Simulation Environment and Parameters

4.1 Simulation Environment and Parameters

In this chapter, we show the simulation results of our VDEB scheme and analysis its performance. We conduct the simulation using the ns-2 simulator [25]. As mentioned in Chapter 3, the simulation scenario is in an 8 km highway with 3 lanes considering single direction. The risk zone of emergency event is 1 km. To evaluate our VDEB scheme, we also simulate the MCDS scheme and DBS scheme for comparison.

To show the tradeoff between overhead and efficiency of MCDS, we vary the hello interval from 0.2s to 3.2s. The VDEB is also simulated with the hello interval from 3.2s to 25.6s (consider that the required hello interval for VDEB is longer than that for MCDS). The impact of mobility is evaluated by two speed scenarios: low speed scenario has a speed in [20km/h, 70km/h]; high speed scenario has a speed in [70km/h, 120km/h]. Table 2 shows the details of simulation parameters.

There are five metrics we observe to compare our VDEB scheme with other schemes:

(1) Average delay: The average of delay time of all the receivers of all the emergency messages.

(2) Farthest end-to-end delay: The average of delay time that the emergency message reaches the farthest receiver for the emergency message.

(3) Average retransmissions: The average number of transmissions for the emergency message.

(4) Delay percentiles: The cumulative distribution function of delay of all the receivers during the message forwarding for the emergency message.

(5) Overhead: The total number of broadcasted hello messages throughout the simulation.

Table 2. Simulation parameters Simulation Scenario Highway Simulation Area 8 km * 3 lanes

Risk Zone 1 km

Transmission Range 250 m

MCDS Hello Interval [0.2, 0.4, 0.8, 1.6, 3.2] s VDEB Hello Interval [3.2, 6.4, 12.8, 25.6] s

Simulation Time 150 s

Vehicle Density [20, 100] vehicles/km Vehicle Speed [20, 70], [70, 120] km/h Contention Window CWmin (32 SIFS)

Slot Time 64 SIFS

4.2 Results Analysis

Before comparing our scheme with the MCDS and DBS schemes, we first observe the performance of MCDS with different hello intervals to conclude its tradeoff.

  30

(a) High speed  (b) Low speed 

Figure 16. Average delay of MCDS

0

(a) High speed  (b) Low speed 

Figure 17. Farthest end-to-end delay of MCDS

Figure 16 shows the average delay of the MCDS scheme with different hello intervals. Clearly, the longer hello interval we set, the lower average delay we have. In the MCDS scheme, the success of virtual backbone is based on the successful forwarding of the designated forwarder. If the selected forwarder is not the best one, the delay might get longer because the hop count of the message reaching the end of the risk zone will arise. But in the worst case, the designated forwarder might, in fact, not in the transmission range of the sender. This results in the retransmission of the sender. The additional time spent for this failure includes the sender’s ACK timeout and the sender’s and the forwarders’ contention time. This might cause even longer delay than simply choosing a non-best forwarder. The impact of different hello intervals on farthest end-to-end delay is also shown in Figure 17. The mobility of vehicles does not influence the performance of delay as shown in Figures 16 and 17.

To achieve better performance, the MCDS scheme selects a shorter hello interval.

As the vehicle density getting higher, the amounts of hello messages getting larger.

Large amounts of hello exhaust the network. Large numbers of broadcastings also influence the general services. This makes the backbone maintenance too costly. The schemes which construct the virtual backbone structures are not reality considering that the network still provides services for other types of applications not only emergency applications.

Our VDEB scheme also adopts the hello message, so we observe the influence of the hello interval on the performance. Since the forwarder in VDEB is not designated, the accuracy of position of neighbor vehicles is not so strict. Vehicle density is measured with long interval hello messages. We use the vehicle velocity and the position provided in the hello message to estimate the current location of the vehicle.

As shown in Figure 18, the interval of hello message has little impact on the average delay. The reason is that the velocity of vehicle in fact does not change violently;

relative speed of vehicles is small. The inaccuracy of vehicle density measurement is small though the hello interval is large.

2

(a) High speed  (b) Low speed 

Figure 18. Average delay of VDEB

The average end-to-end delay of VDEB gets larger as the hello interval grows up.

