Texas Instruments | Imaging Radar Using Cascaded mmWave Sensor Reference Design | User Guides | Texas Instruments Imaging Radar Using Cascaded mmWave Sensor Reference Design User guides

Texas Instruments Imaging Radar Using Cascaded mmWave Sensor Reference Design User guides
Design Guide: TIDEP-01012
Imaging Radar Using Cascaded mmWave Sensor
Reference Design
Description
This reference design provides a foundation for a
Cascaded Imaging Radar RF system. Cascaded
Radar devices can support long-range radar (LRR)
beam-forming applications as well as medium-range
(MRR) and short-range radar (SRR) MIMO
applications with enhanced angle resolution
performance.
The AWR1243P Cascade Radar RF development kit
has been used to estimate and track the position (in
the azimuthal plane) beyond 350 meters with a multidevice, beam-forming configuration. Additionally, this
system has demonstrated azimuth angular resolutions
as small as 1.4 degrees in a TDMA-MIMO
configuration.
Application
• Long range radar
• Imaging radar
• Traffic monitoring camera
Resources
TIDEP-01012
AWR1243
LP87524P-Q1
TMP112
Features
• Two or four-chip FMCW radar sensor for LRR,
MRR and SRR applications
• Detect objects (for example, cars and trucks) at a
distance beyond 350-m with range resolution of 35
cm; human RCS objects detectable at a distance of
150-m
• Antenna field of view ±70º with angular resolution
of approximately 1.4 degrees
• MATLAB MIMO and beamforming example code
provided
• AWR1243P based demonstration design
• Cascaded imaging radar front end beamforming
and MIMO configuration fully explained
Design Folder
Product Folder
Product Folder
Product Folder
ASK Our E2E™ Experts
dN
...
d2
TX1
TX2
TIDUEN5 – June 2019
Submit Documentation Feedback
TXN
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
1
System Description
www.ti.com
An IMPORTANT NOTICE at the end of this TI reference design addresses authorized use, intellectual property matters and other
important disclaimers and information.
1
System Description
ADAS control of a vehicle provides quality-of-life and safety benefits in addition to making the relatively
mundane act of driving safer and less difficult. The quality-of-life features include the ability of a vehicle to
park itself, or to determine whether a lane change is possible, and provide features like automatic cruise
control—where a vehicle maintains a constant distance with respect to the car ahead of it, essentially,
tracking the velocity of the car in front of it. Autonomous braking and collision avoidance are safety
features that prevent accidents caused by driver inattention. These features work by observing the area in
front of a car and alerting the ADAS subsystems if obstacles are observed that are likely to hit the car.
Implementing these technologies requires a variety of sensors to detect obstacles in the environment and
track their velocities and positions over time.
1.1
Why Cascade Radar?
Frequency-modulated continuous-wave (FMCW) radars allow the accurate measurement of range and
relative velocity of obstacles and other vehicles; therefore, radars are useful for autonomous vehicular
applications (such as parking assist and lane change assist) and car safety applications (autonomous
breaking and collision avoidance). An important advantage of radars over camera and light-detection-andranging (LIDAR)-based systems is that radars are relatively immune to environmental conditions (such as
the effects of rain, dust, and smoke). FMCW radars can work in complete darkness and also bright
daylight (radars are not affected by glare) because they transmit and receive electromagnetic waves.
When compared with ultrasound, radars typically have a much longer range and much faster time of
transit for their signals.
Despite the many advantages of radar technology, in many cases, automotive manufacturers today still
use camera sensors as the primary sensor technology used to make final safety decisions in the system.
The radar sensor is being used as the secondary sensor; meaning, the vehicle system receives the Radar
warning, but decides to take an action only upon the camera sensor verification. The main reason is
limitation in radar angular resolution. The radar sensors deployed today in most vehicles lack the ability to
distinguish between static objects with the same range and same relative velocity.
Today, a typical front radar sensor has about a 5-degree angular resolution that corresponds to the ability
of the sensor to distinguish between objects that are 8.5 m apart at 100 m. Objects that are closer than
8.5 m appear as one object. For example, a vehicle stopped in the right lane, might look like a shoulder
road street lamp for example, and therefore would be ignored by the safety system.
This is about to change with the introduction of the Imaging Radar solution from Texas Instruments (TI).
The TI Imaging Radar is a four-chip cascade solution, that acts like a single-chip sensor but achieves
20Log10(NTX) SNR gain in TX beamforming mode and 360/(N*pi) angular resolution (N is the number of
virtual antennas in a MIMO configuration).
The TI Imaging Radar solution, we can distinguish between static objects 0.6 degrees apart with all
antennae placed in single dimension linearly, and reach a 350-m object detecting range(angular resolution
is dependent on the antenna configuration and the number of TX/RX antennae).
This performance enables TI Imaging Radar to become the primary sensor in the vehicle and enhance
safety across weather and visibility conditions by providing a high-resolution image for both static and
moving objects.
1.2
TI Cascade Radar Design
TIDEP-01012 is an introductory application that demonstrates both a long-range beam-forming
configuration, and a shorter range, high-angular resolution MIMO configuration. This reference design can
be used as a starting point to design a standalone sensor for a variety of long range and imaging radar
applications. The TI Cascade RF reference design has demonstrated automobile target detection in
excess of 350 m along with 1.4 degree angular resolution.
