Dual-Band Channel Measurements for an Advanced Tyre Monitoring System

Transcription

1 Dual-Band Channel Measureents for an Advanced Tyre Monitoring Syste Gregor Lasser and Christoph F. Mecklenbräuker Vienna University of Technology Institute of Counications and Radio-Frequency Engineering Gusshausstrasse 5/89, 4 Vienna, Austria Eail: gregor.lasser@nt.tuwien.ac.at Christian Doppler Laboratory for Wireless Technologies for Sustainable Mobility Abstract Due to the legal situation in soe states and the beneficial effects on vehicular safety, fuel consuption and cofort issues, tyre pressure onitoring systes becae very popular in the recent years. Despite this situation, few easureent data about the propagation channel between the sensor units and the vehicle s reader antenna are available, especially for frequency bands above 4 MHz and for reader antennas ounted on the vehicle s botto plate. In this contribution we present static channel easureents for an advanced tyre onitoring syste ATMS in the 868 MHz and the.45 GHz bands, carried out with a vehicle equipped with a single dual-band receive antenna at the vehicle botto. ATM systes obtain sensor data directly in the tyre, so the sensor unit antennas were placed at the tyre sidewalls. I. INTRODUCTION Based on tyre pressure onitoring systes TPMS, which are in widespread coercial use due to legal situations and their positive ipact on vehicular cofort and safety standards, advanced tyre onitoring systes ATMS are currently under investigation [], []. Both, TPMS and ATMS rely on sensor equipped wheel units WUs, which use radio transissions to transfer easureent data. While the WU of conventional direct TPMS is ounted in the ri and often uses the valve s protruding part as antenna, the WU of an ATMS is placed directly in the tyre. This ounting enables to gain new sensor data like tyre tread vibration, vertical load and contact area. Additionally, radio frequency identification RFID functionality can be ipleented. A requireent for this ounting position is a light-weight WU, so it should be battery-less. To power the WU of an advanced tyre onitoring syste, the electroagnetic field of the centralised on-board unit OU could be used like in passive RFID systes. So the radio propagation channel for ATMS is not only iportant for data transfer, but should also be analysed for the purpose of WU powering. There are quite a few papers on siulation odels which capture the channel paraeters for conventional TPMS, ostly in the 5 MHz and 44 MHz bands, but inforation about This work has been funded by Infineon Technologies Austria AG and the Christian Doppler Laboratory for wireless technologies for sustainable obility. The financial support by the Federal Ministry of Econoy, Faily and Youth and the National Foundation for Research, Technology and Developent is gratefully acknowledged. actual easureent setups and their results is quite rare [], [4]. In this paper we present easureents carried out in the 865 to 868 MHz band and in the.45 GHz range, according to current RFID standards. The sensor antennas were directly attached to the inner sidewall of a odified standard tyre of a VW Golf V car. II. UNDERLYING CHANNEL MODEL Current TPMS, but also soe ATMS use a narrow-band uplink technology. For these systes the propagation channel between one OU and several WUs can be described with the following flat fading channel odel [5]: r t = 4 [ ] h i ϕt, Θt + hi t s i t+nt. i= Here r t denotes the received signal at the centralised OU which is the su of four products of the individual OU WU channels and the corresponding WU transit signal s i t, plus a receiver noise ter nt, that is assued to be white and gaussian. The channel coefficient is decoposed in two parts: the deterinistic channel coponent h i ϕt, Θt which captures the variation of the channel caused by changes of the rotational angle ϕ and the steering angle Θ, and the stochastic part h i t that introduces the channel variations caused by wheel and vehicle vibrations and the roughness of the road surface. To easure the deterinistic channel coponent a static easureent scenario can be used. Capturing both, the deterinistic and stochastic coponents, requires a dynaic channel easureent, were the test vehicle is travelling on a specific road. In this work, carried out with an industrial partner, we focus on the deterinistic channel coponent h i ϕ, Θ. The π periodicity in ϕ allows for a decoposition of the channel coponent h i ϕ, Θ in a Fourier series: with h i ϕ, Θ = c i, Θ = π = π c i, Θe jϕ, h i ϕ, Θe jϕ dϕ.

