MICROWAVE RADIO COVERAGE FOR VEHICLE- TO-VEHICLE AND IN-VEHICLE COMMUNICATION

Transcription

1 MICROWAVE RADIO COVERAGE FOR VEHICLE TOVEHICLE AND INVEHICLE COMMUNICATION Marc Heddebaut 1, Jean Rioult 1, Marco Klingler 1, Atika Menhaj 2, Christophe Gransart 3 1 INRETSLEOST 20, rue Elisée Reclus Villeneuve d Ascq France 2 University of Valenciennes IEMNDOAE Le Mont Houy France 3 University of Lille LIFL Villeneuve d'ascq France marc.heddebaut@inrets.fr INTRODUCTION For a few years, autonomous intelligent cruise control (AICC) systems as well as anticollision radars have been developed and are currently sold by manufacturers [Grac 1999]. These systems are working at microwave frequencies. As an example, CEPT (European Post and Telecommunication Conference) has allocated frequency bands at 5.8 GHz for dedicated shortrange communication (DSRC), at 63 GHz for vehicletovehicle communication and at 76 GHz for anticollision sensors. DSRC and AICC equipment are readily available now. However, in the same time very few vehicletovehicle communication systems have been commercially developed and are currently on the bench. More recently new standards have emerged in the telecommunication industry. These standards develop an open global specification that enables mobile devices to access and interact with information and services instantly. These mobile devices are for example cellular phones, personal digital assistants, satellite positioning systems Among these emerging standards, Bluetooth is a candidate operating into the 2.45 GHz band [Blue 2001]. Local Area Network (LAN) working at 64 GHz is also a candidate, higher in frequency. From the Intelligent Transport System (ITS) point of view, invehicle communication as well as vehicletovehicle communication are a major concern. Thus, these emerging standards can probably offer an effective way of communicating inside the vehicle or between vehicles. In order to explore this opportunity, this paper analyzes the microwave radio coverage of these systems applied to invehicle communication and then to vehicletovehicle communication. In the first section, we describe simulation and experimental results of in car radio coverage using the Bluetooth radio frequency (RF) channel. The simulation, as well as broadband experimental measurements explore different transmitterreceiver locations inside a car. In the second part, vehicletovehicle communication is analyzed. The concept of the electronic preview mirror [Hedd 2000] is used to experiment a Bluetooth link in a vehicletovehicle communication scenario. This scenario is compared to the one using an AICC extended sensor. Because of the broad scope of this paper, only significant results can be presented among all the available ones. SIMULATING INCAR MICROWAVE PROPAGATION To determine electromagnetic scattering from complex, lossy/dielectric structures such as a car body, several simulation tools are currently available. Among them, WIPLD, FEKO and SEMCAD codes have been considered. 1

2 As described by their authors, WIPLD is based on the Method of Moments (MoM). It defines the geometry of a structure as any combination of wires, plates, material objects and provides information on that structure s current distribution, farfield radiation pattern, nearfield distribution, and multiport admittance, impedances, and/or sparameters at predefined feed points. FEKO is also based on the Method of Moments. Within FEKO The MoM, which scales poorly with frequency, has been hybridized with two asymptotic high frequency techniques namely: physical optics (PO) and uniform theory of diffraction (UTD). This hybridization enables the solution of much larger problems (in terms of wavelengths). SEMCAD uses a FiniteDifference TimeDomain (FDTD) kernel and focuses on main applications, namely nearfield analysis, antennas embedded in complex environments, EMC/EMI applications and dosimetry. The kernel enables a 3D EM fullwave FDTD formulation. Simulating accurately microwave radio frequency propagation in complex environments of several cubic meters like a car body requires either enormous computational memory resources (MoM Frequency Domain analysis) or a large computational time (FDTD Time Domain analysis). After some initial testing, it was found that the FDTD approach, although requiring large computational time, was the more realistic simulation approach. Anyway, if we consider that these solutions require defining a grid over the entire structure with a side dimension in the order of a tenth of wavelength (1.2 cm at 2.45 GHz), this leads to a fairly huge number of cells. Thus, we consider that simulating with conventional computers/workstations a car body at 2450 MHz is probably very close to the current simulation state of the art. Figure 1 below represents such a grid laid over the simulated car framework used for the study. For readability reasons, the grid step on this representation has been strongly oversized in comparison with the required simulation grid step. Figure 1 Grid over a simulated car (FDTD simulation) BLUETOOTH LINK SIMULATION RESULTS At 2.45 GHz, the simulation is performed using a quarter wave antenna radiating inside the car structure. This transmitting antenna is located either on the dashboard or in the rear boot. A metal sheet is also laid between the passenger cell and the motor compartment. A 10 cm diameter circular hole is perforated into this metal sheet. Otherwise, the passenger cell is empty. Windscreens are considered fully transparent to RF signals and the car body is simulated as a perfect conductor. Figures 2 and 3 show results respectively obtained using the 2.45 GHz radiating source (bright point) situated on the dashboard or in the boot. They are presented with an overall dynamic range representation of respectively 60 db and 40 db. 2