2
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Description
www.ti.com
The flexible chirp and frame timing engine available on the AWR1243P device (similar to other AWR
family mmWave sensors) allows the system to function as a multi-mode radar, interleaving beam-forming
and MIMO configurations on a per frame basis. This enables the sensor designer to achieve best range
and best angular resolution across the array of Cascaded AWR1243P devices as the scene dynamics
requires.
Beamforming antenna across multiple, cascaded, AWR1243P devices provide sensor designers with
higher-output power and therefore lower-detectable target RCS, or increased range detection, or both.
Applications requiring detection of automobile, motorcycle, pedestrian, signage, bridges, and other
roadway objects and barriers at or beyond 350-m range can use this mode of operation.
In medium-range applications (150 m ranges), creating MIMO antenna arrays across multiple, cascaded,
AWR1243P devices allows the sensor designer to maximize the number of active antenna enabling
substantially improved angular resolution. This enables sub 1 degree resolution: true imaging radar
capability.
Figure 1. AWR1243P Four-Device Cascade Radar RF Radar Board
1.3
Key System Specifications
This reference design has two sets of specifications because the radar is used as a multi-mode radar.
MIMO is the first specification. TX beamforming (TXBF) is the second specification,
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
3
System Description
www.ti.com
Table 1. Key System Specifications
PARAMETERS
4
SPECIFICATIONS (MIMO)
SPECIFICATIONS (TXBF)
DESCRIPTION
Maximum Range
150 m
350 m
This represents the maximum
distance that the radar can
detect an object representing
an RCS of approximately 10
m2
Range Resolution
60 cm
150 cm
Range resolution is the ability
of a radar system to distinguish
between two or more targets
on the same bearing but at
different ranges. The resolution
is configurable, so the provided
number is just an example.
Azimuth Angle Resolution
1.4 degrees
1.4 degrees (with multiple
beam steering)
Angle resolution is the ability of
a radar system to distinguish
between two or more targets
with the same range and
velocity but different angles.
The resolution is equivalent in
both applications.
Elevation Angle Resolution
18 degrees
n/a
Elevation angle resolution is
only available for MIMO
application given the antenna
design on the TI cascade EVM
board.
Maximum Velocity
133 kmph
133 kmph
This is the native maximum
velocity obtained using a twodimensional FFT on the frame
data. For TDM MIMO case,
velocity compensation
algorithm is applied to recover
the native maximum velocity.
This specification will be
improved over time by showing
how higher-level algorithms
can extend the maximum
measurable velocity beyond
this limit.
Velocity Resolution
0.53 kmph
0.53 kmph
This parameter represents the
capability of the radar sensor to
distinguish between two or
more objects at the same
range but moving with different
velocities.
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Overview
www.ti.com
2
System Overview
2.1
Block Diagram
Figure 2 shows the block diagram of the cascade RF board.
Figure 2. Cascade Radar RF Board System Block Diagram
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
5
System Overview
2.2
www.ti.com
Highlighted Products
The AWR1243P is an integrated single-chip, frequency modulated continuous wave (FMCW) sensor
capable of operation in the 76 to 81 GHz frequency band. The device is built with TI’s low-power, 45-nm
RFCMOS processor and enables unprecedented levels of analog and digital integration in an extremely
small form factor. The device has four receivers and three transmitters with a closed-loop phase-locked
loop (PLL) for precise and linear chirp synthesis.
Each transmitter includes a programmable 6-bit phase shifter (5.625 degree step) to enable beam-forming
applications. Each devices also includes two 20-GHz local oscillator (LO) output and two 20-GHz LO input
paths for sharing the VCO output with neighboring devices. This enables a cost-effective, totally passive,
cascaded radar architecture.
The sensor includes a built-in radio processor (BIST) for RF calibration and safety monitoring. Based on
complex baseband architecture, the sensor device supports an IF bandwidth of 15 MHz with
reconfigurable output sampling rates in both complex and real sampling modes. Two separate Arm®
Cortex®-R4F based processors run the TI provided radar front-end, calibration, and host processor
interface firmware targeting ASIL-B.
2.3
2.3.1
Design Considerations
AWR1243P Cascade RF Board Features
Cascade Radar RF Board
4 × AWR1243P 76-81GHz Radar SoC
Integrated VCO, LO distribution, PA, LNA, ADC, 3 TX and 4 RX
Arm MCU R4 Controller
AWR RF Peripherals
12 × TX, 16 × RX Antennas
12 total transmitters across all 4 AWR1243 P devices16 total
receivers across all 4 × AWR1243P devices
Azimuth Array
86 element virtual array – enabling 1.4 degree angular resolution
Elevation Array
4 element virtual array – enabling 18 degree angular resolution
Embedded Antenna
Rogers RO3003 4-element, series-fed, patch antenna
20 GHz LOStar Distribution
2 × passive Wilkinson Power dividers fed by the Master
AWR12x device LO output to Slave AWR12x devices and
Master AWR12x device
AWR Digital Peripherals
Clock Distribution
LMK00804B low-jitter clock distribution
Digital Sync Distribution
LMK00804B low-jitter clock distribution
CSI2.0 4-lane
600Mbps/Lane, max 2.4Gbps ADC IF data per device
QSPI Flash
16Mbit QSPI flash for AWR firmware updates
Serial Peripherals
SPI, I2C, UART, GPIO
System Temperature
TMP112 I2C Temperature Sensors
Power
Radar Power Management IC (PMIC) Solution
2.3.2
2 × LP87524P-Q1 Quad-Channel, Integrated FET, Buck
Converters and LC filtering solution
AWR1243P Cascade RF Board Architecture
The AWR1243P Cascade RF board consists of four AWR1243P 77-GHz radar devices and their
associated power, clocking, synchronization, LO, and RF circuits.
