improving reliability of platooning control messages using...

Improving Reliability of Platooning ControlMessages Using Radio and Visible Light Hybrid

CommunicationSusumu Ishihara

Shizuoka UniversityHamamatsu, 432–8561, Japan

Email: [email protected]

Reuben Vincent Rabsattand Mario Gerla

Computer Science DepartmentUniversity of California, Los AngelesLos Angeles, CA 90095–1596, USAEmail: {rrabsatt, gerla}@cs.ucla.edu

Abstract—Autonomous platoon control is a promising autodrive technology. It frees drivers from stressful driving taskand offers comfortable car trip experience. It is also expectedto contribute to energy savings. In platooning control systems,beacon messages have an important role. Each member vehiclecontrols its acceleration according to the information of theposition, speed, and acceleration included in the beacons from 1)the leader vehicle and 2) its preceding vehicle. Since the position,speed, and acceleration of a vehicle can be observed by its directfollower vehicle using on-board sensors, the beacon for precedingvehicle is not so critical. In contrast, the beacon transmitted fromthe leader vehicle to the member vehicles cannot be replacedby on-board sensors and is the key factor that requires highreliability. However, radio communication from the leader to themember vehicles is vulnerable to radio jamming attacks. Thispaper proposes a radio and visible light communication (VLC)hybrid protocol to tackle RF jamming attacks. Simulation resultsshow that the proposed hybrid protocol can avoid long end-to-end delay and achieve high message delivery ratio even underRF jamming attacks.

I. INTRODUCTION

Autonomous platooning control is a promising auto drivetechnology as well as standalone autonomous driving such asGoogle Self-Driving Car [1]. It frees drivers from stressfuldriving task and offers comfortable car trip experience. It isalso expected to contribute to energy savings. So far, variousprojects such as [2], [3], and [4] have struggled with devel-oping autonomous platooning systems. Fig. 1 shows a typicalarchitecture of an autonomous platooning system based oninter-vehicular communication (IVC), or cooperative adaptivecruise control (CACC) [5] [6].

In addition to the sensor-based autonomous cruise control(ACC), in which each vehicle controls its velocity accordingto the distance and velocity of its preceding vehicle, vehiclesin an autonomous platoon have a function to control theirvelocity according to the information from the leader of theplatoon and the preceding vehicle. Since autonomous platooncontrol depends on wireless communication, its performanceand safety are strongly affected by the control of messageexchange in the platoon.

Leader to Followers

Inter-vehicle distance: e.g. 5-7m

To the direct follower

Message update frequency: 10-50Hz

Camera/!RADAR/!LIDAR

Fig. 1. Autonomous Platoon

In typical platooning systems, the leader vehicle periodicallysends its position, velocity, and acceleration to the member ve-hicles periodically. For example, [6] recommends informationupdate frequency of 10Hz. In addition to position, velocityand acceleration, kinematic data of the leader vehicle wouldbe useful. We call a message including these data a beaconmessage in this paper. Control of platooning systems dependson the reliability of the beacon message sent from the leader,because the information delivered by the beacon messagecannot be directory monitored by platoon member vehicles.On the other hand, the position, speed, and acceleration of thepreceding vehicle can be monitored by using on-board sensors,such as millimeter wave radars. Thus, wireless communicationbetween neighboring vehicles in the same platoon for tellingthe position, speed, and acceleration is not indispensable [7].

In addition to guaranteeing the reliability of message ex-change protocols in normal situations, prevention from attacksto the wireless communication, such as forged messages andjamming, is crucial for platooning systems. In this paper, wefocus on RF jamming attacks. Today, PC-based or FPGAbased software radio platforms such as GNU Radio/USRP[8] and Wireless Open-Access Research Platform (WARP)[9] are not difficult to obtain and quickly implement wirelessjamming attacking tools. Punalet al. have developed somejamming attacking tools using WARP board to evaluate the

2015 IEEE Vehicular Networking Conference (VNC)

978-1-4673-9411-6/15/$31.00 ©2015 IEEE 96

risk of various jamming attacking methods using real cars[10]. The risk of jamming attack is inevitable. Thus, weneed countermeasures against such attacks. Even if vehiclesin a platoon are under a jamming attack, the control messagedelivery ratio and end-to-end delivery delay have to satisfy therequirement needed for the platooning control.

