A high performance multisensor system for precise vehicle ground speed measurement.
With increasing traffic flow, the need for higher traffic efficiency and increased driver safety has become a prime concern. By measuring true ground speed with Doppler radar sensors, the capabilities of advanced vehicle control systems, that is, anti-slip control and anti-lock brake systems, can be enhanced in both automotive and railway applications. Furthermore, incremental measurements of the driven distance between fixed reference points are made possible. However, high accuracy and reliability are required, especially in railway applications.
For vehicle ground speed measurements, a microwave beam is emitted obliquely to the ground, as shown in Figure 1. A small part of the wave (wavelength [Lambda]) is scattered back into the sensor antenna by statistically distributed scattering objects within the detection area. The low frequency Doppler signal is obtained by mixing the transmitted microwave signal with the received signal. The Doppler signal frequency [f.sub.d] is proportional to the vehicle speed [v.sub.x] and the cosine of the radiation angle [Alpha] with respect to the horizontal plane. The accumulated phase increments of the Doppler signal are a measure of the distance traveled by the vehicle.
Principally, the Doppler signal is nondeterministic and nonstationary. The true ground speed information has to be extracted from stochastic and noisy Doppler signal data, resulting in limited accuracies at low vehicle speeds and during dynamic vehicle conditions, such as acceleration or tilt-angle variations. In addition, conventional vehicle Doppler sensors exhibit an impact on the estimated speed value from the backscattering properties of the ground. Despite these problems, sophisticated signal evaluation methods enable the true values for speed and displacement to be estimated within an error boundary of less than one percent.
A PROTOTYPE MULTISENSOR CONCEPT
From system experience, built up from long term measurements on German and French railways, it is evident that an advanced sensor concept is necessary to maintain high measurement accuracy and reliability when the system is operated under adverse operating conditions. A multisensor approach has been chosen in order to overcome the typical technical problems of conventional vehicle ground speed sensor systems.
Figure 2 shows a schematic of the prototype measurement setup used during system tests in railway environments. The sensor box, shown in Figure 3, contains three independent microwave front ends with different antenna radiation patterns. Front end 1 utilizes a Janus pater antenna with two symmetrical antenna beams with radiation angles at [Alpha] = 45 [degrees]. This antenna concept compensates systematic measuring errors occurring due to tilt-angle variations. Front ends 2 and 3 are equipped with traveling-wave pater antennas radiating a single antenna beam in a direction of [Alpha] = 15 [degrees] with respect to the horizontal plane. Additionally, the system concept incorporates an accelerometer a, which is intended to improve measurement accuracy and plausibility during rapid acceleration and braking processes.
For an on-line evaluation including the digital signal processor (DSP), the Doppler signals of all three front ends as well as the acceleration signal and the wheel speed transducer signal are recorded simultaneously on digital audio tape (DAT). The signal processing is based on a system model of the sensor configuration[7,8] taking into account a prior knowledge of the system properties and its dynamic behavior. Specific algorithms, based on autoregressive and autoregressive moving average signal models have produced high measurement accuracies. These algorithms provide real-time signal processing with significantly higher accuracy and reliability than conventional methods, and have been implemented on the DSP. Experimental investigations on merging the acceleration and Doppler data with a sophisticated system model incorporating autoregressive models that are combined with Kalman filters offered further improvements of the time response from rapid vehicle speed changes.
The signal processing unit also contains the power supply for the microwave front ends, several output interfaces and a light-emitting diode display indicating the estimated speed value and the system status. For evaluation of the absolute measurement accuracy of the sensor system, both the estimated speed obtained from the sensor front end and the reference speed measured by wheel motion can be monitored on a PC.
THE MICROWAVE FRONT END
A microwave front end, shown in Figure 4, has been developed with the ability to operate under the adverse weather conditions and strong interference typically found in automotive and railway environments. Technically, the new microwave front end's requirements are a highly stable and reliable microwave oscillator, a sensitive receiver within the intended detection area and inherent suppression of any signals from outside this detection area by use of spreading-spectrum coding.
Figure 5 shows a Doppler sensor schematic implemented in the 24 GHz microwave front end, which includes a dielectric resonator oscillator (DRO), Schottky diode receiver, phase shift-keying (PSK) modulator and a traveling-wave patch antenna. The compact microwave module is 60 x 80 x 25 mm and is completely built using low cost hybrid microstrip technology.