Each AWR1243P RF, RX, and TX port is routed to its own set of etched, patch antenna. Each AWR1243P
on the RF board has a 4-port CSI2.0 transmitter that is used for sending radar data to a host processor
CSI2.0 receiver set. The entire AWR1243P configuration, control, and reset lines are made available on
two host-interface connectors implemented with Hirose FX32 series connectors.
6
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Overview
www.ti.com
The AWR1243P devices are separated into master and slave devices classes. AWR1243P #1, the master
device, uses the AWR1243P architecture built in LO distribution, clock distribution and frame
synchronization distribution to provide 40-MHz clock, 20-GHz LO and digital frame synchronization to
other three slave devices – AWR1243P #2, #3 and #4. This allows the system to generate and receive
coherent FMCW chirps across the 4 AWR1243P device array of transmitters and receivers; enabling
beam-forming and MIMO operation across the array of devices.
The 20 GHz LO distribution follows the start-network configuration described in the AWR1243 Cascade
Radar Application Note, SWRA574A. With the master AWR1243P #1 feeding a network of two Wilkinson
power dividers, that provide synchronous LO for the Master and Slave RF PA and mixer subsystems. All
clock distribution, synchronization distribution, and LO distribution requirements are documented in this
referenced application note.
The Cascade RF board accepts 5V DC, 8 A (max) power through the host board connectors. The primary
5 V system rail can be converted into the various AWR12x device rails by a pair of LP87524P, quadchannel, monolithic, buck-converters.
2.3.3
20 GHz Local Oscillator (LO) Distribution
20-GHz LO distribution to all AWR1243P devices is accomplished through an entirely passive,
transmission-line and etched power divider network. By using the dual LO outputs and LO inputs provided
on each AWR1243P device, all devices including the master device, receive the same master device
generated LO output. This results in all package and die routing delays common across all devices. The
PCB designer is left with the task of delay matching only the BGA to BGA delays.
By using both of the 6-dBm (typical) LO output, and minimizing transmission line and power divider losses,
no external amplifiers are required to be added to the LO network. Additionally, by outputting 20 GHz LO
versus full 77-GHz RF for RF synchronization, the LO PCB link budgets can be more relaxed.
The passive LO distribution network and relaxed link budget results in more PCB design flexibility and a
lower overall system cost.
Figure 3 shows the block diagram of the delay matched LO distribution. Figure 4 shows the implemented
Wilkinson power dividers in GCPW transmission-line. Fanout from the AWR1243P devices and power
dividers is through GCPW lines. The majority of the LO distribution length is carried through an internal
Stripline layer.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
7
System Overview
www.ti.com
Figure 3. 20 GHz LO Distribution Block Diagram
Figure 4. 20 GHz LO Distribution Board Implementation
8
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
3
System Design Theory
The Cascade RF MIMO and beamforming antenna design, chirp, and frame configuration are system
variables that must be co-designed for a particular application. Both the antenna array, MIMO chirp, and
beamforming chirp design are presented here.
3.1
Antenna Configuration
Figure 5 shows the antenna array on the Cascade RF board with one master device and three slave
devices. Together these devises create a total of 12 TX channels and 16 RX channels. Among the 12 TX
channels, the three TXs (TX1/2/3) from the master device are placed in the vertical direction for elevation
antenna estimation. The remaining 9 TXs (TX4 through TX12) and all 16 RX channels are placed in the
horizontal plane for azimuth angle estimation. The relative distances between TXs and between RXs are
shown in Figure 4. The three elevation antennas are placed to form a minimum redundancy array to
improve elevation angle resolution. Every two azimuth TXs are spaced 2 wave length apart. RX array A
and RX array C are placed 16-wavelengths apart. RX array C and RX array B are placed fourwavelengths apart. The virtual array in MIMO mode is shown in Figure 5, with 86 virtual antennas in
azimuth direction. The overlapped antenna in azimuth is not shown in the figure. The elevation angle
resolution is equivalent to the resolution achieved with 7 antennas.