Figure 19 shows such variations. We conclude that the hello interval is not so critical

  32

for our VDEB scheme. The impact of hello interval in VDEB scheme is so small that it can be neglected. To achieve the similar delay performance to the MCDS scheme, we can use larger hello interval in our VDEB scheme. VDEB scheme gives the network few overhead.

(a) High speed  (b) Low speed 

Figure 19. Farthest end-to-end delay of VDEB

Table 3. Average retransmissions of VDEB VDEB retransmissions

Table 3 presents the average retransmissions of VDEB scheme. Because the ring width is bounded by the estimated vehicle density, the performance does not get worse as the hello interval grows up. As the hello interval grows up, the average

retransmission only slightly increases due to the misestimated ring width. Ring width that is larger than needed ring width results in more vehicles rebroadcasting the emergency messages.

Figure 20. Average delay comparison

0

Figure 21. Average farthest end-to-end delay comparison

Figures 20 to 22, and Table 6 compare the performance among MCDS scheme, VDEB scheme, and DBS scheme. We first compare MCDS with VDEB under the equal hello interval. When the hello interval is 3.2 seconds, Figure 20 and 21 show the difference of delay time for MCDS, VDEB, and DBS schemes. The average delay of MCDS is even larger than the DBS scheme. This claims that if we want to get the good performance of MCDS, the high frequently broadcasted hello message is needed.

  34

If we want to reach the delay as VDEB scheme with hello interval 6.4 seconds, the hello interval of MCDS should be about 0.4 seconds. The overhead of MCDS with hello interval 0.4 seconds is 16 times that of VDEB with hello interval 6.4 seconds. If the hello interval of MCDS is short enough, the delay of MCDS and VDEB are both shorter than that of DBS. The difference of delay among these schemes results from the longer backoff time of DBS.

As shown in Table 4, the standard deviation of DBS’s average delay is highest among these schemes. The reason is that when the vehicle density is low, the average delay gets obviously high. The standard deviation of average delay of MCDS is slightly higher than VDEB. This is caused by the failure of forwarding in which case the chosen forwarder is not in the transmission range of the sender. Therefore, we claim that VDEB scheme has the most stable delay performance regardless of vehicle density.

Table 4. Standard Deviation of average delay Standard deviation (SD) of average delay

MCDS VDEB DBS

Hello Period (s) SD Hello Period (s) SD SD

0.2 0.2718 3.2 0.1749 0.4 0.2767 6.4 0.1228

0.8 0.307 12.8 0.0688

1.6 0.1608 25.6 0.1507

3.2 0.4141 - -

0.5297

In addition, DBS also performs badly in the number of retransmission. As shown in Figure 22, the number of retransmission of DBS increases when the vehicle density

gets higher. This is because that there are more vehicles with familiar backoff time if the vehicle density gets higher. In the MCDS scheme, because the next forwarder is chosen before broadcasting the message, it will not cause multiple forwarders. Our VDEB scheme is a receiver-oriented method, so the multiple-forwarder problem will occur. With the help of vehicle density information, we can reduce the number of vehicles which have the similar backoff time. In other words, the number of vehicles broadcasting the message is not proportional to the vehicle density. Figure 22 shows such results. Regardless of the vehicle density, the number of average retransmissions is not greater than 15 for both MCDS and VDEB schemes. The number of average retransmissions of VDEB is almost the minimum number of DBS.

0

Figure 22. Average Retransmissions comparison

Besides the average delay, the delay percentile is also an important measure to evaluate the performance of a protocol. Consider the high speed situation. Two vehicle densities 30 and 80 are compared. The hello intervals of MCDS and VDEB are set to 0.4 seconds and 6.4 seconds, respectively. As we can see in Figure 23, the percentiles of delay under 4 milliseconds are 85% for MCDS (Vehicle density is 30), 74% for VDEB, 65% for MCDS (Vehicle density is 80), 47% for DBS (Vehicle density is 80), and 44% for DBS (Vehicle density is 30). In the low mobility case, the

  36

rate of forwarding failure in MCDS is less than that in high mobility case. Therefore, the delay percentile of MCDS with low vehicle density is larger than that with high vehicle density. Delay percentiles of VDEB are not influenced by the vehicle density.