In this paper, we propose a radio and visible light com-munication (VLC) hybrid protocol for delivering the pla-toon leader’s beacon messages reliably under radio jam-ming attacks, and evaluates the effectiveness of the pro-tocol through simulations. Radio communications such asDSRC/IEEE802.11p [11], especially in cases with omni-directional antennas, are vulnerable to jamming attacks be-cause anyone in the range of the radio communication cansend jamming signal to the victims. On the other hand, visiblelight communication (VLC), which is expected to be used invehicle-to-vehicle communications using LED tail and headlights [12], has little influence on jamming attacks becausethe directivity of the VLC receiver is narrow and the attackershave to point the moving victim accurately to saturate the VLCphotodiode receiver with strong light. Such an attack on VLCcan only be performed on a single VLC link, as opposed toall vehicles in the range in the case of RF. This implies thatthe impact on the overall platoon control could be minimal.

It is difficult to send messages directly from the leader toall the members in a platoon through VLC due to the sharpdirectivity and the vehicles’ bodies that work as obstacles.Thus, the communication from the leader to members in theplatoon through VLC has to be in a multi-hop manner. Ifthe link speed is low, the multi-hop transmission leads tolong end-to-end delay. However, the delay can be reduced byforwarding the message using radio communication and VLCcomplementary. Of course, though the radio communicationmay be affected by a jamming attack, the effect can bemitigated by using VLC.

The main contributions of this paper are summarised asfollows.

• We propose a radio and VLC hybrid communicationprotocol for platooning control messages for reliablemessage delivery under RF jamming attacks.

• We developed a simulation model supporting both radiocommunication and VLC.

• Using the simulation model, we simulated message ex-changes in a platooning system under RF jamming at-tacks, and present the proposed protocol works effectivelyto improve the end-to-end delay of the leader messagesto the members and the message delivery ratio.

The organization of the paper is as follows. In the nextsession, we show the related work on autonomous platooningand RF jamming attacks on vehicular wireless networks. Then,we show existing techniques to mitigate the effect of RFjamming attacks and methods to detect such attacks. In Section3, we show the system model of platooning and jammingattacks we treat in this paper. Then in Section 4, we proposea radio and VLC hybrid protocol to enable reliable andfast beacon message delivery under RF jamming attacks. In

Section 5, we present our simulation model that supports bothradio and VLC communications on vehicles, then evaluate theeffectiveness of the proposed method through comprehensivesimulation. Finally, we conclude the paper in Section 6.

II. RELATED WORK

Recently many car and IT industries are aggressively de-veloping autonomous driving technologies aiming for safedriving. For example, Google self-driving car [1] Project hasbeen conducting tests on public roads for several years. Nissanannounced that they will start to sell autonomous driving carin 2020 [13].

As well as a standalone autonomous driving, autonomousplatoon systems have been an active topic due to the chal-lenging problems it raises and the benefits it brings. It has thepotential to improve the traffic flow and to reduce fuel con-sumption, reducing jams on freeways and decreasing pollution[14].

So far, several projects have developed real platooning sys-tems. Sartre [3] lead by Volvo developed a platooning systemand demonstrated a platoon consisting of one truck and threecars with 4-meter gap between vehicles. They use 5.9GHzIEEE802.11p. Energy ITS [4] of Japan developed a platooningsystem that consists of four large trucks. They demonstrateda platoon that keeps 4m-gap at 80km/h moving speed. Theyuse 5.9GHz radio communication that used CSMA/CA basedMAC and infra-red inter-vehicular communication media. Themessage rate used in the system was 50Hz.

Since control of autonomous platoons depends on thereliability of messages exchanged between vehicles in theplatoon, development of wireless communication protocolsfor platooning control has been studied aggressively. Forinstance, Segata et al. have shown that a combined use ofa slotted scheduling mechanism and transmit power controlis highly beneficial in radio-based CACC systems throughrealistic vehicular network simulator [15] [16].

The effect of RF jamming attacks and the method todetect them have been investigated in [17] and [18]. Theirwork is based on simulation. On the other hand, Punalet al.tested three types of RF jammers, reactive, periodical, andconstant jammers in real vehicular network, and confirmedthat periodical jammer is effective even if the jammer’s SINRat the receiver (victim) is very weak.

