Efficient Broadcast Mechanism for Cooperative Collision Avoidance Using Power Control

Efficient Broadcast Mechanism for Cooperative Collision

Avoidance Using Power Control

Andy An-Kai Jeng, Rong-Hong Jan, Chien Chen

National Chiao Tung University Hsinchu, 300, Taiwan, R.O.C. {andyjeng, rhjan, cchen}@cis.nctu.edu.tw

Tsun-Chieh Chiang

Information & Communications Research Laboratories Industrial Technology Research Institute

Hsinchu, 310, Taiwan, R.O.C [email protected]

Abstract—Improving driver’s safety has been an active

research area in wireless communication. In particular, the vehicle cooperative collision avoidance (CCA) is one of the most important issues in safety applications. A variety of broadcast protocols has been proposed for vehicular network. However, there is only a few of them dedicatedly designed for the CCA system. In this paper, we propose a novel broadcast mechanism for CCA using the power control technique. The power control rule is based on the safe distance between vehicles. Simulation results show that our approach can significantly reduce the delivery delay and avoid car collision.

Keywords-vehicular networks; power control; cooperative collision avoidance

I. 0BINTRODUCTION

Traffic accidents have been taking thousand of lives each year, exceeding any deadly disease or natural disasters in many countries. Numerous factors, such as bad weather conditions and mechanical failures, may lead to a traffic accident. In particular, the inability of drivers to react in time to emergency events often creates to a series of car crashes, i.e. the chain car collision. As shown in Fig. 1a, three vehicles A, B, and C are driving on a highway platoon. When vehicle A brakes suddenly, vehicle B can start to decelerateafter a driver reaction time, i.e. the duration when an event is observed and when the driver actually applies the brake, to avoid a collision with A. However, due to the line-of-sight limitation from B, vehicle C may not decelerate until its driver has seen the rear brake light of vehicle B. Studies show [1] that the driver reaction time could range from 0.75 to 1.5s, which means that a trailing vehicle may keep running for a long distance before reacting to an accident ahead. For instance, at a speed of 70 mph, vehicle C may pass through 75 to 150 ft before being decelerated. Consequently, a single emergency event often leads to a string of secondary crashes. Clearly, such an undesirable situation can be substantially avoided or lessened if drivers can be warned earlier.

The Cooperative Collision Avoidance (CCA) is an important class of safety applications in Intelligent Transport Systems (ITS), which aims at offering earlier warning to drivers using vehicle-to-vehicle (V2V) communication [2]. As the example shown in Fig. 1b, once vehicle A confronts

an accident, it can directly send out a warning message to C or quickly forward the warning message hop-by-hop to C whenever their distance is beyond the transmission range. As a result, vehicle C can obtain more chance to stop safely, in contrast to counting on the rear brake lights of vehicle B.

(a)

(b)

Figure 1. (a) alerted by rear brake lights; (b) alerted by warning messages.

However, due to the severe interference in wireless communication, the deliver delay could be intolerably large, especially when many vehicles have to transmit or forward their warning messages. The delay would result in a longer time to response to the emergency event. The interference problem would become more significant in density traffic roads or multi-lane environments.

In this paper, we present a new broadcast mechanism for the CCA system, named PC-CCA. The PC-CCA employs the power control (PC) technique to reduce the physical interference incurred by delivering warning messages. The power control technique has been considered as an effective way to lessen interference in the wireless environments. By reducing the transmission power of each vehicle, the broadcast radius can be smaller, implying fewer warning messages being forwarded and fewer nodes being interfered. The rest of this paper is organized as follows. In Section II, we review recent researches related to broadcasting in the Vehicular Ad Hoc Network (VANET) and the CCA system. Section III presents the proposed broadcast mechanism. In Section V, we conduct simulation results. Conclusion is remarked in the last section.

II. 1BRELATED WORK

VANET Broadcast has been studied in several articles, such as in [3, 4, 5, 6, 7]. Xu et. al. [3] discussed a

to-vehicle location-based broadcast protocol, where each vehicle generates a warning message at a constant rate. The optimum transmission probability at MAC layer for each message is then identified to reduce the packet collision probability. In [4], the authors propose a multi-hop broadcast protocol based on slot reservation MAC. Considering the scenario that not all vehicles will be equipped with wireless transceivers, forwarding in sparsely connected ad hoc network consisting of highly mobile vehicles is studied in [5]. Motion properties of vehicles are exploited in [6] to help with message relay. In [7], the authors proposed efficient protocols to reduce the amount of messages being forwarded. Compared with MANET broadcast, the above protocols concern the mobility of vehicles to achieve more efficient message forwarding. However, these protocols are not specifically for safety applications (e.g. CCA system), where more emphasis should be paid on the emergency of warming messages.

