SDR Hacking - Supplemental 269: Automotive Radar Signal Analysis
S-0269 - Supplemental 269 - Automotive Radar Signal Analysis
Author: Patrick Luan de Mattos
Category Path: sdr-hacking
Audience Level: Advanced
Generated at: 2026-04-03T15:59:15.778Z
Supplemental Chapter: Automotive Radar Signal Analysis
Supplemental Index: 269
Title: Automotive Radar Signal Analysis
Audience Level: Advanced
Focus Keywords: 24 GHz and 77 GHz concepts, FMCW behavior, and safe theoretical study
1) Position of this Supplemental Chapter in the Advanced SDR Roadmap
This supplemental chapter, "Automotive Radar Signal Analysis," is positioned within the advanced stages of an SDR roadmap, building upon foundational knowledge in digital signal processing, radio frequency (RF) principles, and software-defined radio architecture. It assumes a strong understanding of concepts like modulation schemes, Fourier transforms, spectrum analysis, and basic signal detection.
This chapter serves as a bridge to specialized applications of SDR, moving beyond general-purpose signal analysis to a specific, high-impact domain. It complements earlier chapters on topics such as:
- Advanced RF Fundamentals: Deep dives into antenna theory, propagation, and RF front-end design.
- Digital Signal Processing Techniques: Advanced filtering, estimation, and detection algorithms.
- GNU Radio and SDR Implementation: Practical application of SDR hardware and software for complex signal manipulation.
- Communication Systems Analysis: Understanding of various modulation and coding schemes.
By delving into automotive radar, we explore real-world applications that demand precise signal characterization, interference mitigation, and an understanding of the underlying physics of radar operation. This chapter prepares the advanced practitioner for tasks such as reverse engineering, security analysis, and the development of advanced automotive sensing systems.
2) Deep Conceptual Explanation
Automotive radar systems are critical for modern vehicle safety and automation features, including Adaptive Cruise Control (ACC), Automatic Emergency Braking (AEB), Blind Spot Detection (BSD), and parking assistance. These systems rely on emitting radio waves and analyzing the reflected signals to determine the range, velocity, and angle of surrounding objects.
Frequency Bands: 24 GHz and 77 GHz Concepts
Modern automotive radar predominantly operates in two key frequency bands:
24 GHz Band (e.g., 24.05 GHz to 24.25 GHz): This band, often referred to as the "short-range radar" band, is well-suited for applications requiring detection of objects at closer distances, such as parking sensors and BSD. Its advantages include:
- Wider Bandwidth Potential: Allows for better range resolution.
- Lower Component Cost: Historically, components in this band were more mature and cost-effective.
- Regulatory Considerations: Specific regulations govern its use for automotive applications.
77 GHz Band (e.g., 77 GHz to 81 GHz): This is the primary band for longer-range and higher-resolution radar applications, including ACC and AEB. Its advantages are significant:
- Larger Available Bandwidth: The 77-81 GHz range offers up to 4 GHz of bandwidth, enabling very high range resolution (e.g., centimeters).
- Smaller Antenna Size: Higher frequencies allow for smaller antenna elements, facilitating integration into vehicle designs.
- Reduced Interference: Less crowded than lower frequency bands, although interference is still a concern.
- Higher Doppler Sensitivity: Better for detecting subtle velocity changes.
The choice of frequency band directly impacts the radar's capabilities, antenna size, and achievable resolution.
FMCW Behavior
Frequency Modulated Continuous Wave (FMCW) is the dominant waveform used in automotive radar. Unlike pulsed radar, FMCW transmits a continuous wave whose frequency is modulated over time. This modulation allows for simultaneous measurement of range and velocity.
The core principle of FMCW is to transmit a signal whose frequency increases (or decreases) linearly over a specific period, known as the chirp. When this signal reflects off a target and returns to the radar receiver, its frequency will have shifted due to two primary effects:
- Time Delay (Range): The signal takes a finite amount of time to travel to the target and back. This delay causes the received signal's frequency to be lower than the transmitted signal's frequency at any given moment. The difference in frequency is directly proportional to the target's range.
- Doppler Shift (Velocity): If the target is moving, the reflected signal will experience a Doppler shift, causing its frequency to be higher (approaching) or lower (receding) than it would be if the target were stationary. This Doppler shift is proportional to the target's radial velocity.