Recently, research for using VLC in vehicular networks hasbeen active because it has the potential to realize wirelesscommunication that is robust to RF jamming attacks with lowcost devices that can be shared with illumination and visualsignal systems. For example, Yu et al. proposes a model for aVLC link based on measurement campaign of an experimentalVLC system built with off-the-shelf LED and photo diodedevices on a scooter [12]. Tomas et al. developed a simulationmodule of physical and MAC layers of VLC on vehicles forJiST/SWANS wireless network simulator [19].

To the best of our knowledge, this is the first paper thatevaluates the effectiveness of radio and multi-hop VLC hybrid


97

protocol for reliable platooning control under RF jammingattacks.

III. SYSTEM MODEL

A. Platoon Model

We adopt the platooning controller in [5]. The controller ofeach member of a platoon uses the speed and accelerationof the leaders and the preceding vehicle for its input. Weassume that only the leader’s information has to be sent viaa wireless channel because the front vehicle’s speed and ac-celeration can be measured by some on-board sensors such asmillimeter wave radars. We assume that each vehicle is awareof its position in the platoon. We assume the communicationprotocols for platooning are based on the IEEE 802.11p/IEEE1609.4 PHY/MAC. The scheduled messages will contend forthe channel in a CSMA/CA fashion. According to [6] and [15],we assume the information update frequency is 10Hz.

B. Jamming Attack Models

So far various jamming attacks on wireless LAN systemshave been considered. Punalet al. implemented the followingthree types of RF jammers, i) Reactive, ii) Periodical, andiii) Constant Jammers on a real radio devices and measuredthe effect on real vehicular networks that use IEEE802.11p-compatible communication devices. The reactive jammersends a jamming signal when it detects a radio signal of otherstations so that the CSMA/CA algorithm used in IEEE802.11pdoes not work well. The periodical jammer sends short jam-ming signal periodically with short interval. Fig. 2 presentsthe example of the jamming signal of the periodical jammer.The constant jammer sends a jamming signal constantly.

Their experiment results show that the periodical jammeris effective even if the SINR of the jammer is small becausethat the automatic gain controller (AGC) cannot quickly adaptto the change of incoming signal strength and it leads tooverflow of A/D converter if it has received a weak signal.To simulate this phenomenon, the wireless network simulatorhas a function that gives the delay of the change of AGC’sgain. In general, wireless network simulators do not have thefunction for simulating such a behavior. In our simulation, thisfunction is neglected.

In this paper, we focus on the effect of the periodical jammerbecause [10] presented the periodical jammer is effective evenif the SINR of the jammer at the receiver (victim) is small,and thus it can be a strong candidate that attackers will use.

Jam64us

Jam Jam Jam Jam …

Idle 10us-200us

Fig. 2. Periodic Jammer

TABLE IPARAMETERS OF PRELIMINARY SIMULATION

Parameter ValueTx power 20 dBmModulation QPSK R=1/2 (6Mbit/s)Propagation Model Free SpaceFading Model Nakagaimi (m = 2)PHY/MAC Model IEEE 802.11p/1609.4 single channelFrequency 5.89GHzantenna Omni-directional antenna (0dbi)MSDU size 308bytesAccess category AC VIMessage transmission rate 10Hz

CBR Broadcast (MAC Payload 308bytes / 10Hz)

J

S R

: 10m and 50m

Jammer

drds

dsr

Fig. 3. Preliminary Simulation Scenario

C. Effect of Periodic Jammer

For understanding the effect of the periodic jammer, weconducted preliminary simulations using a discrete event sim-ulator Scenargie, a product of Space Time Engineering, LLC[20]. In this simulation, we assumed that a legal sender, areceiver, and a jammer were placed as shown in Fig. 3. Thesender and receiver are IEEE802.11p compliant devices andhave omni-antennas. The detail of the simulation parametersare shown in Tbl. I

We changed the distances between the jammer and thesender, and one between the jammer and the receiver. Inaddition, we changed the idle length of the periodical jammermodel. The periodical jammer model used in their experiment,jamming signal consists of a series of a 64 micro secondOFDM signal and 10 micro second idle time. In our jammermodel used in the simulation, we changed the length of theidle duration from 10 micro seconds to 200 micro seconds,while we used a constant signal length 64 micro seconds. Thesender constantly sends packets with a 308 byte-MAC payloadevery 0.1 seconds.

