Spread spectrum receiver and test method

Abstract

The invention relates to a spread spectrum receiver and a test method. A method of characterizing a LoRa modulated signal or any such signal having multiple chirps as symbols. It foresees: sampling and storing the signal, determining the phase of at least one chirp in the signal, and determining a timing error and / or a frequency error based on the phase, the timing error being extracted by the height of the discontinuity jump in the phase at the cyclic shift position, and the frequency error being obtained by the slope of the phase. The method can be applied to a dedicated receiver for characterizing a LoRa transmitter.

Description

Technical Field

[0001] This invention relates to wireless network systems based on chirp-modulated signals. Specifically, it relates to advanced receivers for such signals and methods for testing and characterizing chirp-modulated transmitters. Background Technology

[0002] Recently, wirelessly connected devices have become the focus of considerable interest and effort. Improved wireless communication technologies contribute to the creation and development of the "Internet of Things" (IoT). In this context, several wireless communication protocols have been proposed and used. LoRa, in particular, is known through patents EP2449690B1, EP2763321B1, EP2767847B1, EP3247046B1, and EP2449690B1. TM The communication system uses linear frequency modulation spread spectrum modulation to achieve long-distance transmission with low complexity and low power consumption.

[0003] In the context of this disclosure, for the sake of brevity, the term "LoRa" refers to a communication system based on the exchange of radio signals comprising multiple frequency linear modulations (FMs), each FM limited to a finite time interval and a finite bandwidth. The FMs include a basic FM and a modulated FM, in which the frequency follows a given function from the beginning to the end of the time interval, and the modulated FM is a cyclic shift of the basic FM. The basic FM and the modulated FM are considered as symbols in the modulation alphabet. This definition includes known LoRa signals. TM Products and standards, as well as possible but not yet implemented variations of this broader concept.

[0004] LoRa TM LoRa modulation is used in many applications and devices, from low-power wide-area networks to long-distance point-to-point communication. Different vendors support the core technologies in chipsets, modules, and reference designs. TM The proliferation of modulation techniques necessitates the standardization of the performance of transmitters and receivers from different sources and models to ensure interoperability.

[0005] Through LoRa TM Routine connectivity tests are performed on various combinations of transmitters and receivers to test interoperability. However, this approach becomes impractical in the long run due to the vast number of possible combinations as the number of available devices increases.

[0006] For example, in other digital communication technologies such as Wi-Fi, performance is quantified using Error Vector Magnitude (EVM), which is a measure of how far the transmit constellation point deviates from its ideal position. To compare with the signal-to-noise ratio (SNR), EVM can be expressed as a percentage of the ideal symbol amplitude, or in dB. The main metrics defining LoRa transmitter performance are the stability of timing and frequency over time, as frames can be very long.

[0007] Conventional LoRa receivers provide a comprehensive estimate of time and frequency errors, rather than attempting to estimate these parameters independently. The object of this invention is to provide a method for testing and characterizing transmitters and receivers of LoRa signals, and for independently measuring their timing and frequency stability. This invention also relates to advanced receiver architectures for LoRa signals. Summary of the Invention

[0008] The purpose of this invention is to provide a method for characterizing the consistency of LoRa signals or LoRa radio transmitting or receiving devices that overcomes the shortcomings and limitations of the prior art.

[0009] Another objective of this invention is to quantify radio impairments in a particular device design. These impairments may arise from phase noise, frequency instability or drift, or nonlinearity. This method provides indicators that can identify the primary sources of transmission errors.

[0010] The above objective is achieved by the subject matter of the appended claims, particularly by a method characterizing a modulated signal comprising a plurality of linear frequency modulateds, each linear frequency modulated being limited to a finite time interval and a finite bandwidth, wherein the linear frequency modulation comprises a basic linear frequency modulation and a modulated linear frequency modulation, in which the frequency follows a monotonic function from the beginning to the end of the time interval, and the modulated linear frequency modulation is a cyclic shift of the basic linear frequency modulation, the method comprising: sampling and storing the signal, determining the phase of at least one linear frequency modulation in the signal, and determining a timing error and / or a frequency error based on the phase, the timing error being the deviation between the timing of the linear frequency modulation and a nominal predetermined timing, and the frequency error being the deviation between the frequency of the linear frequency modulation and a nominal predetermined frequency.