Figure 5. Device and Antenna Array Groups
The Cascade RF board can also be used in two chip cascade mode with 6 TX and 8 RX channels
highlighted in Figure 6. Figure 6 shows the virtual MIMO array in 2-device cascade mode using TX ID of [1
2 3 10 11 12] and RX ID of [ 4 3 2 1 16 15 14 13], with 23 azimuth antennas and same elevation
configuration with 4-device cascade.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
9
System Design Theory
www.ti.com
Figure 6. Antenna Array Positions
26.5
16
0.5 0.5 0.5
4
0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.5 0.5 0.5
RX ARRAY-B
RX ARRAY-C
2-chip cascading antennas
RX ARRAY-A
This distance is to meet 2-chip
cascading requirement (azimuth
and elevation)
1
0.5
Scale unit is
1.5
2
0.5
0.5
0.5
Figure 7. 4-Device Cascade Virtual MIMO Array with 12 TX and 16 RX
10
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
Figure 8. 2-Device Cascade Virtual MIMO Array with 6 TX and 16 RX
3.2
Chirp Configuration
Chirp configurations are different for MIMO and TXBF use cases.
3.2.1
Chirp Configuration for MIMO Radar
For the MIMO application, to achieve the system performance specified above, the chirp configuration in
Table 2 is used.
Table 2. Chirp Configuration - MIMO
PARAMETER
SPECIFICATIONS
Idle time (µs)
4
ADC start time (µs)
5
Ramp end time (µs)
23
Number of ADC samples
256
Frequency slope (MHz/µs)
15
ADC sampling frequency (kSPS)
15
Number of chirps per frame
256
Effective chirp time (µs)
17
Bandwidth (MHz)
256
Frame length (ms)
4.4
With this chirp configuration the MIMO performance shown in Table 2 is achieved. The primary goal was
to achieve a maximum distance of about 150 m and maximum unambiguous velocity of 133 km/h. See
Programming Chirp Parameters in TI Radar Devices for more details. The choice of the chirp periodicity is
a trade-off between range resolution and maximum velocity. The velocity resolution is defined by the
number of chirps per frame.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
11
System Design Theory
3.2.2
www.ti.com
Chirp Configuration for TX Beamforming Radar
For the TX beamforming application, to achieve the system performance specified above, the chirp
configuration in Table 3 is used.
Table 3. Chirp Configuration - TX Beamforming
PARAMETER
SPECIFICATIONS
Idle time (µs)
4
ADC start time (µs)
5
Ramp end time (µs)
23
Number of ADC samples
256
Frequency slope (MHz/µs)
15
ADC sampling frequency (MSPS)
15
Number of chirps per frame
256
Effective chirp time (µs)
17
Bandwidth (MHz)
256
Frame length (ms)
4.4
With this chirp configuration the TXBF performance shown in Table 3 is achieved. The primary goal was to
achieve a maximum distance of about 350 m and maximum velocity around 133 km/h. See Programming
Chirp Parameters in TI Radar Devices for more details.
The amount of phase value to program to each TX channel is computed as a function of array factor and
target angle. Assuming N TX channels, with TX1 as a reference, the distance between every other
antenna and TX1 is the known distance when the antenna array is designed during board development.
As shown in Figure 9 for TI cascade EVM, the distance between any two adjacent TX channels is two
times the wavelength.
Figure 9. TX Array for Beamforming
dN
...
d2
TX1
TX2
TXN
Given the notation in Figure 9, the phase value for each TX channel is calculated as:
(1)
The ideal phase value is further quantified by the allowed phase step size of 5.625 degree to calculate the
integer value to be programed to the registers (TX calibration phase value will also be added as shown in
the next section).
(2)
As an example, for TI cascade EVM, 9 azimuth TX antennas can be used for beam steering, with [d2 =
2λ,d3 = 4λ,…,d9 = 16λ]. If the desired steering angle is 30 degrees, then the phase vector is as shown in
degrees in Equation 3.
12
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
(3)
The TX beamforming application supports both chirp based beam steering and frame based beam
steering. Advanced frame configuration is used for either case.
3.2.2.1
Chirp Based Beam Steering
In the subframe used for TX beamforming, the number of chirps used in this subframe is equal to the
number of different desired steering angles. Each chirp configuration is associated with phase values for
the TX array, calculated based on the corresponding desired steering angle. Within a burst, all chirp
configurations are looped over till the end of the burst. Figure 10 demonstrates the mechanism of chirp
based beam steering.
Figure 10. Chirp Based Beam Steering
angle 1
angle 2
angle N
Chirp
config 1
Chirp
config 2
Chirp
config N
Chirp
config 1
«
«
First loop
Chirp
config 1
Chirp
config N
Chirp
config 2
«
second loop
Chirp
config 2
Chirp
config N
«
Last loop
One subframe (only one burst)
3.2.2.2
Frame Based Beam Steering
In the subframe used for TX beamforming, the number of chirps used in this subframe is equal to the
number of different desired steering angles. Each chirp configuration is associated with phase values for
the TX array, calculated based on the corresponding desired steering angle. The number of burst within
this subframe is equal to the number of different desired angles (same as number of chirp configurations).