3 Figure 2 In car Efield repartition at 2.45 GHz (from the dashboard) Figure 3 In car Efield repartition at 2.45 GHz (from the rear boot) From the dashboard, a fairly good RF coverage is obtained over the whole structure including the imperfectly shielded motor compartment. An overall signal amplitude dynamic of 50 db is deduced from these results. Moving the receiving location a few centimeters apart, even in the vicinity of the transmitting antenna, leads to signal fluctuations in the order of 30 db. This result can be compared to the huge number of propagation modes that exist inside a large perfectly conducting cavity. On figure 3, transmitting from the boot yields to different results. The motor compartment radio coverage exhibits a supplementary attenuation due to the presence of the metal sheet separating the passenger cell from the motor compartment. The lower overall dynamic of the representation emphasizes the fluctuations of signal propagated inside the car body. BLUETOOTH INCAR EXPERIMENTAL RESULTS To validate these simulation results, measurements have been performed on board a real car. Wideband characterization is performed using a network analyzer between 1 GHz and 18 GHz associated to two wideband doubleridged identical horn antennas. Narrowband 2.3 to 2.6 GHz characterization is performed using 2.45 GHz halfwave dipoles antennas. A 3

4 calibration process is performed prior to the measurements in order to integrate all the losses and gains of the setup including the free space loss at the corresponding distance between the two antennas. This distance is set to 1.8 m. S 1,2 parameters are measured. Thus, on the following figures 0 db represents the reference level of free space loss attenuation. 1 0 S 21 (db) FREQUENCY Figure 4 1 to 18 GHz wideband measurements in horizontal polarization S 21 (db) FREQUENCY (GHz) Figure to 2.6 GHz narrowband measurements in horizontal polarization Additional attenuation due to the particular car environment is then effectively measured. Figure 4 results show a highly selective frequency radio frequency link. This was expected from the simulation results obtained in figures 2 and 3. Mean attenuation increases by 10 db over the 1.8 to 18 GHz decade. Moreover, figure 5 shows that on a relatively narrow bandwidth, selective attenuation is highly present. A frequency shift from 2.41 to 2.44 GHz yields to a 30 db fluctuation of the transmitted signal amplitude. Thus, moving mobile devices inside the car heavily modifies the communication properties of the radio channel. This can affect notably the effectiveness of the link. A quick qualitative validation has been performed using two portable computers equipped with PCMCIA Bluetooth cards. A video communication is established between the two PCs. One is set on the dashboard; the other one is moved inside the rear boot. Communication performance is analyzed. The next vehicle to vehicle communication paragraph will show that communication in open area is achievable up to 100 m. Computing this space loss attenuation at 2450 MHz according to 1 gives: 2 λ Att = 80 db (1) 2 ( 4 π d ) Since our simulation results show that an overall 50 db dynamic exists in the car body, communication should be maintained almost everywhere inside the car. 4