[0011] The dependent claims relate to features that may be useful or advantageous but are not essential to the invention. They particularly introduce a signal structure having a preamble of continuous basic linear frequency modulation (LFM), followed by a payload modulated by LFM, and also introduce various methods for obtaining timing and frequency errors from the phase, as well as an iterative process in which the same message can be repeatedly processed in the invention, for example, first determining the frame structure and then recalculating the time and frequency errors of all or some symbols. When analyzing several LFMs, the method may include determining the transmitter's sampling timing drift. It is also advantageous to quantify the transmitter's frequency stability by the ratio between the transmitter's sampling timing drift and the frequency offset of each received LFM, which ideally should yield the center transmission frequency.

[0012] One variation involves synthesizing a recombined modulated linear frequency modulated (LFM) by applying a window to two consecutive basic LFMs, replacing the received LFM. This allows for the determination of timing and / or frequency errors based on phase errors, whereby the phase error is the difference between the phase of at least one LFM and a known phase of the nominal LFM. The timing error can be determined as proportional to the height of the step change in the phase error (corresponding to a cyclic shift), and the frequency error can be determined as proportional to the slope of the phase error.

[0013] Preferably, determining the phase error includes the step of de-linear frequency modulation (LFM), comprising: multiplying the sampled signal sample by sample, by a complex conjugate vector representing the basic LFM, applying a Fourier transform to the product, and detecting two peaks in the result of the Fourier transform with a frequency interval equal to the bandwidth. Preferably, the received LFM is sampled above the Nyquist limit, with an oversampling factor of at least 2 between the sampling frequency and the signal bandwidth to resolve the peaks, but this is not strictly required.

[0014] The method of this invention can be applied to the characterization of LoRa transmitters and receivers, in the latter case, by characterizing the LoRa receiver using a "golden" transmitter whose inconsistencies are negligible. Preferably, the received signal is digitized only in the receiver, and samples are sent to a computing server for determining time and frequency deviations. Attached Figure Description

[0015] Exemplary embodiments of the present invention are disclosed in the specification and illustrated in the accompanying drawings, wherein:

[0016] Figure 1 The structure of a radio modem according to one aspect of the present invention is shown in a simplified schematic manner.

[0017] Figure 2aThe instantaneous frequencies of basic linear frequency modulation and modulated linear frequency modulation according to one aspect of the present invention are shown. Figure 2b The phase of the same signal is shown, and Figure 2c The real and complex components of basic linear frequency modulation and modulated linear frequency modulation are plotted in the time domain and baseband representations.

[0018] Figure 3a and 3b The instantaneous frequency of basic linear frequency modulation (LFM) was compared with that of modulated linear frequency modulation (MCFM).

[0019] Figure 4 This is a plot showing the phase error of a LoRa signal that is inconsistent in timing and frequency.

[0020] Figure 5 A system for characterizing LoRa transmitters and / or receivers is illustrated schematically. Detailed Implementation

[0021] Several aspects of the linear frequency modulation technique used in this invention are described in European Patent EP2449690B 1, which is incorporated herein by reference and will be briefly reviewed herein. Figure 1 The radio transceiver schematically shown is a possible embodiment of the present invention. The transceiver includes a baseband section 200 and an radio frequency (RF) section 100. It includes a baseband modulator 150 that generates a baseband complex signal based on digital data 152 at its input. This baseband complex signal is then converted to a desired transmission frequency by the RF section 100, amplified by a power amplifier 120, and transmitted via an antenna through an RF switch 102.