Each burst is associated with each chirp configuration, so a burst is equivalent to a frame within which
each chirp has the same phase shifter value. Figure 11 demonstrates the mechanism of frame based
beam steering.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
13
System Design Theory
www.ti.com
Figure 11. Frame Based Beam Steering
angle 2
angle 1
Chirp
config 1
Chirp
config 1
Chirp
config 1
«
First burst
Chirp
config 2
angle N
Chirp
config 2
Chirp
config N
Chirp
config 2
«
«
second burst
Chirp
config N
Chirp
config N
«
Last burst
One subframe (N bursts)
3.3
Antenna Calibration
The purpose of antenna calibration is to estimate the frequency, phase and amplitude mismatches across
one master device and three slave devices. The mismatches can be caused by various reasons, such as
path length mismatch, chip to chip variation, antenna coupling, and so on. The frequency mismatch is
usually minimized by routing path length match during the board layout stage. The calibration method
proposed by TI development kit is a one-time boresight calibration method. It is suggested to perform
board specific calibration to achieve best angle performance.
3.3.1
How To Generate Phase/Gain Calibration Matrix
To generate a calibration matrix, it is suggested to put a corner reflector at distance of 5 m and beyond,
with a typical corner reflector RCS of 1 m2 to approximately 2 m2, as shown in Figure 12. The corner
reflector should be aligned with the array center of board in both azimuth and elevation direction. A level
sensor with laser pointer can be used for alignment.
14
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
Figure 12. Antenna Calibration Setup
When the corner reflector is placed at the center position, a few frames of raw ADC data needs to be
collected by using the example script provided in the mmWave Studio release. During the data collection,
turn on 12 TX in a TDM MIMO mode and all 16 RX channels are on at the same time.
The calibration matrix generation follows the sequence described by Figure 13. After the raw binary data
is read and formatted so that data from 192 virtual channels are separated. Range FFT is performed on
each channel and the peak corresponding to the corner reflector is identified as the local maximum within
[D – 1, D + 1] meters, where D is the approximate target range provided by user. The FFT peak index
from all 192 channels form a matrix size of 12 x 16, which will be used for frequency calibration step. The
complex values at the peaks from all 192 channels form a complex calibration matrix of size 12 x 16,
which will be used for phase and amplitude calibration. These two matrixes are saved to a matlab .mat file
to be loaded when applying the calibration results to the ADC data.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
15
System Design Theory
www.ti.com
Figure 13. Flow to Generate Phase and Gain Calibration Matrix
Approximate target
distance D meters
ADC_Channel _1
«
3.3.2
«
«
ADC_Channel
_192
Identify peak
of target within
[D-1 D+1]
meters
Range FFT
Range FFT
Identify peak
of target within
[D-1 D+1]
meters
FFT peak index
matrix size 12x16
Save to a mat file
Complex calibration
matrix
size 12x16
Applying Calibration For MIMO Operation
The saved calibration information is applied to the raw ADC data collected in MIMO mode in two steps on
the flow shown on Figure 14.
3.3.2.1
Frequency Calibration
The FFT index of TX1/RX1 channel is used as reference that is compared with all the other 191 channels.
The FFT index difference and chirp parameters are used to compute the frequency calibration vector
using Equation 4.
where
•
•
•
•
•
•
•
•
∆P is the FFT peak index difference for a virtual channel
fchirp is the FMCW chirp frequency slope
fcalib is the FMCW chirp frequency slope used for calibration which can be different from fchirp
N is the number of ADC samples per chirp
n⃗= [0 N-1] is the ADC sample index
Iinterp is the interpolation factor used for frequency calibration that is further explained in the next section
F⃗is the computed frequency calibration vector
exp(-j×F⃗) is multiplied with the raw ADC data for frequency calibration
(4)
For each virtual channel, F⃗is calculated and applied to the corresponding ADC data
3.3.2.2
Phase and Amplitude Calibration
After frequency calibration, the phase and amplitude calibration value is calculated based on the complex
value of the reference channel
where
•
•
16
Cref is the complex value of the reference channel
Ctxi/rxj is the complex value of the other channels
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
•
Cph_amis the calculated phase and amplitude calibration value
Cph_am is 1 for reference channel and different for every other channel
Cph_amis multiplied with the output from the frequency calibration and the result is the final calibrated
ADC data
(5)
Figure 14. Flow to Apply Calibration Matrix to ADC Data
FFT peak index of
TX1/RX1 reference
channel
Calculate
frequency
calibration
vector
FFT peak index
matrix size 12x16
Raw ADC
data
Perform
frequency
calibration
Chirp
parameters
3.3.3
Complex
calibration matrix
size 12x16
Perform
phase/amplitu
de calibration
Calibrated
ADC data
Phase/amplitude of
TX1/RX1 reference
channel
Applying Calibration For TX Beam-forming Operation
The saved calibration information is applied in TX beamforming operation as shown in Figure 15. The
phase value of the 12 × 16 complex value matrix obtained from section “How to generate calibration
matrix” is averaged across the 16 RX channels to obtain a phase calibration vector for the 12 TX
channels. According to the index of TX channel used in actual TX beamforming mode, the corresponding
phase value within this 12 × 1 vector is used for the phase compensation. The compensation phase value
together with the phase value calculated based on the desired steering angle is used to program the
phase shift register value before starting beam steering process.