By analyzing the frequency difference between the transmitted and received chirps, both range and velocity can be extracted. This is typically done by mixing the received signal with the transmitted signal (or a delayed version of it).
Chirp Structure:
A typical FMCW chirp can be visualized as a rising frequency over time.
Transmitted Signal Frequency
^
| /----/----/----/
| / / / /
| / / / /
| / / / /
| / / / /
+------------------------> TimeBeat Frequency:
When the received signal (delayed and potentially Doppler-shifted) is mixed with the transmitted signal, a new signal is generated whose frequency is the difference between the two. This is called the beat frequency ($f_b$).
For a stationary target at range $R$, the time delay is $\tau = 2R/c$, where $c$ is the speed of light. If the chirp has a slope of $S$ (frequency change per unit time, e.g., Hz/s), the frequency difference at the moment of mixing will be:
$f_b = S \tau = S \frac{2R}{c}$
Thus, the range $R$ can be calculated as:
$R = \frac{c f_b}{2S}$
For a moving target, the Doppler shift ($f_d$) adds to or subtracts from this beat frequency. If the target is moving with velocity $v$, the Doppler shift is $f_d = 2fv/c$, where $f$ is the carrier frequency. This makes the beat frequency for a moving target:
$f_{b,moving} = S \tau \pm f_d$
To disentangle range and velocity, automotive radars typically employ a sequence of chirps with varying modulation characteristics (e.g., up-chirps and down-chirps, or chirps with different slopes and durations). By analyzing the beat frequencies across this sequence, both range and velocity can be precisely determined. For example, using up-chirps and down-chirps allows for separation of range and Doppler effects.
3) Architecture and Signal Reasoning
A simplified automotive radar system architecture comprises the following key components:
- Transmitter: Generates the FMCW chirp waveform. This often involves a Voltage-Controlled Oscillator (VCO) whose frequency is controlled by a precise waveform generator (e.g., DAC output).
- Antenna (Transmit): Radiates the FMCW signal into the environment. Automotive radars often use patch antennas or antenna arrays for beamforming.
- Antenna (Receive): Captures the reflected signals from targets. Often, multiple receive antennas are used (e.g., in MIMO configurations) to achieve spatial resolution (angle estimation).
- RF Front-End (LNA, Mixer, Filter): Amplifies the weak received signals, downmixes them to an intermediate frequency (IF) or directly to baseband, and filters out unwanted noise and interference.
- Analog-to-Digital Converter (ADC): Digitizes the analog intermediate frequency (IF) or baseband signals. The sampling rate and resolution of the ADC are critical for capturing the detailed frequency information.
- Digital Signal Processor (DSP) / FPGA: This is the heart of the radar processing. It performs:
- Chirp Processing: Detecting individual chirps and their characteristics.
- Range FFT: Applying a Fast Fourier Transform (FFT) to the digitized beat signal within a single chirp to identify range bins and their corresponding beat frequencies.
- Doppler Processing: Analyzing the beat frequencies across multiple chirps to extract Doppler shifts. This often involves another FFT on the range bins.
- Target Detection: Identifying peaks in the range-Doppler map that correspond to potential targets.
- Angle Estimation (if applicable): Using techniques like beamforming or MUSIC algorithm on multi-channel data.
- Object Tracking: Maintaining a track of detected objects over time.
Signal Reasoning and Analysis Flow
The signal processing chain can be visualized as follows:
graph TD
A[FMCW Chirp Generation] --> B(Transmit Antenna)
B --> C{Environment}
C --> D(Receive Antenna)
D --> E[RF Front-End (LNA, Mixer, Filter)]
E --> F[ADC]
F --> G[Digital Signal Processor (DSP/FPGA)]
G --> H(Range FFT)
H --> I(Beat Frequency Analysis - Range)
G --> J(Doppler Processing - Multi-Chirp)
J --> K(Doppler Frequency Analysis - Velocity)
I & K --> L(Range-Doppler Map)
L --> M(Target Detection & Tracking)Detailed Signal Analysis Steps:
Acquisition: Capturing raw I/Q samples from the ADC. These samples represent the digitized beat signal after mixing.