[0022] Once the signal is received at the other end of the radio link, it is transmitted... Figure 1 The receiving section of the transceiver processes the signal, which includes a low-noise amplifier 160, followed by a down-conversion stage 170. The down-conversion stage 170 generates a baseband signal consisting of a series of linearly frequency-modulated signals (which are also complex signals, for example, represented by two components I and Q). The baseband signal is then processed by a baseband processor 180, which functions in the opposite way to the modulator 150 and provides a reconstructed digital signal 182.

[0023] As described in EP2449690, the signal to be processed comprises a series of linear frequency modulations whose frequencies vary from an initial instantaneous value f0 to a final instantaneous frequency f1 over predetermined time intervals. For simplicity, it will be assumed that all linear frequency modulations have the same duration T, although this is not an absolute requirement of the invention.

[0024] Linear frequency modulation (LFM) in a baseband signal can be described by its instantaneous frequency profile f(t), or by the signal phase defined as a function φ(t) of time. Importantly, the processor 180 is arranged to process and identify LFMs with multiple different profiles, each profile corresponding to a symbol in a predetermined modulation alphabet.

[0025] According to an important feature of the present invention, the received signal Rx may include a basic linear frequency modulation (hereinafter also referred to as unmodulated linear frequency modulation) or one of a set of possible modulated linear frequencies having a specific and predefined frequency profile, which is obtained from the basic linear frequency modulation by cyclically time-shifting the basic frequency profile. As an example, Figure 2a and 2b The diagram shows the possible frequency and phase curves of a basic linear frequency modulation (LFM) 30 and a modulated linear frequency modulation (MCM) 32 between the start time t = t0 and the end time t = t1 of the LFM. Figure 2c The corresponding baseband signal is shown in the time domain. For example, the horizontal scales correspond to symbols; although these graphs are plotted as continuous, in practice they actually represent a finite number of discrete samples. As for the vertical scales, they are normalized to the target bandwidth or the corresponding phase span. Figure 2b In this context, the phase is represented as if it were an unbounded variable, but in practice, it can actually span several cycles.

[0026] In the depicted example, the frequency of the basic linear frequency modulation (LFM) increases linearly from an initial value of -BW / 2 to a final value of BW / 2, where BW represents bandwidth expansion. However, reduced LFM or other LFM curves are also possible. Therefore, information is encoded in the form of LFM with one of a plurality of possible cyclic shifts relative to a predetermined basic LFM, each cyclic shift corresponding to a possible modulation symbol. In other words, processor 180 needs to process signals comprising multiple frequency LFMs that are cyclically time-shifted copies of the basic LFM curve, and processor 180 needs to extract messages encoded in a series of said time shifts.

[0027] The signal may also include conjugate linear frequency modulation, which is the complex conjugate of the basic unmodulated linear frequency modulation. These can be viewed as down-linear frequency modulation, where the frequency decreases from BW / 2 to -BW / 2.

[0028] The operation of assessing the time shift of the received linear frequency modulation (LFM) relative to the local time base can be referred to hereinafter as "dechirping," and advantageously can be achieved through a despreading step, which involves multiplying the received LFM sample-by-sample by the complex conjugate of the locally generated basic LFM. This produces an oscillating digital signal whose dominant frequency can be expressed as proportional to the cyclic shift of the received LFM. Demodulation can then involve the Fourier transform of the despread signal. The location of the Fourier maximum is a measure of the cyclic shift and the modulation value. In mathematical terms, this is represented by... This represents the k-th received symbol, and the corresponding modulation value is given by m(k) = argmax. n (|X(k,n)|) is given, where express Conjugate with basic linear frequency modulation The Fourier transform of the product. However, other methods are also possible for demodulating the signal and extracting the cyclic shift of each symbol.

[0029] Normal LoRa demodulation does not require explicit extraction. Figure 2b The phase values ​​in the plot. This invention relates to an advantageous method for analyzing LoRa signals using phase information. For this purpose, known techniques can be used to analyze the phase values ​​from the received signal. A time series in which the phase is calculated as a digital value.

[0030] In a variant of the invention, the received linear frequency modulated LoRa signal can be characterized using a special receiver configured to receive radio signals and process them through the following operations / steps.