DBS has the worst performance among these schemes due to its random nature. For the delay under 7 milliseconds; delay percentiles of VDEB outperform both MCDS and DBS schemes. Almost all the receivers in VDEB receive the emergency message within 8 milliseconds. Clearly VDEB broadcasts the emergency message with smaller delay for all the vehicles in the risk zone.

0%

20%

40%

60%

80%

100%

0 2 4 6 8 10 12 14 16 18 20

Delay (ms)

Percentiles (%)

MCDS(HI=0.4, VD=30) VDEB(HI=6.4, VD=30) DBS(VD=30) MCDS(HI=0.4, VD=80) VDEB(HI=6.4, VD=80) DBS(VD=80)

  Figure 23. Delay percentiles comparison

Throughout the simulation, the number of total transmitted hello messages is summarized in Figure 24. When the vehicle density is 100 vehicles per kilometers, the number of total transmitted hello messages in MCDS (hello interval is 0.4 seconds) is three hundred thousands. That is, there are 250 hello messages transmitted per kilometers per second in MCDS while in VDEB (hello interval is 6.4 seconds), there

are only 16 hello messages transmitted per kilometers per second for the similar delay performance. VDEB consumes very low overhead to obtain the even better performance than MCDS at the expense of a little more retransmissions (see Figure 20).

0 50000 100000 150000 200000 250000 300000 350000

20 30 40 50 60 70 80 90 100

Vehicle Density (vehicles/km)

Hello Packets

MCDS(0.4s) VDEB(6.4s)

  Figure 24. Overhead of MCDS and VDEB

  38

Chapter 5 Conclusion  

In this thesis, we proposed a vehicle-density-based multi-hop broadcast scheme, called VDEB, for emergency message forwarding. Because the sender-oriented schemes have the drawback of high overhead and the receiver-oriented schemes causes longer delay, our VDEB scheme resolves both of these problems and provides lower delay and lower overhead for emergency message broadcasting. In the VDEB scheme, receiver-oriented contention mechanism is adopted with the vehicle density measurement component. Vehicle density can help to reduce the number of retransmissions of message and avoid the situation that the delay time gets larger if the forwarder is not far enough. In addition, the number of forwarders is limited by the ring width which is estimated by the vehicle density and neighbor information. In the VDEB scheme, the number of retransmissions is not proportional to the vehicle density. The simulation results show that our VDEB scheme provides a good delay performance with reasonable overhead that do not hurt the general data services in VANETs.

References

[1] L. Armstrong, “Dedicated Short Range Communications (DSRC),” [Online].

Available: http://www.leearmstrong.com/dsrc/DSRCHome.htm

[2] K. V. N. Kavitha, A. Bagubali, and L. Shalini, “V2V wireless communication protocol for rear-end collision avoidance on highways with stringent propagation delay,” In Proceedings of the International Conference on Advances in Recent Technologies in Communication and Computing (ARTCom 2009), pp. 661-663, 2009.

[3] S. Biswas, R. Tatchikou, and F. Dion, “Vehicle-to-vehicle wireless communication protocols for enhancing highway traffic safety,” IEEE Communications Magazine, vol. 44, no. 1, pp. 74-82, 2006.

[4] M. Durresi, A. Durresi, and L. Barolli, “Emergency broadcast protocol for inter-vehicle communications,” In Proceedings of the 11^th International Conference on Parallel and Distributed Systems, vol. 2, pp. 402-406, 2005.

[5] L. Bononi and M. D. Felice, “A cross layered MAC and clustering scheme for efficient broadcast in VANETs,” In Proceedings of the IEEE International Conference on Mobile Adhoc and Sensor Systems (MASS 2007), pp. 1-8, 2007.

[6] T. Taleb, K. Ooi, and K. Hashimoto, “An efficient collision avoidance strategy for ITS systems,” In Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC 2008), pp. 2212-2217, 2008.

[7] L. O. Djedid, N. Lagraa, M. Yagoubi, and K. Tahari, “Adaptation of the MCDS

  40

International Conference on Innovations in Information Technology (IIT 2008), pp. 534-538, 2008.

[8] Z. Shi, F. Liu, and S. Xu, “Novel relay scheme based on traffic type in vehicular networks,” In Proceedings of the International Conference on Intelligent Computation Technology and Automation (ICICTA 2008), vol. 2, pp. 392-397, 2008.

