Radarbook Graphical User Interface (RBK-GUI User Manual)

Radarbook Graphical User Interface (RBK-GUI User Manual) Inras GmbH Altenbergerstraße 69 4040 Linz, Austria Email: office@inras.at Phone: +43 732 2468 6384 Linz, July 2015

Contents 1 Document Version 2 2 Radarbook Graphical User Interface 3 2.1 Upcoming Features..................................... 3 2.2 Accessing the Radarbook................................. 4 3 Doppler Mode 5 3.1 Doppler Processing for a CW Radar........................... 5 3.2 Display IF Signals..................................... 5 3.3 Display Moving Cars.................................... 6 4 FMCW Mode 8 4.1 FMCW Processing..................................... 8 4.2 Display IF Signals in FMCW Mode............................ 9 4.3 Display Moving Persons.................................. 10 4.4 Display Moving Cars.................................... 10 5 Range-Doppler Mode 13 5.1 Range-Doppler Processing................................. 13 5.2 Display Range-Doppler Map................................ 14 6 Digital Beamforming Mode 15 6.1 DBF Processing....................................... 15 6.2 Display Spatial Distribution of Targets in DBF Mode................. 15 7 DBF-MTI Mode 17 7.1 DBF-MTI Processing.................................... 17 7.2 Display Spatial Distribution of Moving Targets in DBF-MTI Mode.......... 17 8 Radarbook Sampling Chain 18 9 List of Abbreviations 21 1 Document Version Version Description Date Author 1.0.0 Initial Version 2015-07-01 Andreas Haderer (AnHa) Inras GmbH 2 AnHa

2 Radarbook Graphical User Interface The software framework is written in Python and can be used to operated different RF frontends (24- and 77-GHz) plugged to the Radarbook from an easy to use tool. The Radarbook can be accessed either via a TCP/IP connection or over an USB 3.0 interface. The software tool provides an easy to use interface without the need to configure all the registers of the synthesizers and the RF transceivers. Hence, radar measurements can be performed within minutes and the implemented processing modes offer the possibility to evaluate the performance of radar sensors in customer specific applications. The user interface is designed to hide the details of the configuration of the transceiver chips and instead the user can focus on the configuration of the transmit waveforms with a minimum set of parameters. The current version of the GUI enables the following measurement modes: ˆ Doppler (CW), ˆ Frequency-Modulated Continuous Wave (FMCW), ˆ Range-Doppler (RD), ˆ Digital Beamforming (DBF), and ˆ DBF with moving target indication (MTI). In the subsequent sections the different modes and the configuration of the transmit waveforms are described. Moreover a few examples measured with a 24-GHz frontend with RF chips from Analog Devices are shown. All modes can be configured independently adjusting the radar system to the application at hand. 2.1 Upcoming Features The current version of the framework supports five measurement modes including various graphs for visualization of the radar signals and the processing results. In the next version of the framework the following new features will be supported: ˆ Measurement data logging with Matlab file export, ˆ Calibration mode to calibrate the frontends for DBF with multiple transmit and receive antennas, ˆ Automatic board finder for the TCP/IP based interface, and ˆ Smart commands for the online configuration of the transmit path and the registers of the transceiver chips. Inras GmbH 3 AnHa

2.2 Accessing the Radarbook After physically connecting the Radarbook to the host computer, the connection to the hardware over the TCP/IP interface must be configured. For this reason the IP address of the Radarbook must be specified as shown in Fig. 1. The dialog for modifing the connection settings is implemented in the System:Board Config menu. Figure 1: Specifying the TCP/IP connection in the System:Board Config menu. The IP address of the board can be altered with the integrated webserver. The default address of the board is set to 192.168.1.1. If the IP address of the RBK has been altered, then the new address must be specified in the user interface. In the subsequent sections the different operational modes are explained. Inras GmbH 4 AnHa