The 24 GHz microwave source is a highly stabilized fundamental-frequency DBO. Conventional DBOs use the fundamental mode T[E.sub.01[Delta]] of the dielectric resonator. However, the use of a higher order mode of the dielectric resonator has led to a new type of DRO high spectral purity (phase noise level equal to -95 dBc/Hz at 100 kHz) and good temperature stability (+10 ppm/K). The receiver is a simple beam-lead Schottky diode detector. The sensor antenna, realizing a pencil-beam antenna pattern, is a traveling-wave type.
In this novel concept, the DRO, Schottky diode receiver and sensor antenna establish the basic Doppler sensing function. The microwave phase shifter, located between the receiver and the antenna, is an advantageous new feature. This modulator offers PSK of the radar signal, which is the key to detecting the driving direction, suppressing flicker noise, obtaining range selectivity and avoiding cross-sensor interference. Figure 6 shows an x-y-plane representation of the complex Doppler signal.
DETECTION OF THE DRIVING DIRECTION
The determination of the driving direction is based on the evaluation of the complex Doppler signal vector [Mathematical Expression Omitted]. A time multiplex between the real and imaginary part of the Doppler vector is performed by PSK of the radar signal using a periodic modulation signal m(t) with a frequency [f.sub.m] much higher than the maximum Doppler signal frequency (that is, [f.sub.m] = 1 MHz). The intended function of a quadrature receiver is achieved when a phase shift of [absolute value of] 2 [center dot] [Delta] [Phi] = 90[degrees] is chosen, whereas the signal components [d.sub.1](t) and [d.sub.2](t) of the signal vector [Mathematical Expression Omitted] are obtained by demultiplexing the phase-modulated Doppler signal using a SPDT switch.
The accumulated phase of the complex Doppler signal is directly proportional to the distance traveled by the vehicle and thus characterizes the vehicle position. To a first approximation, each driven distance of [Delta]x = [Lambda]/2cos([Alpha]) corresponds to a full rotation of [Mathematical Expression Omitted] in a clockwise or counterclockwise direction, depending on the orientation of vehicular motion. Therefore, even small deviations in vehicle position in the order of a wavelength and the no-motion condition are detectable.
FLICKER NOISE SUPPRESSION
Conventionally, the sensitivity of a Doppler sensor is reduced considerably by low frequency flicker noise (1/f noise) exhibited by microwave oscillators. This reduced sensitivity is due to the flicker noise contributions [n.sub.A](t) contained in the Doppler signal, which result in limited accuracies at low vehicle speeds. This disadvantage has been overcome in this front end concept due to its noise-reduction capability. This PSK leads to an upconversion of the Doppler signal into sidebands of the microwave carrier at integral multiples of the modulation frequency [f.sub.m], whereas the oscillator noise remains in the baseband. High-pass filtering of the modulated detector signal will stop the 1/f noise and pass the Doppler signal such that a quasi-heterodyne reception is established. Consequently, the difference term [d.sub.DIF](t) = [d.sub.2](t) - [d.sub.1](t), obtained after demultiplexing the detector signal with an SPDT switch or a mixer, is a noise-reduced Doppler signal with a much higher signal-to-noise ratio. Figures 7 and 8 show the significant 1/f noise suppression attained with this concept.
The detection area of a conventional CW Doppler sensor is determined by the radiation pattern of the sensor antenna. In automotive and railway environments, the Doppler measurement may be distorted, for instance, due to cross-sensor interference or an unintended multipath transmission. To avoid this distortion, the Doppler signal evaluation must be focused on the antenna footprint by establishing a range gate.
A modified version of the microwave front end incorporating the range gate function has been developed by encoding the microwave signal with a spread-spectrum pseudo-random noise (PN) sequence. Spurious Doppler signals from outside the range gate will be inherently suppressed. In accordance with the matched filter theory, the received signal is correlated with a time-delayed replica of the transmitted code. The suppressed spurious Doppler signals are shown in Figure 9 and the schematic of the PN coded front end is shown in Figure 10.
The modulation port of the phase shifter used to encode the microwave signal is fed by a PN generator. A cross correlation of the received encoded Doppler signal and the reference code is achieved with an IF mixer working as a decoder. The measurement distance [R.sub.0] is determined by a delay line ([R.sub.0] = [c[Tau].sub.1]/2), where [[Tau].sub.1] is equal to the time delay and c is equal to the speed of light. The width of the range gate [Delta]R can be controlled by the bit rate of the PN signal ([Delta]R = c/[f.sub.bit]).
Experimental investigations proved a high sensitivity within the focused range combined with a suppression of Doppler signals from outside the range gate. Thus, a significantly higher measurement reliability can be achieved in future railway microwave sensor systems.