5 Bluetooth card Photo 1 Incar Bluetooth video transmission experimental setup We effectively verify that, while moving the second portable computer slowly in the rear boot or around the rear passenger seats, it is effectively possible to establish and maintain the communication with the dashboard computer. Photo 1 shows this equipment laid on the dashboard. The camera is looking ahead and the PCMCIA. A Bluetooth card is inserted in its slot on the left rear side of the computer. VEHICLE TO VEHICLE COMMUNICATION THE ELECTRONIC PREVIEW MIRROR APPLICATION Within a platoon, car and truck drivers use information about the speed and position of the preceding and following vehicles in order to elaborate and update a real time driving solution. AICC systems have been developed and are now currently sold by manufacturers in order to help drivers. Usually, these equipments only track the first preceding vehicle to deduce its speed and position. Nevertheless, this computed information remains on board the vehicle that has performed the measurement [Zhan 1999]. This is not broadcasted to the other platoon vehicles. Furthermore, within a platoon, the frontal road perception of the first vehicle is very particular and highly significant. Thus, it seems to us, that this information can be shared in realtime with the following vehicles within the platoon. This paper demonstrates the technical feasibility of transmitting front scene video information coming from a front vehicle to other following vehicles within a platoon. This concept has been named Electronic Millimetre Wave PreView Mirror (EPVM). Two RF links are considered. The first one uses a Bluetooth 2.45 GHz link, the second one uses a modified extended AICC sensor. This last concept is illustrated by the artist s view provided in figure 4. Using a passive subreflector, some of the millimetric (76 GHz) RF power available in the sensor is transmitted backwards behind the vehicle to following vehicles. Figure 4: AICC converted microwave link vehicletovehicle communication scenario 5

6 The huge available bandwidth (1 GHz) as well as the high data rate modulation capacity of the microwave sensor make it easy to use these sensors simultaneously for measurements of distance to targets and for communication. This particular extension of an AICC is described in [Hedd 2000]. Otherwise, Bluetooth incar equipments dialogue directly between vehicles, from inside the cars, through the different apertures (mainly the windscreens). COMPARISON OF RADIO FREQUENCY CHANNELS Thus, two main radio frequency paths are analyzed: through the windscreens (Bluetooth) and beneath the vehicle, between the underbody and the road surface (AICC). At 2.45 GHz, some results of internal to external coupling loss can already be deduced from figures 2 and 3. As seen in figure 4, beneath the vehicle is the path which can naturally be used by the millimetric wave signal partially redirected from the main AICC front radiation lobe. Characterization between 700 MHz and 18 GHz is performed similarly two the incar broadband analysis. Higher in frequency, between 5075 GHz external mixers are used and connected to Fresnel lens antennas. Figure 5 shows this setup. Network analyser Figure 5. Experimental radio frequency attenuation setup. Calibration is performed as previously for the incar environment. During this phase, the two antennas are separated by the overall length of the vehicle to be analyzed and, taking into account their directivity, placed sufficiently high, to be sure that ground reflection is minimal. The car or van to be tested is then inserted between the two antennas. Thus, we can be reasonably sure that the supplementary measured attenuation is due to the modification of the path in the presence of the vehicle. Figure 6 shows noticeable results obtained in horizontal linear polarization. Beneath a 3.5 m long car and at frequencies below 2 GHz, a waveguide cutoff propagation effect is obtained due to the presence of the two closely spaced ground surface and car under body parallel interfaces. This can be tentatively explained. Let us consider a conducting plane Σ containing yoz (car under body). We can only introduce a parallel plane Σ' to Σ without altering propagation if we meet the following limit conditions: E y = 0 ; E z = 0 and H x = 0 (2) Introducing these conditions into the general electric and magnetic propagation equations, it is obtained that propagation between Σ and Σ ' can only occur if the following condition (3) is observed: 0 < k. λ/2a < 1 (3) Expression where a is the distance between the two planes and k an integer. Thus, a cutoff frequency is obtained. Translated into our application where the distance between beneath the car and the ground surface is 18 cm, we obtain that frequencies under 850 MHz are heavily attenuated. Although a two plane perfect model is rather far from our real model, the experiment effectively shows this cutoff effect below 1 GHz. At higher frequencies, up to 6