Figure 15. Calibration in TX Beamforming
Desired
steering
angle
Phase of
FFT peak
index matrix
size 12x16
Average
across 16
RX
channels
TX phase
calibration vector
size of 12x1
+
Average
across 12
TX
channels
RX phase
calibration vector
size of 16x1
RX data
post
processing
TIDUEN5 – June 2019
Submit Documentation Feedback
TX1 TX2 TX3 TX1 TX2 TX3 TX1 TX2 TX3 TX1 TX2 TX3
Device 4
Device 3
Device 2
Device 1
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
17
System Design Theory
www.ti.com
Similarly, the phase across the 12 TX channels is averaged to obtain a phase calibration vector for the 16
RX channels. This is phase only RX calibration assuming that there is no frequency mismatch across the
16 RX channels. The frequency mismatch across 16 RX channels can be captured by collecting
calibration data by following these steps.
1. Put a corner reflector at a known distance, for example, approximately 6 m
2. Configure all TX channels to steer towards zero degrees
3. Check the FFT peak index of all 16 RX channels; the FFT peak index indicates the frequency
mismatch, similar to the frequency mismatch described in Section 3.3.2.
The phase calibration value is applied during the RX data post processing stage. More details will be
discussed in the signal processing section of TX beamforming.
3.4
Cascade Radar Signal Processing Chains
3.4.1
MIMO Signal Processing Chain
The raw signal collected in MIMO mode is processed in the flow shown in Figure 16.
Figure 16. MIMO Signal Processing Chain
Calibration matrix
ADC_
Ch_1
Frequency/
phase
calibration
Non-coherent
integration
Range/Dop
pler FFT
Binary data file
«
«
ADC data
read/parse
and reformat
Chirp configuration
parameters
ADC_
Ch_1
Frequency/
phase
calibration
Range/Dop
pler FFT
Heat
map
CFAR cross-1D
detection
Point
cloud
GUI
Max velocity
extension and
phase
compensation
ADC data read
and calibration
3.4.1.1
Azimuth/ele
vation angle
estimation
Detection
ADC Data Read and Calibration
After each data collection, a binary data file and the corresponding chirp configuration parameters are
saved. These two files are the inputs to the ADC data read and calibration module. The binary data file is
read/parsed according to the chirp parameters based on the data format, samples per chirp, chirps per
frame, number of TX/RX channels. Then the data is reformatted into a 4D matrix, with the dimension of
samples per chirp, chirps per frame, number of RX channels, and number of TX channels. Each of the
TX/RX channel is calibrated according the pre-calculated calibration matrix that can be obtained using the
procedures as described in Section 3. (reference)
3.4.1.2
Range/Doppler FFT
For each TX/RX channel, range FFT is performed followed by Doppler FFT. The default windowing is a
Hanning window. The FFT size is determined as the closest integer that is power of 2.
18
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
System Design Theory
www.ti.com
3.4.1.3
Detection
The output of range/Doppler FFT is sent to the detection module. The first step is non-coherent integration
across all the virtual channels. The integration output goes through the CFAR detection step that does 1D
range CFAR detection followed by 1D Doppler CFAR detection at the detected range bins. For each cross
detected range/Doppler point, the maximum velocity extension algorithm is applied to correct any possible
velocity ambiguity caused by TDM MIMO. The recovered velocity is used to correct the phase jump that is
again caused by the TDM MIMO for each detected point. The details of maximum velocity extension and
phase compensation will be discussed in a separate section.
3.4.1.4
Angle Estimation
For each detected point, angle estimation is performed based on the signal vector with corrected phase. If
the configured array is azimuth direction only, only azimuth angle estimation is needed. If a 2D configured
array is involved, both azimuth and elevation angle estimation is performed. Note that multiple angles will
be detected in the azimuth direction, but only the maximum peak is selected in the elevation direction.
3.4.1.5
GUI
Point cloud and heat map can be sent to GUI for visualization.
3.4.2
TX Beamforming Signal Processing
This section provides an example signal processing chain in TX beamforming mode. Users might have
their own preferred signal chain depending on the application. This example signal processing chain is
used to generate an azimuth/range heat map by stitching multiple TX beamforming scans together.
Figure 17 shows the block diagram of the example signal processing chain in TX beamforming mode. The
desired steering angles are defined as β1 β2 ... βn. For each angle βi, the phase shifter value is calculated
based on the TX phase calibration vector and βi with given TX antenna positions (see Section 2.3.3). The
received ADC data from each RX channel goes through range and Doppler FFT, and only zero Doppler
bin is selected for range/azimuth heat map generation assuming it is a static test scene. For moving test
scene, user needs to change accordingly. For each range bin at zero Doppler, after phase compensation
using calibration vector, RX beamforming is performed to steering the RX beam towards the same βi
where the TX channels are focusing on. As a result, one range line is generated for each angle βi. By
stitching all the lines generated with different βi angles, a range/azimuth heat map is generated.
Figure 17. TX Beamforming Signal Processing Chain
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
19
System Design Theory
www.ti.com
Figure 18 shows the example heat map results. Two corner reflectors are separated by around 1.8
degrees at 6 m within the anechoic chamber. The anechoic chamber size is around 6 m in width and 12 m
in length. The radar is put around 4 m away from one side. The top left and top right plots are the heat
map results (top-down view) running in MIMO mode and stitched TX beamforming mode. The TX beam is
steering within [–60 60] with step size of 0.5 degree. The rectangular shape corresponding to the anechoic
chamber wall, and the two bright spots in the middle are corresponding to the two corner reflectors. The
bottom left and bottom right plots are the heat map results in a 3D view, with the height indicates the
reflection intensity. The separation of the two peaks is more visible in the 3D view. It is worth to note that
the grading lobes caused by the wide distance of adjacent TX channels are no longer observed because
of the cancellation in the RX beamforming process.