[9] P. Lai, X. Wang, N. Lu, and F. Liu, “A reliable broadcast routing scheme based on mobility prediction for VANET,” In Proceedings of the IEEE Intelligent Vehicles Symposium, pp. 1083-1087, 2009.

[10] G. Korkmaz, E. Ekici, and F. Ozguner, “An efficient fully ad-hoc multi-hop broadcast protocol for inter-vehicular communication systems,” In Proceedings of the IEEE International Conference on Communications (ICC 2006), vol. 1, pp.

423-428, 2006.

[11] M. M. I. Taha and Y. M. Y. Hasan, “VANET-DSRC protocol for reliable broadcasting of life safety messages,” In Proceedings of the IEEE International Symposium on Signal Processing and Information Technology, pp. 104-109, 2007.

[12] M. M. I. Taha and Y. M. Y. Hasan, “A novel headway-based vehicle-to-vehicle multi-mode broadcasting protocol,” In Proceedings of the IEEE Vehicular Technology Conference (VTC 2008-Fall), pp. 1-5, 2008.

[13] S. Yu and G. Cho, “A selective flooding method for propagating emergency messages in vehicle safety communications,” In Proceedings of the International Conference on Hybrid Information Technology (ICHIT 2006), vol. 2, pp.

556-561, 2006.

[14] C. Chiasserini, R. Gaeta, M. Garetto, M. Gribaudo, and M. Sereno, “Efficient broadcasting of safety messages in multihop vehicular networks,” In

Proceedings of the 20^th International Parallel and Distributed Processing Symposium (IPDPS 2006), pp. 8, 2006.

[15] Y. T. Yang and L. D. Chou, “Position-based adaptive broadcast for inter-vehicle communications,” In Proceedings of the IEEE International Conference on Communications (ICC 2008), pp. 410-414, 2008.

[16] H. Alshaer and E. Horlait, “An optimized adaptive broadcast scheme for inter-vehicle communication,” In Proceedings of the IEEE Vehicular Technology Conference (VTC 2005-Spring), vol. 5, pp. 2840-2844, 2005.

[17] S. Oh, J. Kang, and M. Gruteser, “Location-based flooding techniques for vehicular emergency messaging,” In Proceedings of the 3rd Annual International Conference on Mobile and Ubiquitous Systems, pp. 1-9, 2006.

[18] X. Yang, L. Liu, N. H. Vaidya, and F. Zhao, “A vehicle-to-vehicle communication protocol for cooperative collision warning,” In Proceedings of the First Annual International Conference on Mobile and Ubiquitous Systems:

Networking and Services (MOBIQUITOUS 2004), pp. 114-123, 2004.

[19] Y. Zang, L. Stibor, H. Reumerman, and H. Chen, “Wireless local danger warning using inter-vehicle communications in highway scenarios,” In Proceedings of the 14th European Wireless Conference (EW 2008), pp. 1-7, 2008.

[20] Q. Yu and G. Heijenk, “Abiding geocast for warning message dissemination in vehicular ad hoc networks,” In Proceedings of the IEEE International Conference on Communications (ICC 2008), pp. 400-404, 2008.

[21] J. Peng and L. Cheng, “A distributed MAC scheme for emergency message dissemination in vehicular ad hoc networks,” IEEE Transactions on Vehicular Technology, vol. 56, no. 6, pp. 3300-3308, 2007.

[22] L. Yang, J. Guo, and Y. Wu, “Piggyback cooperative repetition for reliable

  42

Consumer Communications and Networking Conference (CCNC 2009), pp. 1-5, 2009.

[23] D. Shin, H. Yoo, and D. Kim, “EMDOR: Emergency message dissemination with ACK-overhearing based retransmission,” In Proceedings of the First International Conference on Ubiquitous and Future Networks (ICUFN 2009), pp.

230-234, 2009.

[24] T. Toyserkani, E. G. Strom, and A. Svensson, “An efficient broadcast MAC scheme for traffic safety applications in automotive networks,” In Proceedings of the IEEE Wireless Communications and Networking Conference (WCNC 2006), vol. 4, pp. 2100-2105, 2006.

[25] “The network simulator: NS-2,” [Online]. Available:

http://www.isi.edu/nsnam/ns