2 Textile sling VNA OU Antenna Ground Plane Plastic Binders OU WU Measureent Cables Plexiglass Sleeve Radiating Eleent Measureent Cable Fig.. Measureent Setup. III. MEASUREMENT SETUP The basic principle of the easureents perfored was to use a vector network analyser to capture the deterinistic channel coefficient for different rotational angles of the front right wheel. While the position of the OU antenna was kept fixed during the whole easureent capaign, the WU antennas were ounted at different positions at the tyre sidewall to evaluate the ipact of this ounting. Both, the OU and WU antenna were connected to the vector network analyser using coaxial cables, as indicated in Fig.. The use of cables in channel easureents is always delicate, because wave propagation can take place along these etallic objects, and the coaxial cables theselves radiate if unbalanced currents are present. To prevent these negative effects the easureent cables were equipped with ferrite beads which constitute a high ipedance for travelling waves. This property is guaranteed for the 865 MHz band where the agnitude of the coplex ipedance of a ferrite bead is at least Ω. For the easureents at the upper operating frequency the ferrite bead s ipedance is unspecified, but due to the longer electrical length between the antennas and the vector network analyser sufficient daping of the creeping waves is assued. The front right wheel was slightly lifted by pulling a textile sling fixed to the ri with a gantry crane. This ethod ensures a constant distance between wheel and wheel house and keeps the wheel s suspension in position, which is essential for achieving realistic easureent results. Further, it avoids any objects like jacks or support racks below the vehicle. A U shaped cut in the tyre enabled to swing out a sall part of the tyre sidewall and provided access for WU antenna ounting. A. OU Antenna For the OU a odified ground plane antenna was used, which had been ounted to the botto of the vehicle. To enable dual-band easureents without changing the OU antenna, it was designed to be in resonance at both frequency bands. At 868 MHz, the onopole acts as a quarterwavelength resonator. The different electric field distributions of the 868 MHz and.45 GHz band enable to selectively increase the electrical length of the radiating eleent for the higher frequency band. This is done by enclosing the radiating eleent s base section in a plexiglass sleeve. Due to the perittivity of the plexiglass, the antenna s three-quarterwavelength resonance is shifted to the.45 GHz band. The Fig.. Investigated Wheel Botto view of easureent vehicle on auto-hoist. antenna was fixed to the easureent vehicle using plastic binders, as can be seen in Fig.. B. WU Antennas Two different WU dipole antennas were used for the easureents, both anufactured fro adhesive copper foil. Dipoles were chosen because previous work based on [6] indicates their efficient ipleentation in this application. The first one is 4, the second one, both with a gap between the dipole eleents. The shorter dipole had been used for all ounting positions and for both frequency ranges where easureents were perfored. The saller size enables easureents in all orientations, but in all cases it is non-resonant. In the 868 MHz band coparison easureents were carried out with the long dipole, which is in resonance here. The dipoles were soldered to sall printed circuit boards wich carried baluns to adapt for the asyetric coaxial feed line. For easureents in the low band a wire-wound transission line transforer was used. The.45 GHz band was covered with a chip hybrid balun. IV. CALIBRATION AND DE-EMBEDDING The vector network analyser was calibrated using short, open, load, and through standards at the end of the easureent cables. The OU antenna exhibits an input return loss greater db, so no further processing is necessary to obtain accurate easureents. The situation is different at the WU antenna, which was directly soldered to a flexible lossy cable, which was in turn connected to the easureent cable. Additionally, the short WU dipole s input ipedance is badly atched to the feeding cable. In a real TPMS scenario, the antenna ipedance would be atched to the TPMS chip antenna port. The de-ebedding on the WU antenna side was done in two steps. First the effects of the lossy feeding cable where reoved in Matlab. The scattering-paraeters S-paraeters of the feeding cable were separately easured with a network

3 S /db Inner parallel Inner orthogonal Outer parallel Outer orthogonal Fig.. Magnitude of h ϕ, for the short dipole at 865 MHz Fig. 4. WU antenna input return loss before de-ebedding analyser and used to calculate the new scattering-paraeter atrix of the channel response without the cable influence. In a second step, a virtual lossless tuner was inserted in Matlab to copensate for the atching losses. The antenna ports were terinated in the network analyser, so the input reflection coefficient at the OU antenna side equals S, when S denotes the S-paraeter atrix of the channel. The tuner s S-paraeter atrix S T is given by S T = S S S S. 4 The tuner was adapted to each easured frequency point for the WU antenna being at the topost position. This tuner setting was then used for all easured rotational steps, to show the consequences of antenna de-tuning caused by proxiity effects. V. MEASUREMENT RESULTS The vehicle equipped with easureent antennas and odified wheel was placed in a large hall with concrete floor. The gantry crane was adjusted to lift the easureent wheel by a few illietres to enable tyre rotation by hand. The channel coefficient h was easured in steps