Windowing: Applying a window function (e.g., Hanning, Hamming) to the digitized signal within a chirp to reduce spectral leakage before the FFT.
Range FFT:
- Input: A sequence of sampled beat signals for one chirp.
- Operation: Compute the FFT of this sequence.
- Output: A spectrum where peaks indicate the presence of targets at specific ranges. The frequency of each peak corresponds to the beat frequency, which is then converted to range.
Beat Signal (Time Domain) -------------------- | . . . . . . . . .| -------------------- | v (FFT) Range Spectrum (Frequency Domain) -------------------- | . . . . . . . . .| <- Peaks at beat frequencies --------------------Doppler Processing (for moving targets):
- Input: The range bins (frequency spectrum) from multiple consecutive chirps.
- Operation: For each range bin, form a new sequence of beat frequencies across the chirps. Apply an FFT to this sequence.
- Output: A Doppler spectrum for each range bin. Peaks in this spectrum indicate the Doppler shift, which can be converted to velocity. This creates a Range-Doppler Map.
Beat Frequencies (across chirps for a single range bin) -------------------- | f_b1 | f_b2 | f_b3 | ... | -------------------- | v (FFT) Doppler Spectrum -------------------- | . . . . . . . . .| <- Peaks at Doppler frequencies --------------------
Data Representation:
- Raw I/Q Samples: Complex numbers (pairs of in-phase and quadrature samples) representing the digitized signal.
- Beat Signal Spectrum: Magnitude and phase values at different frequencies, corresponding to beat frequencies.
- Range-Doppler Map: A 2D matrix where axes are range and Doppler frequency, and cell values indicate the signal strength (detection probability).
4) Python Examples When Applicable
Python, with libraries like NumPy, SciPy, and Matplotlib, is excellent for theoretical analysis and simulation of FMCW radar signals.
FMCW Chirp Generation and Beat Frequency Simulation
This example simulates a single FMCW chirp and the resulting beat frequency from a stationary target.
import numpy as np
import matplotlib.pyplot as plt
# Radar parameters
carrier_frequency = 77e9 # 77 GHz
bandwidth = 4e9 # 4 GHz
chirp_duration = 40e-6 # 40 microseconds
sampling_rate = 2 * bandwidth # Nyquist rate for the beat signal
num_samples = int(chirp_duration * sampling_rate)
# Target parameters
target_distance = 50 # meters
target_velocity = 0 # m/s (stationary)
# Derived parameters
slope = bandwidth / chirp_duration # Hz/s
c = 3e8 # Speed of light
# Time vector for the chirp
t = np.linspace(0, chirp_duration, num_samples, endpoint=False)
# Transmitted signal (simplified, only frequency is relevant for beat freq)
# tx_signal = np.cos(2 * np.pi * (carrier_frequency * t + 0.5 * slope * t**2))
# Calculate beat frequency for a stationary target
time_delay = 2 * target_distance / c
beat_frequency_stationary = slope * time_delay
print(f"Simulated beat frequency for stationary target: {beat_frequency_stationary:.2f} Hz")
# Simulate the beat signal
# For a stationary target, the beat signal is a constant frequency tone
# at the calculated beat frequency.
beat_signal_stationary = np.cos(2 * np.pi * beat_frequency_stationary * t)
# --- Analysis ---
# Perform FFT on the simulated beat signal to find the beat frequency
fft_result = np.fft.fft(beat_signal_stationary)
frequencies = np.fft.fftfreq(num_samples, 1/sampling_rate)
# Find the peak frequency
# We only care about positive frequencies
positive_freq_indices = np.where(frequencies >= 0)
magnitude = np.abs(fft_result[positive_freq_indices])
positive_frequencies = frequencies[positive_freq_indices]
# Find the index of the maximum magnitude
peak_index = np.argmax(magnitude)
detected_beat_frequency = positive_frequencies[peak_index]
print(f"Detected beat frequency from FFT: {detected_beat_frequency:.2f} Hz")
# Calculate range from detected beat frequency
detected_range = (detected_beat_frequency * c) / (2 * slope)
print(f"Detected range: {detected_range:.2f} meters")
# Plotting
plt.figure(figsize=(12, 8))
plt.subplot(2, 1, 1)
plt.plot(t * 1e6, beat_signal_stationary)
plt.title('Simulated Beat Signal (Stationary Target)')
plt.xlabel('Time (µs)')
plt.ylabel('Amplitude')
plt.grid(True)
plt.subplot(2, 1, 2)
plt.plot(positive_frequencies, magnitude)
plt.title('FFT of Beat Signal (Range Spectrum)')
plt.xlabel('Frequency (Hz)')
plt.ylabel('Magnitude')
plt.xlim(0, beat_frequency_stationary * 2) # Limit x-axis for clarity
plt.grid(True)
plt.tight_layout()
plt.show()FMCW with Doppler Shift Simulation
This example adds a Doppler shift for a moving target.