[0031] As disclosed in EP2449690 and EP2763321, including the same preferred unmodulated linear frequency modulated preamble is advantageous for detection. The receiving device applies the above-described delinear frequency modulation process and searches for peaks in the FFT spectrum above the noise level. The detection of the peak tells the receiver that a LoRa signal has been received, and the position of the peak indicates a timing and frequency offset between the transmitter and receiver systems. To improve sensitivity, the receiver can be configured to add several consecutive identical linear frequency modulated FFT outputs of the preamble in the complex domain.

[0032] As disclosed in EP2763321, the receiver can be configured to improve time and frequency synchronization by examining other special characteristics of the preamble, such as symbols with predetermined cyclic shift values ​​and conjugate (falling) symbols. The receiver can also be configured to improve the synchronization figure by comparing FFT peaks generated by different symbols spaced apart in time.

[0033] Within the framework of this invention, the receiver is configured to determine timing errors using methods disclosed in the cited references, and to track timing and / or frequency errors along the data frame or at least along the preamble, and to track them using a suitable tracking algorithm. Characterization can be further improved by applying a systematic offset to the synthesized linear frequency modulation based on the estimated crystal error.

[0034] According to one aspect of the invention, the receiver is configured to analyze the received data frames as described above, and then repeat the analysis on the same series of digital data to determine at least one linear frequency modulated (LFM) phase in the received signal, but preferably several LFM phases, to improve reliability.

[0035] Preferably, in this invention, the received linear frequency modulation (LFM) digital representation is oversampled. Normal LoRa receivers typically use a sampling frequency equal to the input LFM bandwidth BW, while the receiver used in this invention samples the input LFM at a frequency at least twice that bandwidth.

[0036] Figure 3a The instantaneous frequencies of modulated linear frequency modulation (LFM) 32 and basic LFM 30 are shown. The modulated signal can be divided into two continuous parts: a first part before discontinuity 39, whose frequency is a constant displacement Δf higher than the frequency of the reference LFM 30; and a second part after discontinuity, whose frequency is a displacement equal to BW-Δf lower than the frequency of the reference LFM 30. During the delinear frequency modulation process, the conjugate with the basic LFM... Multiplying them together will give ( Figure 3b A signal has a constant positive frequency Δf, followed by a signal with a constant negative frequency Δf-BW. If the signals are sampled at a frequency equal to BW (the Nyquist frequency), these positive and negative components will alias into a single peak. Oversampling can separate the positive and negative frequency peaks.

[0037] The receiver of the present invention is preferably configured to find paired peaks in the FFT derived from the processing of the oversampled signal. As mentioned above, the peak positions are interdependent because f2 = f1 - BW must be maintained. Therefore, the receiver can be configured to (incoherently) sum the contents of the corresponding bins in the positive and negative frequencies.

[0038] Incoherence can be improved by using the fact that the relative amplitudes of the two peaks are a priori determined by the modulation value in a predictable manner. A symbol with a cyclic shift close to N / 2 will have a discontinuity 39 near the middle of the symbol, so the FFT will present two peaks with approximately the same amplitude, while a symbol with a cyclic shift close to the extremes 0 and N will show a large peak and a much smaller peak, the amplitude of which is proportional to the length of the segment before and after the discontinuity.

[0039] Importantly, in this second analysis step, the phase of the received signal and its time unfolding are determined and compared with the nominal phase function to provide phase error. The phase of the received signal may include an unknown constant offset, which can be set to any value without consequence, as will be seen below.

[0040] Figure 4 These are plots showing the expected phase error of a transmitter with timing errors (Plot 61) and the expected phase error of a transmitter with both timing and frequency errors (Plot 62). The expected phase error of a perfectly aligned transmitter would be constant and is not plotted. Noise is ignored in this plot. This is an acceptable simplification because in most test procedures, the transmitter and receiver are very close together and the signal-to-noise ratio is high.

