Chapter 4 Simulations and Numerical Results
4.1 Simulation model
4.1 Simulation Model
In the simulation, we use the NS-2 simulator (version 2.29) [26] with 802.11e supported [27] as platform. The simulation environment is composed of a QAP and variable number of QSTAs (see Figure 4.1).
Figure 4.1: The simulation topology.
The physical and MAC layer parameters are shown in the Table 4.1. The length of a beacon interval is set to 1 second. We use three kind of codec for VoIP: 72 bytes payload with 20ms packet interval, 72 bytes payload with 30ms packet interval, and 72 bytes payload with 50ms packet interval. We use three types of frequency to simulate Video traffic: 10 frame per second (fps), 15 fps, and 30fps. The payload length of each video stream is set to 1000 bytes.
Each station generates variable-bit-rate (VBR) traffic according to the two-state talk-silence speech model [18], so we set the talk-spurts period to 7.24 sec and set the silence period with 5.69 second. All video are set as VBR. For best effort test, we use FTP transmissions with 1000 bytes payload length.
In our simulation, we compared our scheme with round robin polling scheme in the 802.11e reference scheduler.
4.2 Numerical Results
We compared our proposed scheme with round-robin polling based scheduler in access delay, total throughput, and delay jitter for various numbers of stations. The simulation results are presented in the following sections.
4.2.1 Throughput
From Figure 4.2, we can see the decrement of average total throughput as the number of QSTAs increases. The figure shows that the Highly Efficient Polling scheme performs much better than the round-robin polling scheme in total throughput. HEP scheme starts at 19610 K-Bytes and RRP scheme from 18559 K-Bytes when the numbers of CBR and VBR streams are both three, and there are 16 FTP streams. When the number of QoS stations increases, the gap is getting larger between these two polling schemes.
Figure 4.2: The relationship of total throughput and network load.
Table 4.2 shows the average throughput of FTP, VBR and CBR traffic streams in using round robin polling scheme and our proposed polling scheme. In this table, we can find that the throughput in VBR and CBR traffic streams are the same. Hence, we can say that both our proposed polling scheme and RR polling scheme guarantee the transmissions for real-time streams. In Table 4.2, we can also notice that under the same number of traffic streams, our proposed polling scheme has better performance than RR polling scheme in total throughput.
From Figure 4.1 with Table 4.2, as the number of traffic streams increases, our proposed polling scheme has more chance to accomplish FTP (best effort) data transmission than RR polling scheme.
Table 4.2: Average throughput between two polling schemes Number of
TSs
Average Throughput (K‐Bytes/sec)
HEP_FTP RR_FTP HEP_VBR RR_VBR HEP_CBR RR_CBR
6 1039.36 980.95 2.00 2.00 48.56 48.56
12 921.38 806.92 5.58 5.58 96.78 96.83
18 807.16 639.25 8.14 8.14 145.22 145.28
24 685.36 462.57 12.82 12.81 192.67 192.67
30 576.68 307.44 15.56 15.56 240.22 240.22
36 469.04 165.53 18.40 18.40 288.56 288.67
Figure 4.3 shows the throughput of best effort traffic against network sizes. It shows a significantly increased gap between HEP with and without silence detection function in one side and RR polling scheme on the other side. The higher throughput in both two HEP schemes is primarily due to the reduction of polling overheads. With silence detection function, the more silence QSTA been detected, the more polling overhead will be reduced.
Figure 4.3: Throughput against network size of RR, and two types of HEP schedulers.
4.2.2 Access Delay
Figure 4.3 shows the average access delay against different network size for CBR /VBR traffic streams and HE/RR polling schemes. By this figure, the HEP scheme features lower access delay comparing with RR polling scheme due to the reduction of polling overhead. In RR polling scheme, the access delay would be increased with the increase of network size;
however, in HEP scheme, and with the increase of network size, the access delay would be decreased. This is because that the more QSTAs need to poll, the CAP duration will be larger, and QSTAs will have more chance to be polled in current CAP.
