A Low Cost Pulse Doppler Radar Sensor for Short Range Commercial and Industrial Applications
By Kenneth V. Puglia, Principal, E X H Consulting Services
The prototype development and front-end architecture of a low cost, pulse Doppler radar (PDR) sensor is presented. The low cost follows from the utilization of a single RF source for transmit and receive functions in place of the traditional high spectral quality stable local oscillator (STALO) and coherent oscillator (COHO) signal generation. The PDR front-end architecture is specifically addressed with respect to efficient detection of the In-Phase (I) and Quadrature-Phase (Q) Doppler signals, as well as the limitations imposed by the utilization of a semi-coherent local oscillator in the detection process. The single RF source transmit and receive front-end architecture is comprehensibly examined and validated via laboratory measurement
Demodulation of the quadrature signal components is shown to be dependent upon synchronization of the transmitter frequency deviation control signal with the reference oscillator used for quadrature demodulation. The Doppler frequency error and pulse width are demonstrated to have a direct operational relationship to the difference between the transmitter deviation frequency and the quadrature demodulator reference source. A signal processing algorithm is explored and demonstrated to enhance signal detection and provide improved accuracy of range and range-rate measurements.
A number of commercial, government and military applications of the low cost PDR sensor are considered.
The confident detection of vehicles, personnel and other objects of interest provides significant performance advantage to many unrelated applications from vehicle range and closing rate measurement, to traffic monitoring, industrial security, environmental monitoring and robotics – just to name a few of the many solutions offered by microwave and millimeter wave radar sensors. Radar sensor detection systems with the properties of low cost, ease of physical deployment, positive detection – with extremely low probability of false alarm – and mission effectiveness would be of significant benefit to those, as well as many other applications. In addition, the ease of networking data from multiple radar sensor nodes provides the ability to implement process or other controls with the capacity to further enhance system effectiveness and to distribute information quickly and reliably to other related systems and processes.
Unlike infrared or laser sensors, microwave and millimeter wave radar sensors operate effectively under adverse weather conditions including, rain, snow, fog, smoke and particulate matter.
The existence of previously developed technology solutions – specifically automotive radar sensors – provides significant cost reduction in the system development phase and the recurring manufacturing cost category. High volume deployment of automotive radar sensors for collision avoidance, autonomous cruise control and pedestrian detection applications, have facilitated low cost radar sensor technology solutions. In fact, most major automobile manufacturers have integrated radar sensors among their safety and convenience product features. Automobile manufacturers demand high reliability, low cost equipment. These compatible parameters suggest serious consideration of modified or otherwise adapted automotive radar sensors for other numerous and unrelated applications.
Most low cost commercial and industrial radar sensors utilize a CW or FM waveform due to the architectural simplicity and modest bandwidth signal processing requirements; however, CW radar sensors have range measurement limitations and FM radar sensors impose strict frequency linearity and spectral requirements for accurate range and range resolution measurement. In addition, FM radar sensors typically require separate transmit and receive antenna apertures for proper operation; whereas pulse radar sensors may operate effectively using a single transmit and receive antenna via time duplex or sharing.
Within the following text and graphics, a simple, effective pulsed Doppler radar sensor architecture is explored that is suitable for many short range commercial and industrial applications. The pulsed Doppler radar architecture utilizes a single RF source for transmitter and local oscillator signal generation, thereby reducing the complexity and cost typically associated with STALO and COHO components, while retaining and exploiting the signal processing benefits of PDR, specifically, coherent pulse integration and clutter rejection for enhanced signal detection and low transmit power.
2.0 Pulse Doppler Radar Front-End Architecture
This PDR sensor front-end architecture is truly unique to low cost sensor applications.
Pulse Doppler techniques have generally been avoided in low cost sensor applications due to the perceived requirement of high spectral quality transmit and receive oscillators. Figure 2.1 illustrates the architecture of a low-cost, PDR sensor front-end. The unique feature of the PDR front-end is the phase continuous frequency switching of the RF VCO which provides the transmit signal frequency at ƒo + Dƒ, as well as the local oscillator signal frequency at ƒo for the first mixer.
During the transmit time interval, the RF VCO frequency is deviated by an amount equal to the reference oscillator frequency for tw seconds, where tw is the transmit pulse width. In order to demodulate the IF Doppler signal, the following equality must be enforced:
The frequency deviation pulse during the transmit interval is synchronous with the transmitter pulse repetition frequency which is derived from the reference oscillator. This condition assures that the phase relationship of the received IF signal is constant with respect to the reference oscillator for a stationary object; and that the Doppler phase shift from a moving object is efficiently detected. The RF VCO must be sufficiently stable over the two-way time of flight to the object such that effective demodulation of the Doppler frequency in the second converter stage is achieved. The second mixer operates as a quadrature demodulator. Such an arrangement enables true, pulsed Doppler radar implementation using a single RF source and a single IF reference oscillator for quadrature demodulation.