3 Doppler Mode The simplest mode of operation is the Doppler mode, where a Continuous-Wave (CW) radar with constant frequency is used. The Doppler processing implemented in the Radartool includes the following functionalities: ˆ Display IF signals, ˆ Calculate Doppler spectrum, ˆ Display time-dependent Doppler spectrum with adjustable color map, and ˆ Configure parameters of the CW radar system and data visualization. 3.1 Doppler Processing for a CW Radar In the Doppler mode the transmit frequency is held constant. Therefore, the transmit signal s t (t) = A t cos (2πf c t + ϕ 0 ) (1) can be modeled as cosine function with constant frequency f c. In (1), A t denotes the amplitude and ϕ 0 is the initial phase of the transmit signal. In case of a CW radar, the received signal is down-converted with a part of the transmit signal and therefore the IF signal for a single ideal point target s IF (t) A IF cos (2πf D (t)t + ϕ IF ) (2) is approximated with a cosine signal with amplitude A IF, frequency f D (t), and phase ϕ IF. The frequency f D (t) = 2f c v r (t) c 0 (3) is commonly referred to as Doppler frequency and is proportional to the radial velocity v r (t) of the target imaged by the radar. In (3), c 0 denotes the velocity of the electromagnetic wave, which is approximated by the speed of light. In the Doppler mode moving targets can be detected and the velocity of the targets can be estimated. 3.2 Display IF Signals To configure the radar system in Doppler mode the menu Modes:Doppler must be selected or instead the shortcut Ctrl+1 can be used. The Radartool offers the possibilty to display the sampled IF signals for all enabled IF channels as shown in Fig. 2. The IF signals can be enabled or disabled in the Configuration menu. In the Configuration tab the center frequency f c can be altered, too. In addition, the page enables the configuration of the AD8283 analog frontend implemented on the Radarbook. A detailed description of the sampling chain is given in Sec. 8. By altering the Clock divider and the CIC reduction the sampling frequency f s can be adjusted to the application at hand. The calculation of the sampling frequency f s is described also in Sec. 8 in more detail. By clicking the Initialize button the configuration is programmed to the Radarbook and the RF frontend. After finishing the initialization process measurements can be started. If parameters on the configuration page are Inras GmbH 5 AnHa

Figure 2: IF signals in Doppler mode. changed, the initialization must be run again to get the new parameters active. If the Radar is operated in Doppler mode, only moving targets are recognized. The instantaneous frequency of the received IF signals is proportional to the radial velocity v r (t) of the target. To explore the frequency content of the measured IF signals the FFT processing can be selected by clicking the check box. In this case the spectrum is plotted instead of the time signals as shown in Fig. 3. For generating the displayed snapshot a person was moving in front of the radar system. Therefore, a broadened spectrum is observable as multiple reflection centers with different radial velocities are present. 3.3 Display Moving Cars In Doppler mode the hardware complexity can be reduced because the required sampling rates are in the khz range. On the other hand only targets with different velocities can be distinguished and the system has no capabilities to resolve objects regarding their distance. In Fig. 4 the signatures of moving cars are displayed. For conducting the measurements the radar system was mounted next to a road in and distance of approximately 7 m and pointing to the road in an azimuth angle of 45. For conducting the measurements a fixed measurement interval of 75 ms was used. The signatures of the moving cars are clearly visible in the velocity-time map and simple signal processing algorithms can be used to count the number of passing cars. In the shown snapshot the signatures of five objects are visible. All objects were moving in the same direction. The shape of the signatures depends on the orientation of the radar system with respect to the road and the velocity of the targets. Inras GmbH 6 AnHa

Figure 3: Spectrum of the IF signals in Doppler mode. Figure 4: Velocity-time plot of five cars moving in the same direction alongside the radar. Inras GmbH 7 AnHa

4 FMCW Mode In FMCW mode the distance to the targets imaged by the radar system can be visualized. For this reason, a linear frequency modulation is used. The FMCW processing implemented in the framework includes: ˆ Display IF signals, ˆ Calculate range profiles, ˆ Display time-dependent range profile with adjustable colormap, and ˆ Configure FMCW radar system and data visualization. 4.1 FMCW Processing In FMCW mode the transmit frequency is modulated with a linearly increasing frequency ramp. Therefore, the transmit signal ( ) t s t (t) = A t rect cos ( 2πf c t + πk f t 2 ) + ϕ 0 (4) T can be modeled as cosine function with center frequency frequency f c and a quatratic term πk f t 2. In (4), A t denotes the amplitude and ϕ 0 is the initial phase of the tranmsit signal. The time duration of the chirp T is limeted with the rect function. The slope of the frequency rate is commonly referred to as chirp rate k f = f Stop f Start = B T T, (5) and it is defined by the chirp duration T and the bandwith B = f Stop f Start with f Stop beeing the stop frequency and f Start denoting the start frequency of the radar system. The range resolution in FMCW mode R = c 0 (6) 2B is a measure for the ability of the radar system to seperate two targets according to their radial distance. The range resolution is indirect proportional to the utilized bandwidth. The received signal is down-converted with a part of the transmit signal and therefore the IF signal for a single stationary ideal point target ( ) t s IF (t) A IF rect cos (2πf IF t + ϕ IF ) (7) T can be modelled with a cosine signal with amplitude A IF, frequrency f IF, and phase ϕ IF. The IF frequency f IF = k f τ = k f 2R c 0 (8) is proportional to the round-trip delay time τ and therefore to the distance R between the target and the radar system. The distance can be evaluated by determining the frequency of the IF signal. Inras GmbH 8 AnHa

