Patent Application
20230408671
The present invention is directed to radar systems, and more particularly to radar systems for vehicles and robotics.
The use of radar to determine direction, range, and velocity of objects in an environment is important in a number of applications including automotive radar, robotic sensing, and positioning. In practice, radars experience system impairments that degrade their performance.
A radar system consists of transmitters and receivers. The transmitters generate a baseband signal which is upconverted to a radio frequency (RF) signal that propagates according to an antenna pattern. The transmitted signal is reflected off of object or targets in the environment. The received signal at each receiver is the totality of the reflected signal from all targets in the environment. The receiver down converts the received signal to baseband and compares the baseband received signal to the baseband signal at one or more transmitters. This is used to determine the range, velocity, and angle of targets in the environment.
Outside of automotive radar applications, step frequency waveforms have been used by many applications, where such step frequency waveforms are simple frequency tones at each step. Sometimes a simple modulation, such as phase, is modulated on each step frequency. Furthermore, the frequency patterns to which the steps are ordered are also studied, where random pattern is one approach. Costas sequences and maximal-length sequences were also proposed in the past but each scheme has some disadvantages.
Methods and systems of the present invention provide for a radar to transmit and receive a specific waveform that exhibits robustness against external interference that may be unpredictable and at high power level. The transmission of such a waveform requires a specific frequency generation method and frequency patterns to satisfy those requirements. Moreover, the waveform is also produced using specific modulation methods to improve transmitter amplifier performance. At the receiver, the waveform enables receiver designs that are more robust against high power interference.
In an aspect of the present embodiment, an exemplary radar system produces a radar waveform that is used to step a frequency generator to generate a random permutation pattern in frequency sub-bands. Alternatively, a specific pattern could be generated that exhibits low collisions across sub-bands for causal shifts in that pattern.
In another aspect of the present invention, the radar system uses a radar waveform used to modulate a random differential to the phases between each step of the frequency generator. These random differential phases (or phase transitions) suppress auto-correlation sidelobes between delays of the frequency patterns or cross-correlation between multiple transmitted patterns in multiple antenna radar systems.
In a further aspect of the present invention, the radar waveform contains a sub-band modulation which exhibits continuous phase transmission using an arbitrary phase transition function with a desired instance of an exemplary linear frequency transition function. The continuous phase characteristic allows for power saturation of the transmitter amplifiers and less nonlinearities from the entire transmit signal path.
In yet another aspect of the present invention, an exemplary radar system includes filter bank receivers with both analog and digital components constituting the filter bank.
These and other objects, advantages, purposes and features of the present invention will become apparent upon review of the following specification in conjunction with the drawings.
The present invention will now be described with reference to the accompanying Figures, wherein numbered elements in the following written description correspond to like-numbered elements in the Figures. All figures don't draw time, frequency, or magnitudes to scale. Accurate interpretation of scale must be derived from the equations given in this invention. As discussed herein, the exemplary radar system includes a system and methods that provide for a radar system to transmit and receive a specific waveform that exhibits robustness against external interference that may be unpredictable and at high power level. The transmission of such a waveform requires a specific frequency generation method and frequency patterns to satisfy those requirements. Moreover, the waveform is also produced using specific modulation methods to improve transmitter amplifier performance. At the receiver, the waveform enables receiver designs that are more robust against high power interference.
There are several types of signals used in radar systems. One type of radar signal is known as a frequency modulated continuous waveform (FMCW). In this type of system, the transmitter of the radar system sends a continuous signal in which the frequency of the signal varies across some range of frequencies. This is sometimes called a chirp radar system. At the receiver a matched filter can be used to process the received signal. The output of the matched filter is a so-called “pulse-compressed” signal with a pulse duration inversely proportional to the bandwidth used in the chirp signal. Mixing (multiplying) the reflected wave from a target with a replica of the transmitted signal results in a CW signal with a frequency that represents the distance between the radar transmitter/receiver and the target. By sweeping up in frequency and then down in frequency the Doppler frequency can also be determined.