Figure 4.4: Average access delay of time sensitive traffic against number of QSTA
4.2.3 Delay jitter
In our analysis, the jitter is calculated by using the following equation [29]:
(
receive time j_ ( ) send_time j( )) (
receive time i_ ( ) send_time i( ))
jitter
j i
− − −
= −
, where j > i (5)
Figure 4.5 and Figure 4.6 show the delay jitter of one real-time traffic stream in RR polling scheme and our proposed scheme, respectively. They are measured in the network condition with 18 CBR, 18 VBR, and 46 FTP traffic streams. In these two figures, we can see the delay jitter in RR polling scheme (standard deviation S is 10.30) is larger than that in our proposed scheme (standard deviation S is 6.92). In some condition, RRP scheme is even higher than 30 ms (see Figure 4.5).
Figure 4.5: Delay jitter between packets in RR polling scheme.
Figure 4.6: Delay jitter between packets in our proposed scheme.
Chapter 5
Conclusion and Future Work
5.1 Conclusion
In the thesis, we proposed a new polling scheme, called Highly Efficient Polling (HEP) for improving the performance of real-time multimedia traffic over WLAN.
There are several features in our proposed scheme: First, this polling scheme is based on the interval in the estimated polling time of each traffic stream, for the reason, our strategy is to poll the station with earliest-due-time-first instead of round-robin polling of reference scheduler. It can reduce the waste of bandwidth when the polled stations have no data frame to send. Second, in order to precisely detect the actual en-queue time for traffic streams, our scheme uses the QoS control field defined in IEEE 802.11e standard to carry the data arrival information. In this mechanism, QSTAs should encapsulate the actual queueing time to the QoS control field of data frame, HC could use these collected information to adjust the estimated polling time in its Polling List. Third, the proposed scheme could dynamically adjust the service interval of voice transmission according to the characteristics of voice flow.
This silence detection function detects the silence streams by counting the QoS-Null frames.
When it determines that a traffic stream is in silence state, it will switch to a longer polling interval as estimated polling time for reducing the overhead of polls.
One of the main features in our scheme is the low complexity of implementation. Due to limited number of client stations, we use the liner search to find the next target to be polled, so that the implementation complexity is greatly reduced.
We use NS-2 (ns-allinone-2.29) with ns2 IEEE 802.11 support which contains 802.11e EDCA and HCCA modules to evaluate our highly efficiently polling scheme and round-robin
polling scheme. HEP has shown much better performance than the results using round-robin polling scheme, in term of throughput, access delay and jitter variation.
5.2 Future Work
In the future, we will try to do more comparisons with other polling schemes, as well as do some tests for verifying the decisions of our silence detection mechanism to reduce the polling overhead due to null responses.
References:
[1] IEEE Std 802.11e-2005.
[2] IEEE Std 802.11, “Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications”, Aug. 1999.
[3] J. Zheng and E. Regentova, “An Improved Polling Scheme for Voice Support in IEEE 802.11 Wireless Network”, Proceedings of the International
Conference on Information Technology (ITCC’05), Vol. 2, pp.603-608.
[4] Ray Y. W. Lam and Victor C. M. Leung, “Polling-Based Protocols for Packet Voice Transport Over IEEE 802.11 Wireless Local Area Networks”, IEEE Wireless Communications, Vol. 13, pp.22-29, Feb. 2006.
[5] P. Wavg, H. Jiang and W. Zhuang, “IEEE 802.11E Enhancement for Voice Service”, IEEE Wireless Communications, Vol. 13, pp.30-35, Feb. 2006.
[6] D. Chen, D. Gu and J. Zbang, “Supporting real-time traffic with QoS in IEEE 802.11e based home networks”, IEEE CNF, pp.205-209, 2004.
[7] J. Son, H. Choi and S. C. Park, “A Efficient Polling MAC Scheme for IEEE 802.11e”, international Symposium on Communications and Information Technologies 2004 (ISCIT 2004), Sapporo. Japan, Oct. 26-29, 2004.
[8] P. Ansel, Q. Ni and T. Turletti, “An Efficient Scheduling Scheme for IEEE 802.11e”, Planete Project, INRIA Sophia Antipolis, France.
[9] W. Quan and D. M. Hui, “Improving the performance of WLAN to support VoIP application”, Mobile Technology, Applications and Systems, 2005 2nd International Conference, pp.5, 2005 IEEE CNF.
