Now we present and discuss our simulation results. As mentioned before, we evaluate our QCS protocol when it is adopted by IEEE 802.11 TSF and ASP as an enhancement.
First of all, we plot the maximum clock drift between any two neighboring nodes during the whole simulation time when IEEE 802.11 TSF is adopted to fulfill clock syn-chronization in Fig. 4.1. Besides, IEEE 802.11 TSF adopting QCS, where N is set to 32, denoted by QCS(32), is also shown in this figure. As the figure shows, the maximum clock drift suddenly increases over 1240µs around 80 seconds after the simulation be-gins. In this case, the first asynchronous pair appears because a node with a slower TSF timer suddenly moves inside another node’s transmission range. Since the original IEEE 802.11 TSF does not address the issue of clock asynchronism caused by mobility, the maximum clock drift continuously increases. At last the maximum clock drift increases over 70000µs.
After IEEE 802.11 TSF adopts QCS, the maximum clock drift decreases under 20000µs.
Note that N = 32 in QCS implies that nodes only have 63 quorum intervals in every 1024 beacon intervals. In other words, the average of quorum intervals among all nodes is only about 30.76s throughout the entire simulation period (500s). This implies that the clock asynchronism problem can be easily relieved even though N of QCS is large.
Fig. 4.2 shows the beacon sending times throughout the entire simulation period.
From this figure, we can see that when the IEEE 802.11 TSF adopts QCS with a smaller N, the maximum or average beacon sending times increases only by a small amount.
This means that QCS does not cause too much additional beacon transmissions. Note that N = 0 in this figure means that QCS is not adopted by IEEE 802.11 TSF.
Now we discuss how the value of N in QCS can affect the average remaining time of asynchronous pairs of nodes, as shown in Fig. 4.3. Note that when IEEE 802.11 TSF does not adopts QCS as an enhancement, the average remaining time of asynchronous pairs is quite large since IEEE 802.11 TSF does not know how to handle the clock asynchronism
20000
15000
10000
5000
1240
0 1000 2000 3000 4000 5000
Maximum Clock Drift (us)
Time (BIs)
IEEE 802.11 TSF IEEE 802.11 TSF with QCS(32)
Figure 4.1: Maximum clock drift between any two neighboring nodes for IEEE 802.11 TSF and IEEE 802.11 TSF with QCS where N is set to 32.
problem. In addition, it is not hard to see that when the value of N in QCS is larger, the average remaining time of asynchronous pairs becomes longer because a quorum interval appears less frequently. However, it consumes more power resources when N is smaller.
For example, N = 2 implies that there are 3 quorum intervals in every 4 beacon intervals, i.e., nodes stay awake in about 75% of the entire simulation period.
Although ASP has automatic self-time-correcting functions, it is not guaranteed that ASP can completely prevent the clock asynchronism problem. Fig. 4.4 shows the max-imum clock drift between any two neighboring nodes during the whole simulation time when ASP is adopted for clock synchronization. As shown in Fig. 4.4, the maximum clock drift suddenly increases over 1240µs around 150s. The reason of the occurrence of the first asynchronous pair is similar to the one in IEEE 802.11 TSF - mobility. At last, the maximum clock drift increases to 34000µs around. Similarly, after ASP adopts QCS with N = 32, the maximum clock drift decreases under 15000µs.
Fig. 4.5 shows the beacon sending times throughout the entire simulation period.
Compared with IEEE 802.11 TSF in Fig. 4.2, the maximum/average beacon sending times in ASP is fewer than that in TSF. Besides, from this figure, we can see that QCS causes some but not too many additional beacon transmissions. At last, Fig. 4.6 illustrates how
0
Figure 4.2: Beacon sending times for IEEE 802.11 TSF.
580.8
Figure 4.3: Average remaining time of asynchronous pairs for IEEE 802.11 TSF.
35000
Figure 4.4: Maximum clock drift between any two neighboring nodes for ASP and ASP with QCS where N is set to 32.
Figure 4.5: Beacon sending times for ASP.