Several application challenges in the CCA system, such as stringent delay requirement, coexisting abnormal vehicles, and different emergency levels, have been identified in [8]. The authors also designed a protocol comprising congestion control policies, service differentiation mechanism, and methods for emergency warning dissemination. In [9], a broadcast scheme based on a client-server platform is proposed. The rebroadcast probability of each relaying vehicle is changed dynamically according to the number of vehicles insides the transmission zone. The purpose is to avoid relaying redundant warming messages so as to reduce delivery delay. However, this protocol requires each vehicle to acquire information in its two hops range. The control overhead may lead to additional delivery delay.

In order to perform forwarding without prior knowledge about neighbors, Biswas et. al. propose two context-aware protocols [10], named the naive broadcast (NB) and intelligent broadcast with implicit acknowledgment (I-BIA), for the CCA system. In both protocols, when an emergency event occurs, the source vehicle broadcasts a warming message first, and then a recipient will forward the message only if the direction-of-arrival (DoA) of the message is in front of itself. This mechanism ensures that the warming message will be eventually delivered to all vehicles behind the source vehicle and any vehicle which is not endangered will not forward the message. The I-BIA can further avoid redundant retransmission by setting a waiting time to see if there is any vehicle behind a recipient having received the same message. Similarity, three context-award protocols, named weighted p-persistence broadcasting, slotted 1-persistence broadcasting, and slotted p-1-persistence broadcasting, are proposed in [11]. In these protocols, vehicles which are farther away from the previous broadcaster will transmit with higher priority (higher probability or earlier time). The purpose is to avoid redundant retransmissions from intermediate vehicles. A similar protocol is presented in [12] for multi-lane highway.

Not endangered Collided

Not necessary A C B D E (a) 80km/h, 4.9m/s2 Not endangered 70km/h, 4.5m/s2 75km/h, 4.1m/s2 A C D E 60km/h, 2.3m/s2 Safe distance Safe distance B (b)

Figure 2. (a) CCA without power control; (b) CCA with power control.

Although the above protocols can make use of directional or other geographic information to reduce overall delay, the local delay may not meet the requirement for each individual vehicle. Consequently, chain car collision may still occur if even the overall delay is low. To solve this problem, a

risk-aware protocol is presented in [13]. In this protocol, vehicles

are classified into several clusters according to the features of their movement. Then, an emergency level is defined for each vehicle based on the order in its cluster. The emergency level reflects the risk of a vehicle to meet an emergency situation in the platoon. The medium access delay of each vehicle is then set as a function of its emergency level to promptly disseminate warning message. However, the order in cluster cannot explicitly reflect the risk, because in real situation many factors, such as intercar space and velocity, are inconsistent from vehicle to vehicle. Besides, the interference is still severe if the physical transmission range is large. To the best of our knowledge, there is no research using power control technique to improve the CCA system.

III. 2BPROPOSED BROADCAT MECHANISM

In this section, we present our broadcast mechanism for the CCA system. First of all, the basic concept is described. After that, we formally model the safe distance between vehicles in vehicular network environment. The power control rule is then summarized in the last part.

A. 5BBasic Concept

The interference may occur when more than one vehicle has to forward the same message within a short period. An example is shown in Fig 2a. Once vehicles B and C received a warning message from A, because they can not be aware of each other, they may forward the message at the same time to the vehicles behind, resulting in a signal collision at vehicleD.ThePC-CCA employs the power control technique to physically reduce the interference. As shown in Fig. 2b, by reducing the transmission power, vehicle D can avoid receiving messages simultaneously from both B and C, since only vehicle B receives the message from A at the first place.

The major problem is how to guarantee the delivery to all vehicles which are endangered as long as the transmission power is shrunk. To tackle this problem, our protocol will dynamically adjust the transmission radius based on the safe distance between vehicles. As shown in

Fig. 2b, under the given velocities and deceleration rates, we assume that the safe distance between vehicles A and B is d, which means that vehicle B is potentially endangered if its distance to A is shorter than d. In other words, to avoid being collided by vehicle B, the transmission radius of A should be at least d.