Doppler frequency detection in the second mixer is the result of the effective coherence of the reference oscillator signal with the IF output signal from the first mixer. Utilization of a narrow transmit pulse width assures that cycles of the heterodyned frequency difference between the IF at the output of the first mixer and the reference oscillator are not present at the video output; in addition, the frequency relationship between the signals is fixed to the extent of the equality of Equation (1). The requirement of stability of the received IF signal therefore translates to a fixed and accurate frequency deviation of the VCO by an amount equal to the second mixer local oscillator frequency (reference oscillator), and rapid settling to the transmit frequency and back to the receiver local oscillator frequency. Doppler detection capability is reduced if the transmit pulse width is arbitrarily increased to the extent that a partial cycle or cycles of the difference frequency become manifest in the recovered bipolar signals at the video output. An understanding of this operational principle is essential to proper deployment of the PDR front-end architecture.
To validate the unique feature of effective Doppler detection inherent in the low cost front-end PDR architecture, consider the block diagram of Figure 2.2 where the transmit and receive ports of the front-end are connected by a 26.5 meter length of coaxial cable with attendant time delay and insertion loss of 125 nanoseconds and 85 dB, respectively, at 24 GHz. The reference oscillator frequency is 160 MHz and the pulse repetition frequency is 1 MHz, i.e. 160 MHz divided by N, and N = 160.
The graphic of Figure 2.3 captures the quadrature signal components at the video outputs of the receiver front-end.
The following parameters are pertinent to the test execution and measurement data:
• Reference oscillator frequency:160 MHz
• Pulse repetition frequency:1.0 MHz
• Pulse width: 50 nanoseconds
• Input power at Rx: -80 dBm
• IF bandwidth: 40 MHz
• Video Bandwidth: 20 MHz
• Cable length: 26.5 meter
• Cable delay: 125 nanoseconds
• Cable loss: 85 dB at 24 GHz
To summarize: the principal attributes of the low cost PDR front-end architecture which enable coherent quadrature signal detection are:
1. Phase continuous frequency switching of the RF VCO
2. Synchronization of the frequency deviation pulse with the reference oscillator
3. Equality of the frequency deviation (Df) of the transmit VCO to the reference oscillator
4. Narrow transmission pulse commensurate with the equality of Equation 1
5. VCO stability and settling time commensurate with the detection range
3.0 Pulse Doppler Radar Sensor System Description
A block diagram of the prototype pulse Doppler radar sensor is illustrated in Figure 3.1, where the signal processing and PC control features have been added to the front-end assembly block diagram.
The quadrature phase signal components (I-Channel and Q-Channel video) from the PDR front-end are digitized using the 14-Bit AD7357 analog-to-digital converter. Signal sampling is executed synchronously by range bin using a variable time delay gate from the PRF signal.
Upon acquisition of N range bin data sets, a discreet Fourier Transform, or DFT, algorithm is executed for each range bin data set and the results are displayed graphically by Doppler frequency. Data acquisition may be accomplished via individual range bin or range bin sequence. An adjustable CFAR threshold (nominally, 15.5 dB signal-to-noise ratio) is utilized as object detection criterion.
The pulse width and duty cycle establish the range resolution and unambiguous range measurements in accordance with Equations (2) and (3), respectively, as defined in (1).
Additional operational parameters of the prototype low cost pulse Doppler sensor are summarized in Table I.
A photograph of the PDR front-end prototype is shown is Figure 3.2.
The prototype PDR sensor front end is constructed with discreet components which were characterized prior to assembly. Specifically, the RF source, which is a critical element of the front-end architecture as established earlier, utilized a post-coupled, low power Gunn device within a waveguide cavity.5 The FM characteristics exhibited greater than 300 MHz of electronic tuning and frequency settling time of less than 4 nsec.
4.0 Prototype Pulse Doppler Radar Sensor Test
The prototype PDR sensor was tested under controlled atmospheric conditions using the equipment configuration of Figure 4.1 where the PDR sensor is positioned relative to a test chamber containing RF absorbing material in the walls and a sideband modulator at the far end to simulate a moving object with variable velocity.
Testing of the PDR sensor may also be accomplished using a more accurate configuration which consists of detaching the transmit and receive antennas and connecting the Tx and Rx ports with coaxial cables and a sideband generator. With the latter technique, a precise measurement of input power enables an accurate determination of the receive sensitivity as well as the effectiveness of the processing algorithm under ideal conditions.
The effectivity of the coherent pulse integration algorithm was determined by introducing an input signal-to-noise ratio of near zero dB and processing a number of pulses from a single range bin via the DFT algorithm which effectively adds, or in radar parlance, coherently integrates the received signal energy from each pulse. It should be clear that the object must reside within the subject range bin for the entire observation interval. In some detection scenarios, particularly those where high speed object velocity is encountered, a moving object may not reside within a single range bin for the entire data acquisition interval. In those circumstances, the benefit of coherent pulse integration is degraded.
The graphic of Figure 4.2 is the result of the coherent integration of 1024 pulses from a single range bin using the DFT algorithm and an input signal-to-noise ratio of approximately zero dB. The simulated Doppler frequency is 25 KHz, which corresponds to a velocity of 156 meter/second (562 km/hour). An object at this velocity would be observed in a fixed range bin for less than 50 msec. The minimum data acquisition time is the product of the sample rate and number of samples, or 1.0 msec, at a sample rate of one sample per microsecond (the PRF).