SENSOR SIGNAL EVALUATION
Prototype sensor systems have been tested in different railway environments. The data obtained from these long term measurements enabled continuous improvements of the overall system performance. Typical results were obtained during a test ride of approximately one hour. The estimated vehicle speed based on the evaluation of Doppler signals from front end 1 (steep radiation angle, [Alpha] = 45 [degrees]) is shown in Figure 11. The ratio of the estimated and reference speeds measured by the wheel speed transducer is shown in Figure 12. Although an overall measurement error of approximately one percent is achieved, the estimated speed is influenced by backscattering properties of the ground. In the first part of the test ride, changing sleeper (railroad ties) types (metal/wood) and partial snow coverage caused a fairly large standard deviation. For the rest of the track with wood sleepers, only the measurement accuracy is satisfactory. The arrow indicates a transition of metal to wood sleepers. The measurement on metal sleepers leads to a lower speed value, which is a systematic error caused by a shift of the center of gravity of the Doppler spectrum due to specular reflection.
To obtain reliable and precise measurement results, the speed estimation must be independent of the ground properties. Principally, a theoretical upper boundary of the Doppler frequency exists that corresponds to horizontal radiation ([Alpha] = 0 [degrees]). Typical Doppler spectra measured with steep and flat radiations ([Alpha] = 45 [degrees], 15 [degrees]) are compared in Figure 13. The frequency scale normalized to the upper boundary of the Doppler frequency corresponds to the horizontal radiation [Alpha] = 0[degrees]. Using a flat radiation reduces the width of the Doppler spectrum significantly.
Figure 14 shows the improvement obtained from a front end with flat radiation combined with specific evaluation algorithms. The narrow spectrum resulting from flat radiation is less sensitive to varying ground types than the broad spectra obtained for steep radiation. Therefore, appropriate signal evaluation algorithms extracting the speed information exclusively from the narrow, highly significant upper edge of the Doppler spectrum will provide precise speed estimations.
Altogether, different versions of the sensor system have been tested on several thousands of kilometers of railway track. The overall measurement precision is approximately one percent for the ground speed measurement and approximately 2 m/1000 m for the driven-distance measurement. The performance of the system was not influenced by rain or snow. In almost all situations where the accuracy obtained with front end 1 was reduced due to changing ground properties, front ends 2 and 3 (flat radiation angle) produced reliable measurement results. In test rides on the TGV French high speed train, the sensor system proved to work reliably at speeds up to 350 km/h. Moreover, the new concepts have been tested successfully for slippage control on the Euro-Sprinter high performance locomotive and the Metro in Prague.
A high performance sensor system for precise noncontact measurements of ground speed and driven distance has been reported. The data obtained from these long term measurements in various railway environments enabled continuous improvements of the overall system performance. The novel multisensor approach, based on advanced microwave front ends, an acceleration sensor and sophisticated digital signal processing, overcame typical accuracy limitations of conventional ground speed sensors.
1. V. Magor, "How Sensors Can Improve the Efficiency of Transportation Systems," Siemens Review, Spring 1993, pp. 26-30.
2. T.M. Hyltin, T.D. Fuchser, H.B. Tyson and W.R. Regueiro, "Vehicular Radar Speedometer," International Automotive Engineering Congress and Exposition, 1973, Detroit, MI.
3. N. Kees, M. Weinberger and J. Detlefsen, "Doppler Measurement of Lateral and Longitudiual Velocity for Automobiles at Millimeter-waves," 1993 IEEE MTT-S International Microwave Symposium, Atlanta, GA, pp. 805-808.
4. M. Joppich, R. Schubert and V. Magori, "Error Limits of Speed Measurement Using Onboard Doppler Radar Systems from a Theoretical and Practical Viewpoint," Proceedings SENSOR '95, Nurnberg, Germany, pp. 521-526.
5. R. Schubert, P. Heide and V. Magori, "Microwave Doppler Sensors Measuring Vehicle Speed and Traveled Distance: Realistic System Tests in Railroad Environment," Proceedings MIOP '95 Microwave & Optronics Conference, Stuttgart, Germany, pp. 365-369.
6. E. Heidrich, J. Kehrbeck and W. Wiesbeck, "Design Considerations for Comb-Line Microstrip Traveling-wave Antennas," 1992 IEEE International Symposium on Antennas and Propagation, Chicago, IL, pp. 1443-1446.
7. O. Besson and F. Castanie, "On Estimating the Frequency of a Sinusoid in Autoregressive Multiplicative Noise," Signal Processing 30, 1993, pp. 65-83.
8. M. Joppich, "Schatzverfahren zur Genauigkeitssteigerung der Geschwin-digkeitsmessung uber Grund nach dem Dopplerprinzip," PhD thesis, University d. Bundeswehr Munchen, Germany, VDI-Verlag, Fortschritt-Berichte, Reihe 8, Nr. 440, Dusseldorf, 1994.