In general a spectral representation is used to search for dominant reflections, which can be mapped to different targets. The phase ϕ IF can be used to estimate the distance only within the amibuity of the wavelength, but in comparison to the phases from multiple reveive antennas for the angle of incidence. 4.2 Display IF Signals in FMCW Mode To configure the radar system in FMCW mode the menu Modes:FMCW must be selected or the shortcut Ctrl+2 can be used. The framework offers the possibilty to display the sampled IF signals for all enabled IF channels as shown in Fig. 5. As no high-pass filter is implemented in the IF chain Figure 5: IF signals in FMCW mode. of the radar system, a strong low-frequency component is visible in the IF signals. This component is due to the unwanted coupling between the receive and transmit antennas. In the Configuration tab of the FMCW mode, the start frequency f 0, the stop frequency f 1, and the chirp duration T can be adjusted for the application at hand. Additionally, the sampling frequency can be set according to the requirements. The configuration of the sampling frequency is explained in Sec. 8. If the Radar is operated in FMCW mode, the distance to stationary targets can be estimated. The frequency of the received IF signal for a single target is proportional to the its distance to the radar. To explore the frequency content of the IF signals the FFT processing can be selected. In this case the magnitude spectrum is plotted versus the range as shown in Fig. 6. It is also possible to display the average spectrum by selecting the checkbox. In the FMCW mode it is also possible to image moving scenarios. For this reason a range-time image can be visualized. Inras GmbH 9 AnHa

Figure 6: Spectrum of the IF signals in FMCW mode. 4.3 Display Moving Persons If the Range-Time tab is selected, then the history of the range profiles is shown. In this mode it is best to use a fixed measurement repetition rate to ensure a uniform timing. To generate the image the number of stored range profiles and the displayed range interval can be configured. In Fig. 7 the signature of a person moving in front of the radar is shown. The measurements were taken indoor and therefore strong background reflections are observable too. 4.4 Display Moving Cars In contrast to the Doppler processing the FMCW operational mode offers the possibility to resolve targets according to their distance to the radar system. For conducting the measurements the radar system was mounted next to a road in and distance of approximately 7 m and pointing to the road in an azimuth angle of 45. A fixed measurement interval of 50 ms was used. If no visualization is used, then the measurement interval can be reduced below 1 ms increasing the number of measurements. In Fig. 8 the range-time plots for moving cars are shown. The signatures are embedded in clutter from the surrounding as no MTI processing is used. In Fig. 9 the range profile for a single measurement without a car present on the road is shown. Although no car is present, there are strong reflections in the displayed range interval. These components clearly exceed the system s noise floor, which is approximately at 100 dbv for the applied data scaling. Inras GmbH 10 AnHa

Figure 7: Range profile over time for a person walking in front of the radar system. Figure 8: Range-time plot for three cars passing the radar. Inras GmbH 11 AnHa

Figure 9: Range profile for a single measurement without a car on the road. Inras GmbH 12 AnHa