[10] L. W. Lim, R. Malik and P. Y. Tan, “A QoS Scheduler for IEEE 802.11e WLANs”, Panasonic Singapore Laboratories, pp.199-204, 2004 IEEE.
[11] M. Matsumoto and T. Itoh, “QoS-guarantee Method for Public
Wireless LAN Access Environments”, Nippon Telegraph and Telephone Corporation (NTT) Network Service Systems Laboratories, 2005.
[12] G. Bianchi, I. Tinnirello and L. Scalia, “Understanding 802.11e
contention-based prioritization mechanisms and their coexistence with legacy 802.11 stations”, Vol. 19, pp.28-34, 2005 IEEE JNL - ieeexplore.ieee.org.
[13] Q. Ni, “Performance analysis and enhancements for IEEE 802.11e wireless networks”, Network, Vol. 19, pp.21-27, IEEE, 2005 - ieeexplore.ieee.org.
[14] X. Ma, C. Du and Z. Niu, “Adaptive Polling List Arrangement Scheme for Voice Transmission with PCF in Wireless LANs”, 10th Asia-Pacific Conference on Communication and 5th International Symposium on Multi-Directional Mobile Communication.
[15] A. Grilo, M. Macedo, M. Nunes, ” A scheduling algorithm for QoS support in IEEE802.11 networks”, Vol. 10, pp.36-43, 2003 IEEE JNL.
[16] H. Lee and J. Kim, “An optimal CF-Poll piggyback scheme in IEEE 802.11e HCCA”, The 8th International Conference Advanced Communication
Technology, Vol. 3, pp.6, Feb. 2006.
[17] L. Yang, “Enhanced HCCA for real-time traffic with QoS in IEEE 802.11e based networks”, Intelligent Environments, 2005. The IEE International Workshop.
[18] S. Deng, “Traffic characteristics of Packet Voice”, IEEE International Conference on Communication, Vol. 3, pp. 1369-1347, 1995.
[19] S. Tsao, “Research challenges and perspectives of voice over wireless LAN”, IEEE CNF, pp. 4, 2005 - ieeexplore.ieee.org.
[20] P. T. Brady, “A model for generating OF-OFF speech patterns in two-way conversations”, Bell System Technical Journal, Vol. 48, pp. 2445-2472, Sept.
1969.
[21] J. Naoum-Sqwaya, B. Ghaddar, S. Khawam, H. Safa, H. Artail and Z. Dawy,
“Adaptive Approach for QoS Support in IEEE 802.11e Wireless LAN”, IEEE WMob’05, Vol. 2, pp/167-173, Aug. 2005.
[22] L. Romdhani, Q. Ni and T. Turletti, “Adaptive EDCF: Enhanced Service Differentiation for IEEE 802.11 Wireless Ad Hoc Networks”, Pro. IEEE WCNC, New Orleans, LA, Mar. 2003.
[23] M. Barry, A.T. Campbell and A. Veres. “Distributed Control Algorithms for Service Differentiation in Wireless Packet Networks.” In IEEE Computer and Communications Societies INFOCOM 2001, pages 582-590.
[24] I. Aad and C. Castelluccia. “Differentiation Mechanisms for IEEE 802.11”, IEEE INFOCOM 2001, pages 209-218.
[25] G. W. Wong and R. W. Donaldson, “Improving the QoS Performance of EDCF in IEEE 802.11e Wireless LANs”, IEEE Pacific Rim Conference on
Communications, Computers and Signal Processing PACRIM 2003, pages 392-296.
[26] The network simulator – ns-2. Available: http://www.isi.edu/nsnam/ns/.
[27] The 802.11(with 11e) module. Available: http://yans.inria.fr/ns-2-80211/.
[28] http://en.wikipedia.org/wiki/802.11.
[29] http://140.116.72.80/~smallko/ns2/tool.htm.
Vita
Chao-Tsung Sung ([email protected]) received his M.S. degree in Industrial Engineering, from the National Tsing-Hua University, Taiwan (2005~2007) and
B.S. degree in Industrial Engineering, from the Feng Chia University, Taiwan (1996~1999).
His research interests include wireless network, QoS and multimedia communication.