9. P. Heide, V. Magori and R. Schwarte, "Coded 24 GHz Doppler Radar Sensors: A New Approach to High Precision Vehicle Position and Ground Speed Sensing in Railway and Automobile Applications," 1995 IEEE MTT-S International Microwave Symposium, Orlando, FL, pp. 965-968.
10. J. Kehrbeck, B. Zimmermann and W. Wiesbeck, "Mikrowellen-Nahbereichssensoren bei 24 GHz fur die Industrielle MeBtechnik," Proceedings MIOP '95 Microwave and Optronics Conference, Stuttgart, Germany, pp. 370-374.
11. P. Heide, R. Schubert, V. Magori and R. Schwarte, "24 GHz Low Cost Doppler Sensor with Fundamental Frequency GaAs Pseudomorphic HEMT Oscillator Stabilized by Dielectric Resonator Operating in Higher Order Mode," 1994 IEEE MTT-S International Microwave Symposium, San Diego, CA, pp. 945-948.
12. H.J. Siweris and B. Schick, "Analysis of Noise Upconversion in Microwave FET Oscillators," IEEE Transaction on Microwave Theory and Techniques, Vol. 33, No. 3, March 1985, pp. 233-242.
13. R.C. Dixon, Spread Spectrum Systems, John Wiley and Sons, New York, 1984.
14. P. Heide, "Coded Doppler Radar Sensors for Precise Measurements of Position and Speed: System Aspects and Implementation," PhD thesis, University Siegen, Germany, 1994.
15. J.W. Fischer and A. Schaefer-Enkeler, "Euro-Sprinter High Performance Electric Locomotive," eb-Elektrische Bahnen (91) 8, 1993; English version available from Siemens AG.
Patric Heide received his Dipl Ing and Dr Ing degrees in electrical engineering from the University of Siegen, Germany in 1991 and 1994, respectively. In 1991, he joined the Department of Sensor Systems at Siemens Corporate R&D labs in Munich, where he has developed ultrasonic and microwave sensors for railway and automotive applications. Currently, Heide is responsible for various research activities focused on new concepts and key technologies for future mm-wave systems, such as SAW-based methods, flip-chip technology and sensor-specific evaluation algorithms.
Richard Schubert received his Dipl Phys and Dr rer nat degrees in physics from the University of Bayreuth, Germany in 1987 and the Technical University, Munich in 1991, respectively. As an exchange student in Grenoble, France, he obtained the licence de physique diploma in 1984. From 1988 to 1991, Schubert developed a microwave reflectometer for the measurement of electron density profiles in nuclear fusion plasmas at the Max-Planck-Institute, Garching. In 1991, he joined the Department of Sensor Systems at Siemens Corporate R&D labs in Munich, where he works on the physical background, signal processing algorithms and the experimental evaluation of microwave sensors for industrial and traffic applications.
Valentin Magori received his Dipl Physics degree in 1968 from the Ludwig-Maximilian-University, Munich. From 1968 to 1970, he worked on neurophysical experiments at the Gesellschaft fur Strahlen- und Umweltforschung in Neuherberg. In 1970, he joined the Siemens Corporate R&D labs in Munich, where he built up substantial experience on ultrasonic systems and technology for more than 20 years. Since 1980, he has headed the Department of Sensor Systems and worked on ultrasonic sensors and sensor system techniques: Recently, Magori established a new research team working on innovative sensor principles and system concepts based on SAW devices and microwave technology.
Rudolf Schwarte received his Dipl Ing degree in communication engineering in 1965 and his Dr Ing degree in 1972 from the University of Aachen. He has been responsible for projects in the fields of extremely fast pulse techniques and gigabit electronics. Since 1982, he has led the Institut fur Nachrichtenverarbeitung at the University of Siegen, where industry-supported research on measurement techniques, signal processing, laser optical distance sensing and three-dimensional imaging is performed. In 1988, he established the Zentrum fur Sensorsysteme, a new interdisciplinary research center at the University of Siegen, where tight cooperation with industry, sensor systems for production automation, process and quality control, medicine and the prevention of environmental pollution are being developed. Currently, Schwarte is a full professor in electrical engineering and communication at the University of Siegen.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Doppler radar sensors|
|Author:||Heide, P.; Schubert, R.; Magori, V.; Schwarte, R.|
|Date:||Jul 1, 1996|
|Previous Article:||A high gain dual-polarized conical antenna.|
|Next Article:||Design and simulation of an inhomogeneous coupled-line bandpass filter.|