of the rotational angle ϕ. A. Results at 865 MHz With the short dipole four ounting positions within the tyre were easured: Mounted at the sidewall facing the OU antenna inner position, or the opposite outer position, and oriented parallel or orthogonal to the tyre tread. Fig. presents the de-ebedded easureent results for the short dipole and those four ounting positions, where ϕ = corresponds to the WU antenna being at the upperost position. The channel coefficients exhibit fading with a dynaic range of 9 to 7 db. This is due to ulti path propagation in the a b Fig. 5. Magnitudes of spectral coponents c, for inner parallel a, inner orthogonal b oriented antennas at 865 MHz. rich scattering environent. Another indicator of ulti path propagation is the issing window around ϕ = 8 where the WU antenna leaves the wheelhouse. In Section IV we stated, that the badly atched short dipole ounted on the tyre is exposed to a changing environent and therefore these proxiity effects change the atching over rotational angle. This effect is the reason for the variation of the WU antenna input return loss plotted in Fig. 4. For ATMS the WU chip ipedance should be equal to the conjugated coplex WU antenna feeding ipedance. Since this value is not constant, a coproise value has to be used. In Fig. 5 the fourier doain introduced in is used to show the spectral liitation of the channel. The agnitudes of the fourier coponents for the two cases of inner ounted WU antennas are plotted in these diagras. Visual inspection indicates that the channel is bandliited and a representation with ±9 Fourier coefficients is possible with little error. Calculating the captured channel power when suing only over these 9 Fourier coefficients yields to power capturing ratios of 99.9% for the parallel oriented antenna and 99.6% for the orthogonal one

4 For the four different WU antenna orientations, the channel coefficient s phase over rotational angle is plotted in Fig. 6. While the agnitudes plotted in Fig. indicate ulti path propagation, the phase response is a strong evidence for a doinating single path or cluster coponent. Especially for the inner ounted WU oriented orthogonal to the tread, the phase curve can be divided in three segents: First, between a rotational angle ϕ of and 9 the phase response exhibits a alost linear course fro about π/ to π. The second segent ranges fro 9 to 7 and shows a phase rise to ore than π. In the third part the slope is negative again. These three segents correspond to the varying differential oveents between WU and OU antenna. In the first and third segent the WU OU distance increases with rising rotational angle, so the slope of the phase response is negative. In the second segent the situation is reversed. The alost triangular shape of the phase response can be used to calculate the resulting Doppler frequency. In the second segent the doppler angular frequency ω D = d argh dt is represented by the difference quotient v dπ, 5 where v denotes the vehicle speed and d the wheel s diaeter. This yields to a Doppler frequency of ω D = Δargh Δt = Δargh Δϕ f D = ω D π = Δargh v Δϕ dπ 5v dπ. 6 If we neglect the higher Doppler frequency coponents, and use the linearised phase response with a Doppler shift of f D in sections one and three and f D in section two we get a total Doppler Bandwidth B D = 5v dπ. This corresponds to the forula B D = Mv dπ based on the spectral representation of the channel given in [5], when we chose M = 5 for the last Fourier coponent to be included. Fig. 5b shows, that the fifth spectral coponent c,5 is the strongest, indeed, but higher frequency coponents and therefore higher Doppler coponents exist, also indicated by the deviation fro a pure triangular shape in Fig. 6. A coparison between the short antenna discussed before and a resonant λ/ dipole is shown in Fig. 7. This figure copares the agnitudes of the channel coefficient for parallel oriented WU antennas. Due to higher efficiency the resonant dipole in ost cases is superior to the shorter version. The shape of the curves is alost the sae, especially for the case of inner WU ounting. B. Results at.45 GHz At the upper frequency band all easureents were carried out with the short dipole. Fig. 8 shows the agnitude of the channel coefficient at.45 GHz. When copared to the results in the lower frequency band the average channel losses are about db higher. The channel shows ore fluctuations which is also expressed in the Fourier doain, as plotted in Fig. 9. To deterine the bandwidth liitation ore frequency points and corresponding finer angular steps are required. argh /rad 4π π π π π π π Fig. 7. Fig. 6. Channel phase response argh at 865 MHz. Inner parallel short Outer parallel short Inner parallel long Outer parallel long Magnitude of h ϕ, for the short and long dipoles at 865 MHz. As was stated earlier, the channel response was not only easured for different WU antenna orientations, but also for three steering angles Θ: and ±5. One full turn of the steering wheel to the left hand side atches the Θ= 5 easureent curve. As an exaple of the ipact of the steering angle, Fig. shows a coparison between the agnitudes of the channel coefficients for the three easured angles at.45 GHz for a inner, parallel ounted WU antenna. There are soe changes in aplitude and curve progression, but the overall behavior over ϕ reains the sae. For the sae antenna position, Fig. depicts a broadband easureent fro.4 to.6ghz. There is soe frequency selective fading, but the aount of deviation in the channel coefficient curve progression between the two frequencies is siilar to the deviation between different steering angles.