7 18 GHz, since the measured attenuation is less than the quasifree space attenuation used in the calibration process, a low attenuation oversized waveguide propagation effect is observed. Through the front to back windscreens of the car structure, the attenuation is more or less constant between 700 MHz and 18 GHz and equals to 10 db. At 63 GHz, in horizontal polarization, measurements show a 0 db attenuation over the beneath the car RF path and 15 db through the windscreens db FREQUENCY (GHz) 1 Beneath the car 2 Through the windscreen Figure 6. Measured attenuation (HH). EPVM USING A MICROWAVE BENEATH THE CAR LINK To validate the EPVM concept, a microwave transmitter is used. This transmitter develops an output power of + 10 dbm which feeds a 17 dbi gain plated antenna. This antenna is situated below the front bumper of the car radiating between the underbody and the road surface towards the following platoon vehicles. This vehicle is the head of the platoon. A video camera visualizing the scene in front of the vehicle modulates this transmitter. A second car is equipped with a receiver fed by a similar mounted antenna but directed towards the head of the platoon (like a radar). Figure 7 Experimental implementation of the EPVM concept Figure 7 represents this setup. Using this setup, good quality video transmission is obtained up to 300 m using the beneath the car radio path. 7

8 EPVM IMPLEMENTATION USING A BLUETOOTH LINK An experimental setup has been built using two portable computers running Bluetooth through the available PCMCIA cards. One camera is installed on the dashboard of a front car simulating the head of a platoon. This camera is connected to the USB port of a portable computer laid on the back shelf. It sends data corresponding to the front scene perceived by this vehicle. A second computer is installed on the dashboard of a following vehicle visualizing the front scene perceived by the preceding vehicle. Experiments are performed, trying to establish and maintain a link between the computers while moving at various distances between the two vehicles. Up to m video communication ranges associated to 128x128 pixels definition at a rate of 56 images per second were consistently obtained. CONCLUSION The use of Bluetooth has been investigated in order to evaluate its potential for some ITS applications. Simulating and experimenting these systems for communicating inside a vehicle show that the propagation channel is harsh but that it is possible to maintain, for the chosen experimented locations a good radiocoverage inside the whole car. For vehicletovehicle communication, Bluetooth 1.1 standard available equipment seems also promising. Using available PCMCIAs cards, video transmissions up to 100 m have been achieved. Of course, the limited standard data rate means limited video resolution and frame rate in comparison to the use of the wide bandwidth provided by an AICC extended sensor. Bluetooth v 2.0, yet to be fully defined, announces a communication range greater than 100 m and a transmission rate of 10 Mbps thus suggesting a good candidate for vehicletovehicle communication. Finally, we really do need to indicate that, during the tests, actually seeing the scene in front of the preceding vehicle is a very strange situation for a human driver and that in the EPVM concept this raw video information in not intended to be directly presented to the drivers but preprocessed in order to extract significant parameters to be presented to the driver or used for example by the AICC signal processing unit. REFERENCES [Grac 1999] M. Grace, R. AbouJaoude, and K. A. Harvey Adaptive Cruise Control Radar Test System. 6 th World Congress on Intelligent Transport System Toronto Canada November Proceedings on the ITS America Official CDROM. [Blue 2001] The official Bluetooth web site. [Hedd 2000], M. Heddebaut, J. Rioult, M. Cuvelier, S. Ambellouis, M. Saint Venant, A. Rivenq Technical Evaluation of an Electronic Millimeter Wave PreView Mirror IEEE Vehicular Technology Conference (VTC2000/Fall) ISBN n IEEE paper Boston Sep. 2428, [Zhan 1999] Y. Zhang, E. B. Kostamopoulos, P. A. Ioannou and C. C. Chien Autonomous Intelligent Cruise Control Using Front and Back Information for Tight Vehicle Following Maneuvers, IEEE Transactions on Vehicular Technology Vol. 48 No. 1 January 1999 pp BLUETOOTH, WIPLD, FEKO and SEMCAD are registered trademarks. 8