587.9
40 35 30 25 20 15 10 5
32 20 10 5 4 3 2 0
Average Remaining Time (BIs)
N of QCS
Figure 4.6: Average remaining time of asynchronous pairs for ASP.
the value of N in QCS can affect the average remaining time of asynchronous pairs of nodes. As only ASP is adopted, the average remaining time of asynchronous pairs is up to 59s. However, the situation becomes totally different when ASP adopts QCS. From this figure, we can see that the average remaining time of asynchronous pairs is smaller than 4s when N ≤ 32 in QCS.
Chapter 5 Conclusions
In this paper, we point out that the current clock synchronization protocols lack an ex-ceptional handling mechanism to solve the clock asynchronism problem in IEEE 802.11 MANETs. Therefore, we propose a compatible protocol called QCS to address the clock asynchronism problem in IEEE 802.11 multi-hop MANETs. Through our simulations, we show that our proposed scheme can assist the existing clock synchronization proto-cols in solving the clock asynchronism problem successfully.
Bibliography
[1] Lifei Huang and Ten-Hwang Lai. On the Scalability of IEEE 802.11 Ad Hoc Net-works. In ACM Int’l Symp. on Mobile Ad Hoc Networking and Computing (Mobi-HOC), pages 173–182, 2002.
[2] Dong Zhou and Ten-Hwang Lai. Analysis and Implementation of Salable Clock Synchronization Protocols in IEEE 802.11 Ad Hoc Networks. In 2004 IEEE In-ternational Conference on Mobile Ad-hoc and Sensor Systems (MASS’05), pages 255–263, 2004.
[3] Dong Zhou and Ten-Hwang Lai. A Compatible and Scalable Clock Synchronization Protocol in IEEE 802.11 AD Hoc Networks. In Proceedings of the 2005 Interna-tional Conference on Parallel Processing (ICPP’05), pages 295–302, 2005.
[4] Jang-Ping Sheu, Chih-Min Chao, and Ching-Wen Sun. A Clock Synchronization Algorithm for Multi-hop Wireless Ad Hoc Networks. In Proceedings of the 24th International Conference on Distributed Computing Systems, pages 574–581, 2004.
[5] Dong Zhou and Ten-Hwang Lai. A Scalable and Adaptive Clock Synchronization Protocol for IEEE 802.11-Based Multihop Ad Hoc Networks. In 2005 IEEE mobile ad hoc and sensor system conference, 2005.
[6] Jungmin So and Nitin Vaidya. MTSF: A Timing Synchronization Protocol to Sup-port Synchronous Operations in Multihop Wireless Networks. In Technical ReSup-port, 2004.
[7] Divyakant Agrawal and Amr El Abbadi. An Efficient and Fault-Tolerant Solution for Distributed Mutual Edxclusion. In ACM Transactions on Computer Systems, pages 1–20, 1991.
[8] Hector Garcia-Molina and Daneil Barbara. How to Assign Votes in a Distributed System. In Journal of ACM, pages 1029–1040, 1985.
[9] Jehn-Ruey Jiang, Yu-Chee Tseng, Chih-Shun Hsu, and Ten-Hwang Lai. Quorum-Based Asynchronous Power-Saving Protocols for IEEE 802.11 Ad Hoc Networks.
In ACM Mobile Networking and Applications (MONET), pages 169–181, 2005.
[10] Mamoru Maekawa. A√
N Algorithm for Mutual Exclusion in Decentralized Sys-tems. In ACM Trans. Comput. Syst., pages 145–159, 1985.
[11] S. D. Lang and L. J. Mao. A Torus Quorum Protocol for Distributed Mutual Exclu-sion. In Proc. of the 10th Conf. on Parallel and Distributed Computing and Systems, pages 635–638, 1998.
[12] Wai-Shing Luk and Tien-Tsin Wong. Two new quorum based algorithms for distrib-uted mutual exclusion. In Proc. of International Conference on Distribdistrib-uted Com-puting Systems, pages 100–106, 1997.