Fig. 4 shows the results of the simulation. If the distancebetween the jammer and the receiver/sender is shorter than acertain distance, the packet reception ratio quickly becomeszero. There are two reasons for the low packet receptionratio. The first is that the sender cannot find the channel idlecondition if the jammer is close to the sender and the idlelength is short. As shown in the results of dsr = 10m case,even if the distance between the sender and the receiver isshort and the SINR of the sender’s signal at the receiver is


98

sufficiently high (the distance between the receiver and thejammer is longer than 20m), the packet reception ratio is zeroif the idle length of the jamming signal is shorter than 122micro seconds. The second reason is that the low SINR of thesender.

From the viewpoint of attackers, it is clear to use idle timeshorter than 122 microseconds. Thus we use 10 microsecondsidle time in our simulation for the rest of the paper.

IV. SECURING AUTONOMOUS PLATOONING SYSTEMSWITH HYBRID VEHICULAR COMMUNICATION

A. Problem FormulationAs stated above, if an autonomous platoon that depends

on only radio communication is under a jamming attack, thesafety and the stability of control of the platoon are posed athreat. The messages from the leader vehicle may be lost orthe delivery delay become longer than the requirement of theplatoon control system.

Though one hop communication from the leader vehicleto member vehicles is useful for shortening the end-to-enddelay of the message, if the SINR of a jamming signal ata member vehicle is sufficiently high compared with themessage signal from the leader, the member will fail toreceive the message from the leader. In addition, generally-used omni-directional antenna has a vulnerability of beingattacked from any directions, while it enables to transmit andreceive legitimate messages from/to any directions.

On the other hand, high-frequency electromagnetic wave in-cluding visible light inherently has high directivity. Especiallyvisible light communication (VLC) is useful because it canuse LED tail and head lamps for transmission device, and thussave the device cost. The communication between two VLCnodes has to be in a line of sight. Thus, if the transmitter andthe receiver of VLC signal are at the tail and head lamp ofa vehicle, it is difficult to send jamming signal so that thereceiver can receive the jamming signal.

In compensation for the resistance to RF jamming attacks,VLC using head and tail lamps is hard to use broadcast fordelivering the leader messages to the member vehicles in onehop because the vehicle bodies block the line of sight of theVLC. Therefore, message delivery using VLC from the leaderto the members has to be done in a multi-hop manner. Thisleads to long end-to-end delay especially when the VLC linkspeed is low.

The problem to be resolved by the proposed protocol is toshorten the end-to-end delay from the leader to the membersin a platoon and improve the message delivery ratio by usingboth radio and visible light communications when the radiocommunication is under a jamming attack.

B. Basic IdeaThe basic idea to shorten the message delivery delay under

RF jamming attacks is forwarding messages from the leaderby using both radio and VLC interfaces. The leader vehiclealways sends its message including its ID, position, velocity,acceleration, etc., periodically via both a radio and a VLC

interface. Even if a member vehicle fails to receive a messagefrom the leader vehicle sent via its radio interface, it will beable to receive the same message via a VLC interface if theleader has sent the message via VLC and upstream vehiclesforwards the message via VLC. If a vehicle receives a newbeacon message of the leader, it forwards the message viaboth radio and VLC message. If a member vehicle receives amessage with a timestamp older than the newest message thatthe vehicle has forwarded, it does not forward the receivedmessage.

Since the effective range of a single jammer is limited, byusing both radio and multi-hop VLC, we can improve the end-to-end message delivery delay of control messages from theleader vehicle as well as improving the message delivery ratio.

C. Restraining radio-based forwardingForwarding messages via radio interface may increase the

radio traffic even under RF jamming attacks. This may leadto further congestion of the radio link and decrease theradio message delivery ratio from the leader. For example,if ten vehicles other than a leader vehicle are in a platoonand the half of them have successfully received a beaconfrom the leader directly via radio interface, then five vehicleswill try to forward the message simultaneously. Under RFjamming attacks, the simultaneous attempts of transmissionfrom multiple vehicles will make the transmission more longer.

It can be reasonable to avoid the frequency that the membervehicles forward messages received from the radio interface.From this view point, we adopt the following strategies.