The transmitted signal from each radar transmitter consists of a baseband signal which is upconverted to an RF signal by an RF upconverter followed by an antenna. The received signal at each radar receiver antenna is downconverted by an RF downconverter to a complex baseband signal. The baseband transmitted radio signals and the reflected radio signals after downconversion in the receiver are provided to the processor. As an example, a baseband signal used for transmission might consist of repeated sequences of random or pseudo-random binary values for one transmitter, e.g., (−1, −1, −1, −1, 1, 1, 1, −1, 1, 1, −1, −1, 1, −1, 1), although any sequence, including non-binary sequences and non-periodic sequences could be used and different sequences would be used for different transmitters. The use of truly random number generators and pseudo random number generators to produce the values used to phase modulate the radio signal before transmission is described in detail in U.S. patent application Ser. No. 15/204,003, filed Jul. 7, 2016, which is hereby incorporated by reference herein in its entirety.
The exemplary radar sensing system may also utilize aspects of the radar systems described in U.S. provisional applications, Ser. No. 62/319,613, filed Apr. 7, 2016, Ser. No. 62/327,003, filed Apr. 25, 2016, Ser. No. 62/327,004, filed Apr. 25, 2016, Ser. No. 62/327,005, filed Apr. 25, 2016, Ser. No. 62/327,006, filed Apr. 25, 2016, Ser. No. 62/327,015, filed Apr. 25, 2016, Ser. No. 62/327,016, filed Apr. 25, 2016, Ser. No. 62/327,017, filed Apr. 25, 2016, Ser. No. 62/327,018, filed Apr. 25, 2016, Ser. No. 62/332,544, filed May 6, 2016, Ser. No. 62/336,966, filed May 16, 2016, Ser. No. 62/338,792, filed May 19, 2016, Ser. No. 62/816,941, filed Mar. 12, 2019, Ser. No. 63/167,347, filed Mar. 29, 2021, Ser. No. 63/140,567, filed Jan. 22, 2021, Ser. No. 63/194,267 filed May 28, 2021, which are all hereby incorporated by reference herein in their entireties.
Fast Frequency Hopping with Continuous Phase Modulation:
In automotive radar applications, one challenge is the need to have a short step duration of the transmission frequencies. This requirement is important to provide frequency diversity in the dynamic environment of interference and changing targets. Unfortunately, conventional techniques for step frequency waveforms are silent on how to deal with the speed of frequency steps. Hence this problem is related to issues in fast frequency hopping.
With fast hopping, or stepping, straightforward assumptions of ideal analog to digital conversion followed by digital signal processing of the waveform using FFT's is not necessarily a good solution. Again, conventional systems and methods do not propose new solutions here and always use the digital FFT assumption as the receiver solution. For automotive radar, alternate receiver architectures can help solve many application specific problems like multiple radar interference and power coupling from transmitter to receiver.
Additionally, the conventional techniques for step frequency are scarce on the issue of multiple antenna support by the waveform. While adding phase codes in addition to step frequency codes have been suggested before, they have not necessarily been applied to the multiple antenna requirement nor any consideration of transmitters. An important property in automotive radar's transmitter is to exhibit continuous phase such that amplifiers maintain power saturation. The challenge remains to modulate the phase codes to exhibit continuous phase for any given coding pattern, such as random coding. Currently, there are no known proposals to solve this problem.
The exemplary waveform will satisfy the following formula:
T
ping
Δf=1
Each code word contains Nb quanta. In this code word, each unique quanta is assigned an integer to represent its sub-band location offset by an integer multiple of Δf.
To generate a randomly permuted code of integers from 1 to Nb for a code word of Nb in length, a uniform distribution random generator is used that can be stepped Nb times. At the first step, the Nb integers are randomly drawn, each with 1/Nb likelihood, and then that integer is removed from all following draws. On the second step, there are only Nb−1 integers remaining. Each of those integers with 1/(Nb−1) likelihood are randomly drawn to remove the outcome from the remaining set. This procedure is continued until no integers remain from the drawing set.
To generate multiple code words for a radar scan duration, the previous procedure is executed to generate a randomly permuted code word multiple times. Every random code word is statistically independent from all other code words. For practical reasons, pseudo-random code words that's almost statistically independent are sufficient for implementation.
To generate low-collision code words, the following process steps are followed.
In a first step, a randomly permuted code word of length Nb is generated, as discussed herein. This first code word is code word C1.
For the next Nb−1, code words, each new code word is shifted one position with an anti-causal shift, which is defined as a left-ward shift, as opposed to a right-ward shift or a causal shift that has a positive time delay. Each shift is also a wrap-around shift where the tail end of the code word is shifted to the beginning position of that code word.