[0041] Figure 61 illustrates an example where, for a spreading factor N = 1024, the analyzed linear frequency modulation has a modulation of 512, thus the discontinuity is in the middle of the linear frequency modulation and has a timing error equal to 1 / 8 of the sample. In the continuous representation, the received symbol is represented by BW-Δf, and the transmitted symbol by Δf, where Δf-BW is the start of the symbol, which would be BW if there were neither timing nor frequency errors. If a timing error R(t0+t) exists, then T(t0+t) will be obtained after sampling, where t0 is the sample index.

[0042] If we consider the sample R(t0+t) = T(t0+t) exactly before the discontinuity and the sample Δt exactly after the discontinuity, then the instantaneous frequency, which is the time derivative of the phase, will be very close to R(t0+k / BW) = T(t0+Δt+k / BW) for the former, and very close to k for the latter. Therefore, we can write:

[0043] as well as

[0044]

[0045] This means that the sampling error Δt introduces phase errors of opposite sign before and after the frequency discontinuity. The phase error in... Figure 4 The frequency transition moment shown by Δ1 has a step, and the height of the step is directly proportional to the sampling error Δt. Because of this, the receiver of the present invention can obtain an independent estimate of the sampling error.

[0046] The frequency error introduces a phase error that is linearly related to time. This can be seen in Figure 62, which shows the sign of the cyclic shift of 896 for a spreading factor of 1024, the timing error of 1 / 16 of the sample, and the frequency error. The timing error gives a step error Δ2. The frequency error is directly proportional to the difference Δ3 between the phase errors of the first and last samples. Thus, the receiver of the present invention can obtain an independent estimate of the frequency error from the phase error.

[0047] Figure 5 This illustrates a possible implementation of the invention in a system used to characterize transmitter 350. The transmitter is controlled by workstation 385 and sends LoRa packets to receiver 360, which is configured to perform the aforementioned signal processing and determine whether the received signal conforms to various specifications, including frequency error and timing error. The frequency error and timing error are determined based on the aforementioned phase error function; for example, the frequency error is proportional to the difference in phase error between the start and end of the symbol error, and the timing error is proportional to the step of the phase error at the frequency transition position.

[0048] Data processing in receiver 360 is performed automatically in server 380, which uses a set of programming scripts 382 to capture digital baseband I and Q signals from receiver 360. The results can be securely transmitted to the workstation via WAN 384, which can be the Internet.

[0049] Preferably, the receiver is configured to automatically compensate for frequency and phase errors, as well as timing offsets and sampling timing drift in the captured signal. Sampling and sampling drift compensation is preferably calculated using time interpolation. The receiver can be configured to calculate transmitter accuracy metrics after the compensation has been applied to the captured signal.

[0050] The receiver of this invention can independently determine timing and frequency errors in a number of consecutively received LoRa symbols (i.e., linear frequency modulation) from a given transmitter, or even all LoRa symbols received from a given transmitter. Preferably, the receiver is configured to track the drift of these errors over time. Variations in timing errors in the captured data produce sample timing drift. For any symbol in the captured data, the ratio of sample timing drift to frequency offset is calculated. Since sample timing and frequency synthesis are typically driven by the same frequency reference in the transmitter, it is expected that sample timing drift and frequency error should be correlated. Therefore, the frequency offset should be equal to the center transmit frequency multiplied by the sample timing drift, and the aforementioned ratio should give the center transmit frequency. The receiver of this invention can be configured to track variations in the ratio around this expected value, and these will give a measure of frequency consistency in the transmitter.

[0051] Frequency reference drift is a measure of the change in frequency reference over captured data. It comprises two metrics: one calculated from the estimated frequency offset change and the other from the estimated sample timing drift.

[0052] Modulator bandwidth loss is a measure of transmitter accuracy loss caused by the limited modulator bandwidth. It is measured from the demodulated values ​​after frequency and time compensation. The received signal is compared to an ideal transmitted signal with the same modulation sequence, retaining only the samples corresponding to frequency jumps.