5 Range-Doppler Mode In range-doppler mode the radial distribution of the targets as well as their instantaneous velocities can be analyzed. This mode combines the advantages of FMCW and Doppler processing, which have been described in the previous sections. The range-doppler mode is one of the most powerful modes of operation but it requires a high-performance platform for processing the measured data in real-time. To enable range-doppler processing multiple subsequent upchirps are processed simultaneously by evaluating a two-dimensional Fourier transform. The range-doppler processing implemented in the framework includes the following functionalities: ˆ Display range-doppler map, and ˆ Configure range-doppler processing and data visualization. 5.1 Range-Doppler Processing In the range-doppler mode the transmit signal consists of multiple upchirps. The instantaneous frequency of the transmit waveform is shown in Fig. 10. Figure 10: Transmit waveform in range-doppler mode. The mathematical description for a single chirp has already been stated in Sec. 4.1. In range- Doppler mode N adjacent frames with a chirp repetition interval T p are used for processing. As the phase information between subsequent chirps is used to extract the velocity information, a precise timing relation is required. For this reason the measurement timing is generated with the programmable timing unit in the FPGA. This unit enables to specify the timing in clock cycles of the ADC without any additional timing jitter. The timing unit enables the specification of arbitrary uniform and non-uniform timing patterns. To configure the timing in range-doppler mode the chirp duration T (Duration) and the chirp repetition T p (Period) can be specified in the configuration page of the user interface. A two-dimensional FFT is used to calculate the range-doppler map and the magnitude of the map is displayed in the measurement view of the software framework. Before calculating the range- Doppler map the stationary targets are subtracted from the IF signals in order to remove the clutter. In addition, only the first receive channel is used for generating the range-doppler map and the residual receive channels are omitted. By evaluating the residual channels also the angle to the targets can be estimated. The Radarbook platform supports the measurement of all receive channels in the range-doppler mode but as a three-dimensional plot would be required only the results of the first channel are displayed. Inras GmbH 13 AnHa

5.2 Display Range-Doppler Map To configure the radar system in range-doppler mode the menu Menu:Range-Doppler must be selected or the shortcut Ctrl+3 can be used. The Radartool offers the possibilty to display the range-doppler map for a single receive channel as shown in Fig. 11. The snapshot was generated Figure 11: range-doppler map for a moving person in front of the radar. for a person swinging its arms in front of the radar. Therefore, dominant reflections can be observed in a distance of 2.5 m. As the body and the hands move with different speeds the reflections are spread over an extended interval of velocities. The range-doppler processing is one of the most powerful processing modes as it allows to separate targets with different velocitíes even if they are located at the same distance. This is especially useful to resolve complicated traffic scenarios with cars moving in opposite directions or during overtaking manouvers. Inras GmbH 14 AnHa

6 Digital Beamforming Mode In the digital beamforming (DBF) mode the distance to the targets and the angle are visualized. For this reason the receive signals from the eight spatially separated receive antennas are used to estimate the reflectivity of the imaging scene. The framework includes the following functionalities: ˆ Display spatial distribution of targets in the xy-plane, and ˆ Configure FMCW radar system and image visualization. 6.1 DBF Processing In the DBF mode the radar system is configured similar to the FMCW mode, which has been explained in Sec. 4.1. The only difference resides in the processing of the sampled IF signals. After calculating the range profiles the angle of incidence is calculated by evaluating the phase differences between the receive channels. This task can be performed with either an additional FFT or a back projection algorithm (BPA). In the software framework a BPA is used to evaluate the reflectivity of the imaging scene. For this reason the imaging scene is divided into equally spaced pixels and in the next step the reflectivity of the pixels is evaluated. High values for the reflectivity indicate the presence of targets and the maximum position can be used to estimate the angle of incidence. In DBF mode a calibration of the radar frontend is required to eliminate unwanted deterministic phase variations between the receive channels. The frontends are assembled with an EEPROM, which is used to store the calibration data. After initialization the calibration data is loaded from the EEPROM and the sampled IF signals are corrected before evaluating the reflectivity of the imaging scene. 6.2 Display Spatial Distribution of Targets in DBF Mode To configure the radar system in DBF mode the menu Modes:DBF must be selected or the shortcut Ctrl+4 can be used. In the measurement view the reflectivity of the illuminated scene is displayed. In Fig. 12 the reflectivity for an indoor measurement is shown. The radar system is pointing towards the corner of the room. In the reflecitivity map strong reflections from the walls building the corner can be observed. Inras GmbH 15 AnHa

Figure 12: Revlectivity map of the indoor target scene in DBF mode. Inras GmbH 16 AnHa