After generating Nb−1 code words, the code word is saved as final code word Cz. During these last two steps, if the total number of code words exceeded what's needed for this scan duration, the process stops generating code words. Otherwise, the process continues to the next step.
The process then generates a new word using random permutation. The process then compares the final code word Cz against this new word such that no code symbols are equal at each corresponding position. The process also performs the same symbol comparison for some known number. This number should be greater than or equal to 1 and much smaller than Nb, of anti-causal shifts of the new word against Cz and ensure there are no code symbols equal at each position. If the new word doesn't satisfy these checks, the process generates a new word and checks again. Once the conditions are satisfied with this new code word, the process returns to generate the next code words.
After FH code words are generated, the step frequency generator 502 in
e
j2πΔfnt
Where t is time variable, n is the quanta integer in the code word. Each integer n produces only Tping duration of that tone.
Multiplier 503 is a normal multiplier that modulates the tones from the step frequency generator 502 with a phase signal (from the phase transition function 506).
Starting from the random phase code generator 504, a random, or practically pseudo-random, binary sequence is output. This sequence determines the direction of phase transitions, of either +/−0, between each quanta. An example of θo, is π/2.
Following the phase code generator 504 is a configurable delay 505, a configurable delay that adjusts a start timing between the phase transitions relative to the start timing of the step frequency generator 502. Since both phase and frequency generators are timed to the quanta time units, the absolute time dimension of these generators is pegged to the quanta time line. The frequency generators will start at the start of each quanta and the phase generator may adjust its start time relative to that.
Next is the phase transition function 506. This function 506 acts like a smoothing filter such that the modulated phases are continuous in time. Unlike a filter, the phase transition function's output is defined as follows:
Output phase(t′)=θ0(nTping)+θ(t′)
Where n is an integer that counts the quanta time units, t′ is time units normalized to each quanta start t′=t−nTping with t being absolute time. So the
Where K is a normalization factor such that at θ(Tping)=1. fθ(τ) is called the phase-change density function.
In the example plot 602 illustrated in
Another specification for phase-change density function that is especially important for efficient implementation is fθ(t)=τ.
The waveform receiver processor 705 performs the waveform processing that will be described further and illustrated in
An exemplary strong return estimator 723 and a squelch control 724 are not mandatory to have a working filter bank receiver 722. However, the squelch control 724 and the strong return estimator 723 are needed to improve the performance of the filter bank receiver 722 in presence of strong input signals, either from strong reflections or power coupling from radar transmitter to radar receiver.
The down-converted signal (received at the input 801) is fed into a bank (the repeated instances are indicated by 807) of analog mixers 802. Each mixer 802 is a demodulating multiplier with its coefficients chosen to yield demodulation of the subband carriers of the waveform quanta. Because these coefficients are dependent on digital mixers 805, they will be described shortly.
Block 803 is an analog low pass filter to remove out-of-band signals. The bandwidth of this filter 803 is determined by the number of quanta being demodulated by this particular bank.
After the low pass filter 803, an exemplary analog-to-digital converter (ADC) 804 performs analog to digital conversion. Another bank of digital mixers 805 follows the ADC 804. The coefficients of these digital mixers 805 are chosen such that each bank exactly demodulates a specific subband to baseband. The product fmfmn, is chosen such that a specific sub-band shown in
After each digital mixer 805, there is a digital filter 806 that serve to low pass filter the subband signal. Cascaded integrator—comb (CIC) designs are efficient digital low pass filter implementations for this application.
In this embodiment of the analysis stage, there is no squelch switch or signals from the squelch controller 724. The use of a squelch controller 724 is described in
Following the bank outputs from the analysis stage is a bank of synthesis filters. Each synthesis path consists of an up-sampler 810, a CIC low-pass filter 811, and a digital mixer 815. Combining the bank is a special summation function 812.
The up-sampler 810, increases the digital sampling rate of its input by inserting zero samples to the data stream. The resulting sampling rate is equal to the total bandwidth of the waveform that was present at input 801. After up-sampling, the signal images are filtered by the CIC low-pass filters 811. The mixers 815 up-convert that sub-band to the intended sub-band frequency.
The up convert frequency fk for each synthesis path is determined by the connection between the first section 809 and the second section 813. Suppose analysis path fmfmn demodulates sub-band nΔf, then fk=nΔf for that connected synthesis path.