1) Forward via RF only if VLC is faster: If a platoonadopts Forward via RF only if VLC is faster strategy, onlyif a member vehicles receives a beacon message via its VLCinterface earlier than via its radio interface, it forwards themessage via both its radio and VLC interfaces. Otherwise, itforwards the message only via the VLC interface.

The condition that a beacon arrives via its VLC interfaceearlier than via the radio interface implies that the RF channelis congested and many vehicles have failed to receive RFmessages. Thus, the newer message should be forwarded assoon as possible. On the other hand, if a beacon message viaRF arrives faster than VLC, the radio channel is not severelycongested, the necessity to forward the message via RF is nothigh. Of course, sending a message via radio interface undera severe radio channel congestion would worsen the channelcondition.

2) Always forward via RF: If a platoon adopts Alwaysforward via RF strategy, every time a member vehicle receivesa new forwarded message either from its radio or VLCinterfaces, it immediately forwards the message via both itsradio and VLC interfaces.

V. EVALUATION

A. Simulation SetupTo evaluate the effectiveness of the proposed method,

we conducted simulations of communication for autonomousplatooning under jamming attack. We assumed each platoon


99

10us 120us 150us 200usIDLE = 122us

75m65m

15m75mColor = Packet Receive Ratio (Yellow = 100%)

ds [m]

d r [m

] 10m

50m

75m65m 65m

15m75m

dsr

dsr

10us 120us 150us 200usIDLE = 122us

160140120100

80604020

0

160140120100

80604020

0

100

80

60

40

2010

90

70

50

30

100

80

60

40

2010

90

70

50

30

65m 65m15m 15m

0 20 40 80 100 120 140 16060 0 20 40 80 100 120 140 16060 0 20 40 80 100 120 140 16060 0 20 40 80 100 120 140 16060 0 20 40 80 100 120 140 16060

Fig. 4. Effect of Periodic Jammer

Radio

LeaderVLC

Fig. 5. Proposed Method

TABLE IISIMULATION PARAMETERS

Parameter ValueVehicular size 5m × 2.5mNumber of platoons 1 and 4× 4Distance between the 5mcenters of neighboring lanesVehicle head-to-head distance 10mDistance between platoons 10mVehicular speed 17.0m/sBeacon frequency 10HzMAC Algorithm VLC interfaces ALOHA (no retransmission)Bit rate of VLC link 50kbpsSize of VLC packet 200bytesRF Tx power 20 dBmRF Modulation QPSK R=1/2 (6Mbit/s)RF Propagation Model Free SpaceRF Fading Model Nakagaimi (m = 2)RF PHY/MAC Model IEEE 802.11p/1609.4 single channelRF Frequency 5.89GHzRF antenna Omni-directional antenna (0dbi)RF MSDU size 200bytesRF Access category AC VI

consists of 10 vehicles and the head-to-head distance was10meters. We simulated a case where only 1 platoon exists ona road and a case where there are 4 lanes and 4 platoons areon each lane. Each vehicle has an IEEE802.11p radio interfaceand VLC transmitter and receiver. The VLC transmitter isattached at the center of the tail of the car, and the VLCreceiver is attached to the center of the head of the car. Tbl. IIsummarizes the simulation conditions.

To investigate the effectiveness of reducing the frequencyof forwarding beacon frames by member vehicles, we used

500m10m

10m

…

t=010 vehicles

10m

…

Periodic Jammer

…

5m

12.5m

17.0m/s

Fig. 6. Simulation Scenario

TABLE IIIVARIANTS OF THE PROPOSED METHOD

Model name OptionsUse VLC Use RF Forward via RF only

forwarding if VLC is fasterFFF False False FalseFTF False True FalseTFF True False FalseTTF True True FalseTTT True True True

variants of the proposed shown in Tbl. III.

B. Simulation ResultsFig. 7 and 8 show the time series of the end-to-end delay

from the leader to 5th and 10th car when there is only 1platoon on the road and when there are four platoons oneach of four lanes respectively. Fig. 9 shows the number ofmessages delivered to the 5th and 10th cars every one second.

In these graphs, each dot shows that the beacon messagewith the timestamp at the leader shown by the x-axis of thegraph has been received by the 5th or 10th car. When thebeacon messages are sent only via the RF interface of theleader vehicle (FFF, (a1), (b1)), we can see that messages arenot received after 26 seconds. The leader vehicle approachesclosest to the jammer at 29.4sec, and the 5th and the 10thvehicles approach there at 31.7sec and 34.7sec. We can seethat the beacon messages are not received by the vehiclesduring the time period centering the timing when they areat the closest point to the jammer.