import numpy as np
import matplotlib.pyplot as plt
# Radar parameters (same as before)
carrier_frequency = 77e9
bandwidth = 4e9
chirp_duration = 40e-6
sampling_rate = 2 * bandwidth # Needs to be high enough to capture beat freq + doppler
num_samples = int(chirp_duration * sampling_rate)
# Target parameters
target_distance = 50 # meters
target_velocity = 15 # m/s (approaching)
# Derived parameters
slope = bandwidth / chirp_duration # Hz/s
c = 3e8 # Speed of light
# Time vector for the chirp
t = np.linspace(0, chirp_duration, num_samples, endpoint=False)
# Calculate time delay and Doppler shift
time_delay = 2 * target_distance / c
doppler_shift = 2 * target_velocity * carrier_frequency / c
# Calculate beat frequency for a moving target
# For an approaching target, Doppler adds to the beat frequency
beat_frequency_moving = slope * time_delay + doppler_shift
print(f"Simulated Doppler shift: {doppler_shift:.2f} Hz")
print(f"Simulated beat frequency for moving target: {beat_frequency_moving:.2f} Hz")
# Simulate the beat signal for a moving target
# This is a tone at the combined beat and Doppler frequency
beat_signal_moving = np.cos(2 * np.pi * beat_frequency_moving * t)
# --- Analysis ---
# Perform FFT on the simulated beat signal
fft_result = np.fft.fft(beat_signal_moving)
frequencies = np.fft.fftfreq(num_samples, 1/sampling_rate)
positive_freq_indices = np.where(frequencies >= 0)
magnitude = np.abs(fft_result[positive_freq_indices])
positive_frequencies = frequencies[positive_freq_indices]
peak_index = np.argmax(magnitude)
detected_beat_frequency_with_doppler = positive_frequencies[peak_index]
print(f"Detected beat frequency (with Doppler) from FFT: {detected_beat_frequency_with_doppler:.2f} Hz")
# --- Range and Velocity Calculation (requires multiple chirps) ---
# For this single-chirp example, we can only see the combined frequency.
# To separate range and velocity, we need to analyze multiple chirps (e.g., up/down chirps)
# and perform a 2D FFT (Range-Doppler Map).
# For illustration, if we *knew* the Doppler shift, we could extract the range beat freq
# and calculate range.
# If we assumed this detected freq was *only* beat frequency (incorrect for moving target):
inferred_range_incorrect = (detected_beat_frequency_with_doppler * c) / (2 * slope)
print(f"Inferred range (assuming no Doppler): {inferred_range_incorrect:.2f} meters (Incorrect!)")
# Plotting
plt.figure(figsize=(12, 8))
plt.subplot(2, 1, 1)
plt.plot(t * 1e6, beat_signal_moving)
plt.title('Simulated Beat Signal (Moving Target)')
plt.xlabel('Time (µs)')
plt.ylabel('Amplitude')
plt.grid(True)
plt.subplot(2, 1, 2)
plt.plot(positive_frequencies, magnitude)
plt.title('FFT of Beat Signal (Combined Frequency)')
plt.xlabel('Frequency (Hz)')
plt.ylabel('Magnitude')
plt.xlim(0, beat_frequency_moving * 1.5) # Limit x-axis for clarity
plt.grid(True)
plt.tight_layout()
plt.show()5) GNU Radio Examples When Applicable
GNU Radio is powerful for real-time signal acquisition and processing. While a full automotive radar implementation in GNU Radio is complex and hardware-dependent, we can illustrate key signal processing blocks that would be used.