Figure 18. Heat Map – MIMO and Stitched TX Beamforming – Top Down View
Figure 19. Heat Map – MIMO and Stitched TX Beamforming – 3D View
This experiment result proves that the receiver angle resolution in these two operation modes are
equivalent since the effective aperture size is the same. Further, the signal SNR in the beamforming is
much higher due to the coherent gain in TX beamforming mode. The SNR in MIMO mode can be
improved by increasing the chirp integration time.
20
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
4
Hardware, Software, Testing Requirements, and Test Results
4.1
Required Hardware and Software
4.1.1
Hardware
The tests were performed using the AWR1243P four-device Cascade Radar RF board from Texas
Instruments together with TDA2 based capture System.
Figure 20. AWR1243P Four-Device Cascade Radar RF Board with TDA2 Based Capture Board
4.1.2
Software
The cascade board was configured using an updated version of the MMWAVE-STUDIO tools. Algorithms
implemented in Matlab were used to post process the captured raw data.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
21
Hardware, Software, Testing Requirements, and Test Results
4.2
www.ti.com
Testing and Results
Multiple test scenarios were setup to explore the capabilities of the Cascade Radar RF design in both
MIMO and TX beamforming operation.
All data presented here was collected with the Hardware mentioned above.
4.2.1
Test Scenarios
4.2.1.1
•
•
•
•
•
•
•
•
•
•
•
•
4.2.1.2
•
•
•
•
4.2.2
MIMO Test Scenarios
In-lab Angular Resolution
Side-by-side Car Detection Resolution
Side-by-side Car Detection Resolution – MUSIC Algorithm
Car, Pedestrian and Other Targets Close Range Separation
Car Contour and Orientation
Car Door Contour
Bike Contour and Orientation
Fence Contour Detection
Curb Contour Detection
Manhole Contour Detection
Height Measurement
Single-Device vs. Cascade Devices
TX Beamforming Test Scenarios
In-lab Beamforming Control Pattern
Pedestrian Long-Range Detection
Car Medium-Range detection
Car Long-Range Detection
MIMO Test Results
4.2.2.1
In-lab Angular Resolution
A basic azimuth separation test was performed within an anechoic chamber radar test range. Two corner
reflectors were placed approximately 8 m from the AWR1243P radar sensor. 1.5 degrees of azimuth
separation was applied to the reflectors and the resulting azimuth resolution was measured.
In the current configuration, with 86 virtual array elements in the azimuth axis, and lambda/2 spacing of
each element, a best-case 1.4 degree angular resolution is possible. This system measured a 1.5 degree
separation, which is in close alignment to the expected.
22
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
Figure 21. Two Corner Reflectors Separated by 1.5 Degrees in Azimuth
Figure 22. Range-Doppler FFT Plots Showing Detected, Separated Peaks From Both Reflectors
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
23
Hardware, Software, Testing Requirements, and Test Results
4.2.2.2
www.ti.com
Car Angular Resolution Scenario
A test was performed with two cars located at 112 m range from the sensor at varying angular
separations, again showing the angular resolution capabilities of the AWR1243P MIMO radar operation.
Figure 23. Two-Car Angular Separation Test Setup
Figure 24. Two-Car Angular Separation Test Results
24
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
4.2.2.3
Car, Pedestrian, and Other Targets Close Range Separation
A test was performed showing the AWR1243P MIMO radar is able to separate a person and a bicycle at
different lateral distances away from the car. This represents a more challenging scenario due to the
pedestrian (or another low RCS object) occupying the same range bin as the high reflecting point from the
car (high RCS object).
Figure 25. Bicycle and Person 1.5 Meters Away From Vehicle
Figure 26. Bicycle and Person 1.0 Meter (left) and 0.5 Meter (right) Away From Vehicle
4.2.2.4
Car Contour and Orientation
A test was performed showing the AWR1243P MIMO radar detecting the contour and orientation of a car.
The azimuth and range resolution allows for the detection of the contour of many vehicle surfaces.
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
25
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
Figure 27. Car at 0 Degrees (Top), 90 Degrees (Middle) and 135 Degrees (Bottom) Orientation and
Corresponding Azimuth Range Heatmap
26
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
4.2.2.5
Curb Contour Detection
A test was performed showing the AWR1243P MIMO radar detecting the contour and orientation of
multiple curbs in a parking lot setting. The azimuth and range resolution allows for the detection of the
contour of these shorter driving obstacles.
Figure 28. Parking Lot Curb Scene (Top), and Resulting Azimuth Range Heatmap (Bottom)
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
27
Hardware, Software, Testing Requirements, and Test Results
4.2.3
www.ti.com
TX Beamforming Results
4.2.3.1
In-lab Beamforming Control Pattern
A basic beam steering test was performed within an anechoic chamber radar test range. A single corner
reflector was placed approximately 8 m from the AWR1243P radar sensor. Beamsteering vectors were
then programmed into the AWR1243P devices to achieve a 15 degree beam rotation. The resulting target
return versus angle graph was then compared against the simulated 15-degree beam rotation pattern.