100

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a1) 5th car False/False/False (b1) 10th car False/False/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a2) 5th car False/True/False (b2) 10th car False/True/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a3) 5th car True/False/False (b3) 10th car True/False/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a4) 5th car True/True/False (b4) 10th car True/True/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a5) 5th car True/True/True (b5) 10th car True/True/True

Fig. 7. Time series variation of end to end delay of leader messages(1 platoon)

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a1) 5th car False/False/False (b1) 10th car False/False/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a2) 5th car False/True/False (b2) 10th car False/True/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a3) 5th car True/False/False (b3) 10th car True/False/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a4) 5th car True/True/False (b4) 10th car True/True/False

0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec] 0

0.5

1

1.5

2

2.5

20 25 30 35 40 45 50

End-

to-e

nd d

elay

[sec

]

Time [sec]

(a5) 5th car True/True/True (b5) 10th car True/True/True

Fig. 8. Time series variation of end to end delay of leader messages (4× 4platoons)


101

0

2

4

6

8

10

20 25 30 35 40 45 50

Pack

et d

eliv

erie

d in

1 s

ec

Time [sec]

False/False/FalseFalse/True/FalseTrue/False/FalseTrue/True/FalseTrue/True/True

(a) 5th car / 1 platoon

0

2

4

6

8

10

20 25 30 35 40 45 50

Pack

et d

eliv

erie

d in

1 s

ec

Time [sec]


(b) 10th car / 1 platoon

0

2

4

6

8

10

20 25 30 35 40 45 50

Pack

et d

eliv

erie

d in

1 s

ec

Time [sec]


(c) 5th car / 4× 4 platoons

0

2

4

6

8

10

20 25 30 35 40 45 50

Pack

et d

eliv

erie

d in

1 s

ec

Time [sec]


(d) 10th car / 4 platoons

Fig. 9. Packet delivery ratio

By forwarding beacon messages using RF at each membervehicle (FTF, (a2) (b2)), the number of beacon messagesreceived by the members increases, but the end-to-end delayis long and the number of received messages during a fewseconds when the vehicle at the closest point of the jammer.If the leader vehicle cannot send any beacon message due tothe busy state of the RF channel caused by the jammer, thefollower vehicle cannot forward the beacon message, thus thepacket received by the member vehicles is still small.

By using only VLC for forwarding beacon message ofthe leader vehicle (TFF (a3), (b3)), the reachability of thebeacon messages improves. It can be confirmed clearly inFig. 9. However, the end-to-end delay is long for long duration,especially when the position of the vehicle in the platoon isbackward and the number of platoons is large.

Using both RF and VLC for forwarding beacon messages ofthe leader (TTF, (a4), (b4)), the duration of end-to-end delayis shortened. When using the forward via RF only if VLS isfaster strategy (TTT, (a5) ,(b5)), due to the smaller number ofRF forwarding messages caused by the strategy, the end-to-enddelay is longer than TTF case when the number of platoon is1. The idea behind the strategy is to prevent congestion causedby the messages forwarded by member vehicles under RFjamming attacks. Thus, the effect of the strategy is expected topresent when the number of vehicles/platoons is large. Whenthe number of the platoon is 16 (Fig. 8), the difference betweencases of TTF and TTT is very small. We cannot see neithernegative impact nor positive impact of the forward via RF onlyif VLS is faster strategy in this case.

Note that in the simulation of in this paper, we assumedvery slow VLC link (50kbps) that will be realized by today’soff-the-shelf LED and photo-diode devices. The delay factor ofthe system strongly is dependent on the link speed. To balancethe requirement of the end-to-end delay that the platooningalgorithm requires and the link speed and technology used inVLC should be carefully chosen.

VI. CONCLUSION

We proposed a radio and visible light hybrid protocolfor improving the reliability of control message used in au-tonomous platooning systems and evaluated the effectivenessof the strategies through simulations. The simulation resultsshow that by using both radio communication and multi-hop visible-light communication, the reliability of platooningcontrol messages from the leader vehicle can be significantlyimproved and the end-to-end delay can be shortened. Futurework will be more simulation with realistic scenarios thatinclude the real platoon mobility and precise VLC commu-nication model including the interference of VLC signals,especially in congested and/or curved lanes.