The DFT algorithm achieves processing gain, i.e. improved signal-to-noise ratio by effectively reducing the detection bandwidth. Radar detection is a statistical process and is highly dependent upon the signal-to-noise ratio. The DFT filter bank bandwidth is the sample rate divided by the number of samples: 1.0 MHz/1024, or 976 Hz, which is also the Doppler frequency measurement resolution.
While the indicated signal-to-noise ratio results in the expected improvement of 30 dB (10*LogN), results from a real-world operating environment may not achieve this ideal value due to the statistical nature of object cross section, weather conditions, and propagation path variations.
The range bin data acquisition and DFT algorithm are basic elements of radar sensor development and serve only as initial signal processing tools to evaluate and quantify sensor operation. Signal processing and algorithm development are application specific, and typically require data acquisition sets for various anticipated scenarios under actual operating conditions. For example, consider a radar sensor for use in an autonomous cruise control (ACC) application where the lead vehicle is tracked and a fixed distance is to be maintained to the tracked vehicle. The algorithm and parameters for the simple scenario of an open road in the absence of road curvature, surface composition, terrain variations, guardrail and bridge clutter and on-coming traffic would represent the most elementary and least threatening operational scenario. When one considers the detection, classification, accounting and discrimination and decision tasks required for successful operation under the more complex scenarios, the algorithm designer’s task becomes significantly more complex and may require additional capability within the sensor front-end, e.g., increased dynamic range, antenna beam steering and a more complex transmit waveform to better define object measurement.
5.0 Pulse Doppler Radar Sensor Applications
As mentioned within the introduction, pulse Doppler sensors have not been extensively utilized in commercial, short range sensor applications due to the perceived requirements of high spectral quality, coherent RF sources. That perception has been demonstrated to be false.
Now, many short range sensor applications may benefit from simultaneous, high resolution measurement of object range and velocity. The additional benefits of coherent pulse integration, clutter suppression, singular and modest spectral source requirements and low transmit power are also notable.
The author enjoyed the privilege of membership on the Daimler + M/A-COM (now Autoliv Electronics) team that developed the first automotive autonomous cruise control (ACC) radar. The ACC radar system utilized a 76.5 GHz, pulse Doppler front-end architecture and serves as a model for advanced technological development and achievement of high volume manufacturing cost targets.
The model ARS200 pulse Doppler radar sensor system is illustrated in Figure 5.1.The development and production experience of the ARS200 sensor system may be leveraged to facilitate pulse Doppler radar sensors for a variety of commercial, industrial, government and military applications.
1. Marine radar sensors intended for navigation are an exception; however, although utilizing a pulse radar waveform, Doppler velocity measurement is not featured. See for example, Furuno marine radar at www.furunousa.com.
2. See patent number US5150126, 1992, Dornier GmbH, Friedrichshafen, Germany.
3. Separate transmit and receive antennas are illustrated; however, the PDR operates effectively with a single duplexed antenna. Optional transmit power amplifier and low noise receive amplifier are also shown.
4. Range resolution is degraded to a limited extent due to CFAR as explained in , p.299.
5. The Gunn VCO is a commercially available 24 GHz RF transceiver from M/A-COM Technology Solutions. See data sheet MACS-007802-0M1R1V. The Schottky detector diodes were removed to increase the output power.
6. Utilization of the Tx to Rx cable configuration, as previously described in section 2, eliminates background noise and clutter from the tested range bin as well as adjacent range bins from which additional reflection of transmit energy may enter the receiver. Although a more accurate characterization of sensor parameters is afforded, the technique does not fully and accurately predict sensor performance in a real operational environment where variations in propagation path, clutter and electromagnetic interference may seriously degrade sensor measurements.
A window or weighting function increases the resolution bandwidth of the DFT.
7.0 Additional Resources
1. Skolnik, M. I., Introduction to Radar Systems, 3rd ed., McGraw-Hill, New York, NY, 2001.
2. Stimson, G. E., Introduction to Airborne Radar, 2nd ed., SciTech Publishing, Mendham, NJ, 1998.
3. Mahafza, B. R., Introduction to Radar Analysis, CRC Press LLC, Boca Raton, FL, 1998.
4. Levanon, N., Radar Principles, John Wiley & Sons, Inc., New York, NY, 1988.
5. Richards, M. A., Fundamentals of Radar Signal Processing, McGraw-Hill, New York, NY 2005.
The author gratefully acknowledges the services of Microwave Support Systems in Nashua, NH for prototype assembly and test of the low cost Pulse Doppler Radar. The author is thankful for the many benefits of working with M/A-COM (now Autoliv Electronics) and Daimler AG during the development and manufacture of the 76.5 GHz automotive autonomous cruise control radar system – their product development vision and implementation are responsible for significant advances in automotive safety and convenience.
E X H Consulting Services
this article to a friend!