Close alignment between the simulated and measured system was observed.
28
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
Figure 29. Simulated (Top) and Measured (Bottom) 15 Degree Beam Steering Test
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
29
Hardware, Software, Testing Requirements, and Test Results
4.2.3.2
www.ti.com
Pedestrian Long-Range Detection
In this test a pedestrian jogging from 30 m to 140 m. The AWR1243P Cascade RF board is running in 9TX beamforming mode.
The range profile shows the pedestrian at 120 m with 20-dB SNR. The Doppler-Range plot is shown to
the right. Slight Doppler displacement can be seen corresponding to the pedestrian velocity.
Figure 30. Range Profile (top left), and Doppler-Range Plot of a Pedestrian Jogging Away From Radar at
120 Meters Range
4.2.3.3
Car Long-Range Detection
Similar to the pedestrian long range detection test, this test observes a car at 350-m range. The
AWR1243P Cascade RF board is running in 9-TX beamforming mode.
The range profile shows the vehicle at 350 m with 14-dB SNR.
30
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
Hardware, Software, Testing Requirements, and Test Results
www.ti.com
Figure 31. Range Profile (left), and Doppler-Range Plot of a Vehicle 350 Meters Range
TIDUEN5 – June 2019
Submit Documentation Feedback
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
31
Design Files
www.ti.com
5
Design Files
5.1
Schematics
To download the schematics, see the design files at TIDEP-01012.
5.2
Bill of Materials
To download the bill of materials (BOM), see the design files at TIDEP-01012.
5.3
Altium Project
To download the Altium Designer® project files, see the design files at TIDEP-01012.
5.4
Gerber Files
To download the Gerber files, see the design files at TIDEP-01012.
5.5
Assembly Drawings
To download the assembly drawings, see the design files at TIDEP-01012.
6
Software Files
The cascade board was configured using an updated version of the MMWAVE-STUDIO tools.
7
Related Documentation
1.
2.
3.
4.
7.1
Texas
Texas
Texas
Texas
Instruments, Programming Chirp Parameters in TI Radar Devices Application Report
Instruments, AWR1243 76-GHz to 81-GHz high-performance automotive MMIC
Instruments, AWR1243 Cascade Application Report
Instruments, MMWAVE-STUDIO
Trademarks
E2E is a trademark of Texas Instruments.
Altium Designer is a registered trademark of Altium LLC or its affiliated companies.
Arm, Cortex are registered trademarks of Arm Limited.
7.2
Third-Party Products Disclaimer
TI'S PUBLICATION OF INFORMATION REGARDING THIRD-PARTY PRODUCTS OR SERVICES DOES
NOT CONSTITUTE AN ENDORSEMENT REGARDING THE SUITABILITY OF SUCH PRODUCTS OR
SERVICES OR A WARRANTY, REPRESENTATION OR ENDORSEMENT OF SUCH PRODUCTS OR
SERVICES, EITHER ALONE OR IN COMBINATION WITH ANY TI PRODUCT OR SERVICE.
32
Imaging Radar Using Cascaded mmWave Sensor Reference Design
Copyright © 2019, Texas Instruments Incorporated
TIDUEN5 – June 2019
Submit Documentation Feedback
IMPORTANT NOTICE AND DISCLAIMER
TI PROVIDES TECHNICAL AND RELIABILITY DATA (INCLUDING DATASHEETS), DESIGN RESOURCES (INCLUDING REFERENCE
DESIGNS), APPLICATION OR OTHER DESIGN ADVICE, WEB TOOLS, SAFETY INFORMATION, AND OTHER RESOURCES “AS IS”
AND WITH ALL FAULTS, AND DISCLAIMS ALL WARRANTIES, EXPRESS AND IMPLIED, INCLUDING WITHOUT LIMITATION ANY
IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT OF THIRD
PARTY INTELLECTUAL PROPERTY RIGHTS.
These resources are intended for skilled developers designing with TI products. You are solely responsible for (1) selecting the appropriate
TI products for your application, (2) designing, validating and testing your application, and (3) ensuring your application meets applicable
standards, and any other safety, security, or other requirements. These resources are subject to change without notice. TI grants you
permission to use these resources only for development of an application that uses the TI products described in the resource. Other
reproduction and display of these resources is prohibited. No license is granted to any other TI intellectual property right or to any third
party intellectual property right. TI disclaims responsibility for, and you will fully indemnify TI and its representatives against, any claims,
damages, costs, losses, and liabilities arising out of your use of these resources.
TI’s products are provided subject to TI’s Terms of Sale (www.ti.com/legal/termsofsale.html) or other applicable terms available either on
ti.com or provided in conjunction with such TI products. TI’s provision of these resources does not expand or otherwise alter TI’s applicable
warranties or warranty disclaimers for TI products.
Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright © 2019, Texas Instruments Incorporated
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Related manuals

Download PDF

advertising