ACKNOWLEDGMENT

This work is supported by KAKENHI 15H02689.


102

REFERENCES

[1] Google Self-Driving Car Project, http://www.google.com/selfdrivingcar/(retrieved 2015/9/14)

[2] X. Lu, and S.E. Shladover, “Automated truck platoon control,” CaliforniaPATH Research Report, UCB-ITS-PRR-2011-13, June. 2011.

[3] E. Chan, P. Gilhead, P. Jelinek, P. Krejci, T. Robinson, “Cooperativecontrol of SARTRE automated platoon vehicles”, in ‘Proc. of the 19thITS World Congress, pp. 22–26, Oct. 2012.

[4] S. Tsugawa, An overview on an automated truck platoon within theEnergy ITS Project, in Proc. 7th IFAC Symposium on Advances inAutomotive Control, pp.41–46, 2013.

[5] R. Rajamani, H.-S. Tan, B.K. Law, and W.-B. Zhang, “Demonstration ofintegrated longitudinal and lateral control for the operation of automatedvehicles in platoons,” IEEE Trans. on Control Systems Tech., vol. 8, no. 4,pp. 695–708, July 2000.

[6] J. Ploeg, B. Scheepers, E. van Nunen, N.-van de Wouw, and H. Ni-jmeijer, “Design and experimental evaluation of cooperative adaptivecruise control,” in Proc. IEEE International Conference on IntelligentTransportation Systems (ITSC 2011), pp. 260–265, Oct. 2011.

[7] H.H. Meinel, “Evolving automotive radar — From the very beginningsinto the future,” in Proc. 8th European Conference on Antennas andPropagation (EuCAP) 2014, pp. 3107–3114, Apr. 2014.

[8] GNU Radio, http://gnuradio.org/[9] P. Murphy, A. Sabharwal, and B. Aazhang, “Design of WARP: A Flexible

Wireless Open-Access Research Platform,” in Proc. EUSIPCO 2006,2006.

[10] O. Punal, C. Pereira, A. Aguiar, and J. Gross, “Experimental Character-ization and Modeling of RF Jamming Attacks on VANETs,” IEEE Trans.Vehicular Technology, vol. 64, no. 2, pp. 524–540, Feb. 2015.

[11] IEEE 802.11p, Wireless Access in Vehicular Environments, Amendment6 to 802.11, Jul. 2010.

[12] S-H Yu, et al., “Smart automotive lighting for vehicle safety,” IEEECommunications Magazine., vol. 51, no. 12, pp.50–59, 2013.

[13] Nissan’s autonomous drive car, http://www.nissanusa.com/blog/autonomous-drive-car, Mar. 2014.

[14] B. van Arem, C. van Driel, and R. Visser, “The impact of cooperativeadaptive cruise control on trafic-flow characteristics,” IEEE Trans. onIntelligent Transportation Systems vol. , no. 4, pp. 429–436, Dec. 2006.,

[15] M. Segata, et al., “Towards Inter-Vehicle Communication Strategies forPlatooning Support,” Nets4Cars 2014-Fall, pp. 1–6, 2014.

[16] M. Segata, S. Joerer, B. Bloessl, C. Sommer, F. Dressler, and R.L. Cigno,“PLEXE: A Platooning Extension for Veins,” in Proc. of 6th IEEEVehicular Networking Conference (VNC 2014), pp. 53–60, Dec. 2014.

[17] A. Hamieh, J. Ben-Othman, and L. Mokdad, “Detection of radiointerference attacks in VANET,” in Proc. Int. IEEE Global Commun.Conf. 2009, pp.1–5, 2009.

[18] E. Schoch, F. Kargl, and T. Leinmuller, “Vulnerabilities of geocastmessage distribution,” in Proc. IEEE Globecom Workshops, 2007, pp. 1–8, Nov. 2007.

[19] B. Tomas, H.-M. Tsai, M. Boban, “Simulating Vehicular Visible LightCommunication: Physical Radio and MAC Modeling,” in Proc. of IEEEVehicular Networking Conference (VNC), pp. 228–231, Dec. 2014.

[20] Space Time Engineering, http://www.spacetime-eng.com/


103

improving reliability of platooning control messages using...

Documents