Because the expected code word is known by the receiver from the frequency code generator, and the desired range delay to correlate with, the combiner 812 is able to produce one stream of data samples that represents the reconstruction of the waveform at a desired range delay.
During continuous radar transmission, close range reflectors such as radome or fascia, and coupling from radar transmitter to receiver can produce very strong signals that can corrupt analog components and distort signal quality. As such, the invention includes a technique called “Strong Return Squelching.”
In practice, the preferred realization is to place the switch 902 as close to the input path as possible, may be before the LPF 904. When the switch 902 is open, strong signals that may saturate the ADC are no longer allowed to pass through. Furthermore, signal coupling between the analog components in the banks of the analysis stage are also reduced.
Squelch control 903 turns on the coupled squelch switch 902 by calculating the timing and the frequency of each sub-band. For timing, squelch control 903 can get that from either calibrated delay of coupling or radome reflection, or the Strong Return Estimator or both. The Strong Return Estimator is described later.
Once timing is known, squelch control 903 must calculate the sub-band location as follows:
In one exemplary embodiment, the strong return timing is provided as tn, where n is the sub-band from the frequency hop code at time t. Squelch control 903 would activate the squelch switch 902 for a sub-band m where |m−n|<squelch delta-subband threshold. The squelch delta-subband threshold is an integer greater than or equal to 0 that's configured by the radar.
The Strong Return Estimator starts by correlating one PRI worth of synthesized signal from the filter banks.
A
v(τ)=∫t
Where ti is start of ith PRI, zq(t) is the complex output values from the filter bank receiver q, cp*(t) is the complex conjugate of the transmitted waveform by transmitter p. p,q indices together form the multi-antenna radar system which maps to a single index v which is called the virtual receiver (VRX) index. τ is the desired range bin delay for the correlation output and it's swept for each range bin within the given PRI. The mapping of v from p,q depends on the antenna design and isn't of direct consequence to Strong Return Estimator.
The Strong Return Estimator can use any one of the VRX as long as the average power per PRI of Av(τ) is greater than a threshold.
To detect the strong target, the noise floor must be estimated across the PRI for the chosen VRX, v.
Noise Floor(v)=Median(Av(τ)Av*(τ))
Where the Median( )function can be, for example, either the conventional median or the Hodges-Lehmann median technique. Av*(t) is the complex conjugate of Av(τ).
The Strong Return Estimator then outputs the strong return timing tn for any τ where:
A
v(τ)Av*(t)>KsNoise Floor(v).
KS is the threshold for detecting strong returns in the Strong Return Estimator.
Thus, the exemplary embodiments discussed herein, include an exemplary radar system that provides for a radar system to transmit and receive a desired waveform that exhibits robustness against external interference that may be unpredictable and at high power levels. The transmission of such an exemplary waveform can require a specific frequency generation method and frequency patterns to satisfy those requirements. Moreover, the waveform is also produced using specific modulation methods to improve transmitter amplifier performance. At the receiver, the waveform enables receiver designs that are more robust against high power interference. The exemplary radar system also produces a radar waveform that is used to step a frequency generator to produce a random permutation pattern in frequency sub-bands. Alternatively, a specific pattern could be generated that exhibits low collisions across sub-bands for causal shifts in that pattern. The radar waveform could also be used to modulate a random differential to the phases between each step of the frequency generator to suppress auto-correction sidelobes between delays of the frequency patterns or cross-correlation between multiple transmitted patterns in multiple antenna radar systems. The radar waveform may also contain a sub-band modulation exhibiting continuous phase transmission using an arbitrary phase transition function with a desired instance of an exemplary linear frequency transition function and allowing for power saturation of the transmitter amplifiers and less nonlinearities from the entire transmit signal path. Lastly, the radar system may include filter bank receivers with both analog and digital components constituting the filter bank.
Changes and modifications in the specifically described embodiments can be carried out without departing from the principles of the present invention, which is intended to be limited only by the scope of the appended claims, as interpreted according to the principles of patent law including the doctrine of equivalents.
The present application claims the filing benefits of U.S. provisional application, Ser. No. 63/353,044, filed Jun. 17, 2022, which is hereby incorporated by reference herein in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 63353044 | Jun 2022 | US |