7 DBF-MTI Mode In the DBF moving target indication (MTI) mode the distance to the targets and the angle are visualized for moving targets only. For this reason N adjacent measurements for all receive signals are used to implement the moving target indication. The framework includes the following functionalities: ˆ Display spatial distribution of targets in the xy-plane, and ˆ Configure FMCW radar system and image visualization. 7.1 DBF-MTI Processing In the DBF-MTI mode the radar system is configured similar to the range-doppler mode, which has been explained in Sec. 5.1. Multiple adjacent chirps are processed simultaneously. The N chirps are used to estimate the signature of the non-moving targets, which is subtracted before evaluating the reflectivity of the imaging scene. 7.2 Display Spatial Distribution of Moving Targets in DBF-MTI Mode To configure the radar system in DBF-MTI mode the menu Modes:DBF-MTI must be selected or the shortcut Ctrl+5 can be used. The framework displays the reflectivity of the moving targets in the illuminated scene. In Fig. 13 the reflectivity for an indoor measurement is shown. The non-moving background is not visible in the measurement view. Figure 13: Reflectivity map of the moving targets. Inras GmbH 17 AnHa

8 Radarbook Sampling Chain The analog frontend of the Radarbook is built with two AD8283 ADCs from Analog Devices. In this section the configuration of the sampling chain is explained in more detail. In Fig. 14 a block diagram of the analog frontend including the processing in the FPGA is shown. The gray colored blocks are implemented inside the FPGA and they are used to process the discretized values. The Figure 14: Sampling chain in the FPGA framework. AD8283 analog frontend samples the differential IF input signals sequentially and the sampling frequency of the ADC 80 MHz f ADC = (9) N Div can be controlled with an AD9512 clock distribution circuit. The clock distribution circuit can be used to divide the 80-MHz crystal clock, where integer divider values in the range from 1 to 32 are allowed. In the software framework the divider can be set for every measurement mode independently. The clock divider N Div can be altered on the configuration page as shown in Fig. 15. In addition the ADC clock source can be configured to either use the clock from the baseband board or to use the clock from the frontend. If the clock source is set to Baseband. Then the RF chips and the sampling frequency are clocked from different sources. Most of the frontends from Inras support ramp synchronous sampling as they output an LVDS clock which can be used by the baseband board to drive the sampling circuits. By setting the clock source to Frontend the clock from the frontend is used. Apart from the clock setting the software framework can be used to set the gain, the low-pass cutoff frequency, the number of channels, and the input impedance of the AD8283. The different parameters can be altered by the drop down elements and the configuration is programmed to both AD8283 devices. Inras GmbH 18 AnHa

Figure 15: Configuring the sampling chain with the GUI. s The gray colored blocks are synthesized in the FPGA. The AD8283DatInt MMP is used to copy the serial data stream of the AD8283 devices on parallel channels. The values after the AD8283DataInt MMP are represented as signed 12-bit integer. As the AD8283 is a sequential sampling ADC the sampling frequency for a single channel f Chn = f ADC N Chn (10) depends on the number of activated channels N Chn. For a single data channel the sampled values are filtered with a CIC filter, with programmable filter parameters. The filter delay, the number of filter stages and the sampling rate reduction value can be programmed on the configuration page of the software framework. In addition, the filter can be bypassed. In this mode the input data is copied to the output without any signal processing performed on the data values. The sampling rate reduction value R must be an integer in the range from 1 to 65535. By setting the sampling rate reduction value the sampling frequency f s = f Chn R = 80 MHz N Div N Chn R (11) can be set to the desired value. The data of the CIC filters are output on a streaming interface in a signed 12.4 fractional number representation. The Frame Control (FrmCtrl) block can be used to insert a frame number in the data stream or to replace the sampled data with a defined data sequence in order to test the communication or Inras GmbH 19 AnHa

the signal processing in the final application. A detailed description of the function can be found in the FrmCtrl User Manual. After the FramCtrl MMP a FIFO is implemented using the external SRAM, which provides a buffer for the measured data. After the FIFO the data is transferred to either the ARM processor or USB interface. If the FIFO is not required then it can be bypassed. Inras GmbH 20 AnHa

9 List of Abbreviations ADC... Analog to Digital Converter CIC... Cascaded Integrator Comb Filter DBF... Digital Beamforming FIFO... First In First Out FMCW... Frequency-Modulated Continuous Wave FPGA... Field Programmable Gate Array IF... Intermediate Frequency LCDS... Low Voltage Differential Signaling MMP... Memory Mapped Peripheral MIMO... Multiple Input Multiple Output MISO... Master In Slave Out MOSI... Master Out Slave In MTI... Moving Target Indication RD... Range-Doppler RF... Radio Frequency RX... Receive SEQTRIG... Sequence Trigger Unit SPI... Serial Peripheral Interface TX... Transmit USPI... Universal Serial Peripheral Interface Inras GmbH 21 AnHa