5 Steering angle Θ= Steering angle Θ= 5 Steering angle Θ= Fig. 8. Magnitude of h ϕ, for the short dipole at.45 GHz Fig.. Magnitude of h ϕ, Θ for the inner, parallel ounted short dipole at.45 GHz for different steering angles a b Fig. 9. Magnitudes of spectral coponents c, for inner parallel a, inner orthogonal b oriented antennas at.45 GHz. VI. CONCLUSION In this contribution we discuss static channel easureents between a single dual-band OU antenna, and two types of WU antennas ounted at four positions at the wheel of a easureent vehicle. To retrieve accurate channel easureent data, a special ounting ethod of the wheel under investigation, and de-ebedding of cable and antenna isatching losses are used. The easureent results for both frequency bands indicate ulti-path propagation with fading up to db. For three of four antenna orientations a doinating path coponent is identified in the channel s phase response, for which the resulting doinant Doppler frequency was derived. With the presented easureent results we deonstrated that the channel s agnitude, phase response and antenna isatching are highly influenced by WU antenna ounting. ACKNOWLEDGMENT The authors would like to thank Michael Urbanek, Stefan Firsching and Christian Bauer fro the Institute for Internal Cobustion Engines and Autootive Engineering at the Vienna University of Technology for their valuable support and providing an essential vehicular hardware work environent f / GHz ϕ /Deg Fig.. Magnitude of h ϕ,,f for the inner, parallel ounted short dipole over frequency f and rotational angle ϕ. REFERENCES [] S. C. Ergen, A. Sangiovanni-Vincentelli, X. Sun, R. Tebano, S. Alalusi, G. Audisio, and M. Sabatini, The tire as an intelligent sensor, Coputer- Aided Design of Integrated Circuits and Systes, IEEE Transactions on, vol. 8, pp , July 9. [] R. Matsuzaki and A. Todoroki, Intelligent tires based on easureent of tire deforation, Journal of Solid Mechanics and Material Engineering, vol., no., pp. 69 8, 8. [] H. J. Song, J. S.Colburn, H. P. Hsu, and R. W. Wiese, Developent of reduced order odel for odeling perforance of tire pressure onitoring syste, Vehicular Technology Conference, 6. VTC-6 Fall. IEEE 64th, pp. 5, Septeber 6. [4] M. Brzeska and G.-A. Chaka, RF odelling and characterization of a tyre pressure onitoring syste, The Second European Conference on Antennas and Propagation, 7. EuCAP, pp. 6, Nov 7. [5] G. Lasser and C. F. Mecklenbräuker, Channel odel for tyre pressure onitoring systes TPMS, in Proc. EuCAP The 4th European Conference on Antennas and Propagation, Barcelona, Spain, April. [6] J. Grosinger, L. W. Mayer, C. F. Mecklenbräuker, and A. L. Scholtz, Input ipedance easureent of a dipole antenna ounted on a car tire, in Proc. 9 International Syposiu on Antennas and Propagation, Bangkok, Thailand, October 9, pp