Using the power control technique can also avoid transmitting to vehicles which are not endangered. As shown in Fig. 2a, vehicle E is far away from the platoon, i.e. it is out of the safe distance to D. But, if vehicle D always transmits at the maximum transmission power, vehicle E will eventually receive a warning message from D even if it is not endangered. In contrast to Fig. 2b, if vehicle D shrinks its power according to its safe distance to E, vehicle E will never be disturbed and it can avoid relaying useless messages to the vehicles behind any further. In other words, the covered area can be confined into a smaller zone to avoid redundant bandwidth usage.

B. 6BModeling Safe Distance

Before presenting our power control rule, the safe distances between vehicles in vehicular network environments should be carefully modeled. As shown in Fig. 3, three vehicles Ci+1, Ci and Ci-1 are on a highway platoon,

where Ci-1 is in front of Ci and Ci+1 is behind Ci. Assuming

that Ci-1 is the vehicle confronted an accident, we aim to

formulate the safe distance Si,i+1 that Ci+1 should be kept

from Ci. The safe distance Si,i+1 is then used to model the

necessary transmission radius Ti,i+1 between Ci and Ci+1 and

broadcast radius Bi of Ci. Other symbols used in our model

are listed in Table 1. Note that we assume each vehicle can obtain its current position and the UTC time from a Global Positioning System (GPS).

TABLE I. SYMBOLS

Symbols Meanings

Vi Velocity of Ci;

Di Deceleration rate (regular or emergency deceleration) of Ci;

G Average driver reaction time;

L Car length;

ti UTC time when Ci applies emergency braking or receives a warming message from Ci-1 at network queue;

'i-1,i 'i-1,i = ti – ti-1: delivery delay from Ci-1 to Ci;

di-1,i Distance between the position of Ci at ti and the position of

Ci-1 at ti-1;

Mi Moving distance of Ci after ti;

Si,i+1 Safe distance between Ci+1 and Ci at time ti;

Ti,i+1 Transmission radius from Ci to Ci+1 at time ti;

Bi Broadcast radius of Ci at time ti;

First of all, we need to estimate the moving distance Mi

for Ci. The Mi represents the distance that Ci has to run

through after Ci-1 confronted an accident. The model of Mi

has three cases, corresponding to the cases of soft brake, medium brake, and hard brake in Fig. 2.

Case 1: Ci stops safely

Case 2: Ci collides with Ci-1 after (or when) Ci-1 stopped;

Case 3: Ci collides with Ci-1 before Ci-1 stops;

In case 1, Ci applies a hard brake so that it can stop

safely before colliding with Ci-1. Therefore, after Ci received

a warning message from Ci-1, it will move at the original

velocity Vi during the driver reaction time G and then move

at the decelerated velocity for a period of Vi/Di before

stopping. Let l(V, D, t) stand for the moving distance of a vehicle with velocity V and deceleration rate D during a period of time t. That is,

2 2 ) , , (V Dt Vt Dt l  .

The moving distance of Ci after ti can be represented as

i i i i i i i i i D V V D V D V l V M 2 ) / , , ( 2   G G .

In case 2, since Ci collided with Ci-1 after (or when) Ci-1

stopped, its moving distance is depending on the moving distance of the vehicle ahead, i.e. Mi-1. Therefore, assuming

that the distance between the position of Ci at ti and the

position of Ci-1 at ti-1 is di-1,i, the moving distance of Ci after

ti is the moving distance of Ci-1 (i.e. Mi-1) plus their distance

di-1,i. Note that the car length L should be subtracted. That is,

the Mi in this case can be given by

L M d

Mi i1,i i1 .

Case 3 further has three subcases: 3.1, 3.2 and 3.3. Let tx

denote the moving time of Ci before collided and Fi–1,i

temporally denote the moving distance in this case. In subcase 3.1, Ci collides with Ci-1 before both of them

decelerate, which means that

i x i i ,1 tV

F ,

where tx satisfies that

1 1 ,    ii x i i xV L d tV t .

