] S. Garg and M. Kappes. Admission Control for VoIP Traffic in IEEE 802.11 Networks. In IEEE Globecom’03, 2003.
[7] S. Garg and M. Kappes. Can I Add a VoIP Call ? In IEEE ICC’03, 2003.
[8] A. Grilo, M. Macedo, and M. Nunes. A Scheduling Algorithm for QoS Support in IEEE 802.11e Networks. IEEE Wireless Communications, June 2004.
[9] T. Hiraguri, T. Ichikawa, M. Iizuka, and M. Morikura. Novel Multiple Access Protocol for Voice over IP in Wireless LAN. In IEEE ISCC’02, 2002.
[10] D. P. Hole and F. A. Tobagi. Capacity of an IEEE 802.11b Wireless LAN supporting VoIP. In in proc. of IEEE ICC’04, 2004.
[11] F.-W. Jung, R. Muddumbai, D. Montgomery, and H.-K. Kahng. Performance Evaluation of Two Layered Mobility Management using Mobile IP and Session Initiation Protocol. In IEEE Globecom’03, 2003.
[12] L. Lim, R. Malik, P. Tan, C. Apichaichalermwongse, K. Ando, and Y. Harada.
A QoS scheduler for IEEE 802.11e WLANs. In IEEE CCNC’04, 2004.
[13] M. Malli, Q. Ni, T. Turletti, and C. Brakat. Adaptive Fair Channel Allocation for QoS Enhancement in IEEE 802.11 Wireless LANs. In IEEE ICC’04, 2004.
[14] T. D. T. Ming He, D. Zhao, and V. Kezys. Ad Hoc Assisted Handoff for Real-time Voice in IEEE 802.11 Infrastructure WLANs. In IEEE WCNC’04, 2004.
[15] J. Rosenberg, H. Schulzrinne, E. Schooler, M. Handley, G. Camarillo, A. John-ston, J. peterson, and R. Sparks. SIP: Session Initiation Protocol, RFC3261 in IETF. June 2002.
[16] Y. Xiao, H. Li, and sunghyun Choi. Protection and Guarantee for Voice and Video Traffic in IEEE802.11e Wireless Lans. In IEEE Infocom’04, 2004.
[17] J. Yu, S. Choi, and J. LEE. Enhancement of VoIP over IEEE 802.11 WLAN via Dual Queue Strategy. In IEEE ICC’04, 2004.