Conceptual GNU Radio Flowgraph for FMCW Analysis:
This flowgraph focuses on receiving a signal and performing an FFT to observe its frequency content, analogous to the range FFT step.
graph LR
A[SDR Source (e.g., USRP, RTL-SDR)] --> B{Quadrature Demod}
B --> C[Low Pass Filter]
C --> D[FFT Sink]
D --> E(Spectrum Plot)Explanation of Blocks:
- SDR Source: Configured to receive RF signals at the automotive radar band (e.g., 24 GHz or 77 GHz, if hardware supports it, or a simulated/downconverted signal). Crucially, it needs to be synchronized with the transmitted chirp.
- Quadrature Demod: If the source outputs RF, this block downconverts it to baseband (I/Q samples). If the source is already at baseband, this might be bypassed or used for further processing.
- Low Pass Filter: Essential for removing unwanted high-frequency components and aliasing after downconversion. Its cutoff frequency would be related to the maximum expected beat frequency.
- FFT Sink: This block takes the I/Q samples and performs a real-time FFT, displaying the magnitude spectrum. The number of FFT points determines the frequency resolution. For FMCW, this FFT would ideally be performed within the duration of a single chirp.
GNU Radio Companion (GRC) Snippet (Conceptual):
A simplified GRC flowgraph might look like this, focusing on the FFT analysis of a received signal. This assumes you have a way to capture a signal that resembles the beat frequency output of an FMCW radar.
<?xml version="1.0" encoding="utf-8"?>
<flow_graph>
<port_register>
<port name="in" type="complex"/>
</port_register>
<port_register>
<port name="out" type="complex"/>
</port_register>
<block type="grc.usrp_source" id="usrp_source_0">
<property name="freq" value="77e9"/> <!-- Example: 77 GHz -->
<property name="samp_rate" value="10e6"/> <!-- Example: 10 MSps -->
<property name="gain" value="30"/>
<property name="channels" value="[0]"/>
<property name="type" value="complex"/>
</block>
<block type="grc.low_pass_filter" id="low_pass_filter_0">
<property name="gain" value="1"/>
<property name="cutoff_freq" value="5e6"/> <!-- Example: Max beat freq ~5 MHz -->
<property name="decimation" value="1"/>
<property name="num_taps" value="101"/>
<property name="type" value="fir"/>
<property name="window" value="window.WIN_HAMMING"/>
</block>
<block type="grc.fft_sink_c" id="fft_sink_0">
<property name="fft_size" value="1024"/>
<property name="type" value="real"/>
<property name="average" value="0"/>
<property name="win_size" value="1024"/>
<property name="plot_title" value="FFT Spectrum (Range Analysis)"/>
</block>
<connection>
<source block="usrp_source_0" port="out"/>
<sink block="low_pass_filter_0" port="in"/>
</connection>
<connection>
<source block="low_pass_filter_0" port="out"/>
<sink block="fft_sink_0" port="fft_in"/>
</connection>
</flow_graph>Important Considerations for GNU Radio:
- Hardware Support: 77 GHz operation requires specialized SDR hardware (e.g., some high-end USRPs with daughterboards). For 24 GHz, more common hardware might be available. Often, analysis is done on recorded data or with downconverted signals.
- Synchronization: Precise timing synchronization between the transmitted chirp and the SDR receiver is paramount. This is a significant challenge in practical implementations.
- Real-time Processing: The high sampling rates and processing demands for automotive radar (especially for 4 GHz bandwidth) can push the limits of real-time processing on standard PCs. FPGAs are often used in actual automotive ECUs.
- Waveform Generation: A complete system would also require a GNU Radio flowgraph for generating the FMCW chirps, synchronized with the receiver. This is typically done using high-speed DACs.
6) Visual Examples When Applicable
Visualizing the signals and processing steps is crucial for understanding.
ASCII Signal Diagram: FMCW Chirp and Beat Frequency
This diagram conceptually shows the transmitted chirp frequency and the received signal's frequency after a time delay. The difference is the beat frequency.
Transmitted Signal Frequency (Tx)
^
| /----/----/----/
| / / / /
| / / / /
|/____/____/____/_________ Time
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
---
This chapter is educational, lab-oriented, and constrained to lawful, defensive, and controlled research contexts.