In subcase 3.2, Ci collides with Ci-1 before Ci-1 decelerates

and after Ci-1 decelerated, which means that

i x i i ,1 tV

F ,

where tx satisfies that

) ( ) , , ( 1 1 1 1 ,i i i i x t i i xV L d V lV D t t   G      G' . In subcase 3.3, Ci collides with Ci-1 after both of them

decelerated, which means that

) , , ( , 1 G G F_{i i} V_ilV_i D_i t_x ,

where tx satisfies that

)). ( , , ( ) , , ( 1 1 1 1 ,i i i i x t i x i i i t D V l V d L t D V l V '            G G G G

Combining the above cases, we have the following function for the moving distance Mi:

° ° ¿ ° ° ¾ ½ ° ° ¯ ° ° ® ­       i i i i i i i i i d M L D V V M , 1 1 , 1 2 , 2 min F G

Figure 3. Safe distance and Transmission radius

Based on the moving distance Mi, we can now model

the safe distance Si,i+1 between Ci and Ci+1 for Ci. Assuming

that Ci can obtain the velocity Vi+1 and deceleration Di+1 of

vehicle Ci+1, it can know that Ci+1 will move for a distance

of l(Vi1,Di1,Vi1/Di1) after reacting the warning message from itself. On the other hand, before reacting to the warning message, Ci+1 has to move for a distance of GVi1 during the driver reaction time. Furthermore, there is a propagation delay 'i,i+1 so that Ci+1 has to move at the

original velocity Vi+1 for a distance of 'i,i+1Vi+1. As a result,

the safe distance Si,i+1 between Ci and Ci+1 can be modeled

as i i i i i i i i i i V lV D V D L M S,_1 (',_1G) _1 ( _1, _1, _1/ _1)  .

C. 7BPowr Control Rule

As mentioned above, to send a warning message to Ci+1,

the transmission radius of Ci should be at least the safe

distance Si,i+1. Furthermore, because the velocities of Ci and

Ci+1 are not always the same, vehicle Ci+1 may not receive

any message from Ci if their distance was enlarged during

the message propagation, i.e. the duration 'i,i+1. For this

reason, the transmission radius of Ci to Ci+1 should add the

enlarged gap. That is,

} 0 , max{ 1 1 , 1 , 1 ,i ii 'ii i i i S V V T .

Now, assume that Ci can be aware of the statuses of all

vehicles behind. The broadcast radius of Ci can be set as

^

`

where Pi is the set of vehicles behind Ci and Ht 0 is a factor

to cope with the possible wireless channel fading.

However, if the statues of the trailing vehicle are unknown, we can estimate the safe distance, transmission power broadcast radius, respectively, by

i r

r

i V lV D V D L M

Sˆ (WG) max ( max, , max/ )  , } 0 , max{ ˆ ˆ min V V S Ti iW i , and i i T Bˆ (1H)ˆ ,

where Vmax denotes the maximum velocity (or upper speed limit), Vmin denotes the minimum velocity (or lower speed limit), and Dr is the regular deceleration. Note that the

optimal value of H can be turned by simulation or some rational function. The above models are also applicable to any vehicle Ci in a platoon. In such a case, the Ci-1 presents

the vehicle that has received a warning message from a vehicle ahead (e.g. Ci-2).

IV. 3BSIMULATIONS

In this section, we conduct simulations to evaluate the proposed mechanism. We use the ns-2 network simulator [14] to simulate a highway scenario, where 50 vehicles driving on a highway platoon toward the same direction. Vehicle emergency situations are created by forcing the vehicle at the front of the platoon (i.e. vehicle 0) to rapidly decelerate (8m/s2_{), which triggers a CCA process by initiating a} warning message. Any vehicle behind will decelerate at the regular rate (4.9m/s2_{) whenever it has received a warning} message for a driver reaction time, randomly chosen from 0.75s to 1.5s.

The transmission medium is IEEE 802.11 MAC. The broadcast throughput is throughput is 1Mbps and the

maximum transmission range is 250m. We will compare the cases with and without the power control mechanism. In order to evaluate the performance under the same base, we employ the naive broadcast [10], a direction-ware broadcast protocol for CCA, to forward any warning message at the network layer. Other parameters used in our simulation are listed in Table 2, which are mostly adapted from [10]. Note that to add dynamics in our test vehicle speed and inter-car spacing have 10% variations. Besides, we assume the maximum speed (Vmax) and minimum speed (Vmin) are available to each vehicle so that each vehicle Ci can estimate

its broadcast radius Bi. The channel fading factor H is set as

0.1 in our test. All results are averaged from 10 runs.