[0053] The disclosed system can be used to characterize the transmitter's performance against a high-quality calibrated receiver, or to characterize the receiver using a calibrated "gold" transmitter.

[0054] The reference numerals in the figure

[0055] 30 Basic Linear Frequency Modulation

[0056] 32 Modulation Linear Frequency Modulation

[0057] 39 Frequency jump

[0058] 61. Phase error with timing error

[0059] 62. Phase error with timing and frequency errors

[0060] 100RF section

[0061] 102RF switch

[0062] 110 frequency conversion

[0063] 120 power amplifier

[0064] 129 oscillator, time base

[0065] 150 modulator

[0066] 152 digital signals to be transmitted

[0067] 154 buffer

[0068] 160LNA

[0069] 170 downconverter

[0070] 180 processor, demodulator

[0071] 182 reconstructed digital signal

[0072] 190 controlled oscillator

[0073] 200 baseband section

[0074] 350 launcher

[0075] 385 workstation

[0076] 360 receiver

[0077] 380 server

[0078] 381 computing units

[0079] 382 script

[0080] 384 Internet

Claims

1. A method for characterizing a modulated signal comprising a plurality of linear frequency modulations, each linear frequency modulated being limited to a finite time interval and a finite bandwidth, wherein the linear frequency modulations include a basic linear frequency modulation and a modulating linear frequency modulation, wherein the frequency in the basic linear frequency modulation follows a monotonic function from the beginning to the end of the time interval, and the modulating linear frequency modulation is a cyclic shift of the basic linear frequency modulation, the method comprising: The method involves sampling and storing a signal, determining the phase of at least one linear frequency modulated (LFM) in the signal, and determining a timing error and / or a frequency error based on the phase, wherein the timing error is the deviation between the timing of the LFM and a nominal predetermined timing, and the frequency error is the deviation between the frequency of the LFM and a nominal predetermined frequency, characterized in that the at least one LFM is a modulated LFM or a recombined modulated LFM obtained by applying a window to two consecutive basic LFMs, and the timing error and / or frequency error are determined based on the phase error, which is the difference between the phase of the at least one LFM and a known phase of the nominal LFM.

2. The method according to claim 1, wherein the signal comprises a preamble of continuous basic linear frequency modulation, followed by a payload of modulated linear frequency modulation.

3. The method according to claim 1, wherein the timing error is determined to be proportional to the height of the step change in the phase error.

4. The method of claim 1, wherein the frequency error is determined to be proportional to the slope of the phase error.

5. The method according to claim 1, wherein, The sampling frequency has at least twice the bandwidth. The method includes: multiplying the sampled signal sample by a complex conjugate vector representing a basic linear frequency modulation sample on a sample-by-sample basis, applying a Fourier transform to the product, and detecting two peaks in the result of the Fourier transform with a frequency interval equal to the bandwidth.

6. The method according to claim 1, wherein, The signal is processed repeatedly.

7. The method according to claim 1, comprising: The signal is transmitted by a radio transmitter, received and sampled by a radio receiver, and then transmitted to a computing server to determine the time and frequency deviation.

8. The method according to claim 1, comprising: Determine the timing errors of multiple linear frequency modulations received from a given transmitter, and determine the sampling timing drift of the transmitter based on the timing errors.

9. The method of claim 8, comprising: For each received linear frequency modulation (LFM), the ratio of the transmitter's sampling timing drift to the frequency offset of the received LFM is determined, and the ratio is used as a measure of frequency stability in the transmitter.

10. A method for characterizing the performance of a radio transmitting device, comprising: The signal generated by the radio transmitter is received by a reference radio receiver, and the timing error and / or frequency error is determined by the method according to any one of claims 1 to 9.

11. A method for characterizing the performance of a radio receiving device, comprising: A signal is generated using a reference radio transmitter, the signal is received using the radio receiver, and a timing error and / or frequency error is determined by the method according to any one of claims 1 to 10.

12. A radio receiving device for LoRa signals, configured to perform the method according to any one of claims 1 to 9.

Images