TABLE II. PARAMETERS SETTINGS

Parameters Values Number of vehicles on each lane 50

Vehicle length 4m

Vehicle speed 32m/s r 10%

Regular deceleration 4.9/m/s/s

Emergency deceleration 8/m/s/s

Inter-car spacing [9.6 – 28.8] m r 10%

Driver’s reaction time [0.75 – 1.5] s

Radio model Two ray ground

MAC protocol IEEE 802.11 DCF

Broadcast protocol Naïve broadcast

Message size 20 bytes

Random wait time [0 –10] ms

Simulation runs 10

Fig. 4 shows the number of collided vehicles under varied average inter-car spacing. We can see that there are no more than a half of collisions being avoided if each vehicle always transmits or forwards at the maximum transmission radius. Contrarily, by using the power control technique, the possibility of a car collision can be greatly reducedespecially when the inter-car spacing is reasonably large.

Figure 4. Number of collided vehicles under varied inter-car spacing

Such an impressive improvement is primarily the consequence of the reduced delivery delay. As shown in Fig. 5, with an average inter-car space of 28.8 m, the maximal delay required to delivery a warning message all vehicles can be confined in 8 ms if the PC-CCA is used. By contrast, the delivery delay without the PC-CCA increases drastically to

the trailing vehicles, implying that more vehicles are not able to receive a warning message in time and brake safely.

Figure 5. Delivery delay for each vehicle in the platoon (inter-car spacing: 28.8 m r 10%)

V. 4BCONCLUSION

In this paper, we have proposed an efficient broadcast mechanismfortheCCAsystemusingpowercontrol technique. The main idea for controlling power is based on the safe distance between vehicles. Simulation results show that our mechanism indeed helps to reduce delivery delay and car crashes.

8B

ACKNOWLEDGMENT

This research was supported in party by the National Science Council, Taiwan, ROC, under grants NSC97-2221-E-009-048-MY3 and NCS97-2221-E-009-049-MY3.

9B

REFERENCES

[1] M. Green, “How long does it take to step? Methodological analysis of driver perception-brake times,” Transportation Human Factors, vol. 2, no. 3, pp. 195-216, 2000.

[2] ASTM E22213-03, “Standard specification for telecommunication and information exchange between roadside and vehicle sytem – 5GHz band dedicated short range communications (DSRC) MAC and PHY specifications,” ATM International, July, 2003.

[3] Q. Xu, R. Sengupta, and D. Jiang, “Design and analysis of highway safety communication protocol in 5.9 GHz Dedicated short range communication spectrum,” Proc. of IEEE Vehicular Technology Conference, Spring, 2003.

[4] L.B. Michael and M. Nakagawa, “Non-platoon inter-vehicle communication using multiple hops,” IEICE Transacations on Communications, vol. E82-B, no. 10, pp. 1651-1658, October, 1999. [5] L. Briesemeistr and G. Hommel, “Role-based multicast in highly

mobile but sparsely connected ad hoc networks,” Proc. of the 2000 ACM International Symp. on Mobile Ad Hoc Networking & Computing, pp. 45-50, August, 2000.

[6] Z.D. Chen, H. Kung, and D. Vlah, “Ad hoc relay wireless networks over moving vehicles on highways,” Proc. of the 2001 ACM International Symp. on Mobile Ad Hoc Networking & Computing, pp. 247-250, 2001.

[7] M.T. Sun, W.C. Feng, T.H. Lai, K. Yamada, and H. Okada, “GPS-based message broadcast for adaptive inter-vehicle communications”

Proc. of IEEE Vehicular Technology Conference, pp. 2685-2692, 2000.

[8] X. Yang, J. Liu, F. Zhao, N.H. Vaidya, “A vehicle-to-vehicle communication protocol for cooperative collision warning”, Proc. of 1st _{Annual International Conference on Mobile and Ubiquitous} Systems, pp. 114-123, 2004.

[9] H. Alshaer, and E. Horlait, “An optimized adaptive broadcast scheme for inter-vehicle communication,” Proc. of IEEE Vehicular Technology Conference, pp. 2840-2844, Jun, 2005.

[10] 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. 535-547, 2006.

[11] N. Wisitpongphan, O.K. Tonguz, J.S. Parikh, P. Mudalige, F. Bai, and V. Saderar, “Broadcast storm mitigation techniques in vehicular ad hoc networks”, IEEE Wireless Communications, vol. 14, no. 6, pp. 84-94, 2007.

[12] Y.T. Yang, L.D. Chou, “Position-based adapted broadcast for inter-vehicle communications”, Proc. of IEEE International Conference on Commuinations, pp. 410-414, 2008.

[13] T. Taleb, K. Ooi, and K. Hashimoto, “An efficient collision avoidance strategy for ITS systems”, Proc. of IEEE Wireless Communication & Networking Conf., pp. 2212-2217, 2008.