JPH0821963B2 - Frequency detector - Google Patents

Description

The present invention relates to a frequency detecting device, and particularly demodulates a frequency signal having a poor C / N, for example, a PSK (phase shift keying) signal. It is suitable when applied.

[Conventional technology]

Although various methods are used as a method of detecting a frequency when demodulating a frequency signal that has been conventionally transmitted,
In principle, if the C / N of the received frequency signal is bad, there are many cases where the frequency cannot be detected.

However, FFT (fast Fourier transform) or DFT (di
The screte Fourier transform) method can be transformed into a frequency domain representation by obtaining the Fourier transform of the waveform from a time-series signal obtained by sampling a continuous waveform at regular intervals, and thus performing frequency spectrum analysis. In principle, even if the C / N of the input signal is bad, the frequency of the input frequency signal can be detected by narrowing the frequency resolution.

[Problems to be solved by the invention]

However, this method requires a considerably large amount of hardware or software to analyze the frequency spectrum, and has a problem that a frequency detection device having a practically simple structure cannot be obtained.

The present invention has been made in consideration of the above points, by converting the input frequency signal component carried by the carrier signal into a signal that is not affected by the noise component, C / Of course, when N is 0 [dB] or higher, C / N is 0 [d
B] An attempt is made to propose a frequency detection device capable of surely detecting an input frequency signal even in the following cases.

[Means for solving problems]

In order to solve such a problem, in the present invention, the first complex signal I representing the input frequency signal RVS at the time point t
_1K , Q _1K and the input frequency signal RVS at the time point when the first complex number signals I _1K , Q _1K differ by a predetermined time Δt
The second complex signal I _{1 (K-1)} , Q _{1 (K-1),} which represents the conjugate complex number of
And a signal processing means 10 for taking an average value for each time Δt based on a multiplication output of the multiplication means 5, and an input based on a processing output F _D of the signal processing means 10. The frequency of the frequency signal Δω is detected.

[Action]

The input signal RVS is converted into a signal by being multiplied by a conjugate complex number in the multiplication means 5. This converted signal is
When the signal component γ _ss including the input frequency signal Δω and the signal components γ _nn , γ _sn and γ _ns including the noise component are included and the C / N of the input signal is 0 [dB] or less, By taking an average value for each time Δt based on the converted multiplication output, the signal component of the average value output becomes the input frequency signal Δ
It includes a signal component ▲ containing ω and signal components ▲ ▼, ▲ ▼ and ▲ ▼ containing a noise component. Of the average value of each of these signal components, the signal component including the noise component becomes substantially zero, and as a result, the input frequency signal Δω
Can be detected without being affected by noise components.

〔Example〕

An embodiment of a frequency detecting device according to the present invention will be described in detail below with reference to the drawings.

The frequency detection device according to the present invention converts a frequency input signal into a signal that is not affected by noise components according to the following principle. First, the frequency input signal St is represented by a vector as in the following equation S _t = Ae ^{j ((ωc + Δω) t + φ0)} + n _t e ^{j (ωct + φt)} (1). Where ω _c is the center frequency of the carrier signal component, Δω is the frequency offset, Δω is the input frequency signal carried by the carrier signal, φ ₀ is the initial phase, and φ _t is the phase of the random noise component. (This phase is uniformly distributed), A is the amplitude of the carrier signal component, and n _t is the amplitude of the noise component (this is the ray distribution).

The received signal transmitted in this manner arrives at the time t in a state where the signal component represented by the first term is mixed with the noise component represented by the second term of the equation (1). Therefore, a time point that differs from the time point t by Δt, for example, the previous time point t
The received signal S _{t-t at} −Δt is S _t-Δt = Ae ^{j ((ωc + Δω) (t−Δt) + φ0)} + n _{t −Δt} e ^{j (ωc (t−Δt) + φt−Δt)}
…… (2) can be expressed. Here, n _t-Δt is the amplitude of the noise component, and φ _t-Δt is the phase of the noise component.

The frequency detection device according to the present invention is provided by the equations (1) and (2).
The conversion processing described below is performed on the received signal based on the equation. That is, the conjugate complex number S _t-Δt * of the second signal component represented by the equation (2) is multiplied by the first received signal represented by the equation (1) to obtain the following equation _St · S _t-Δt. * = [Ae ^{j ((ωc + Δω) t + φ0)} + n _t e ^{j (ωct + φt)} ] × [Ae- ^{j ((ωc + Δω) (t-Δt) + φ0)} + n _t-Δt e
^{-J ([omega} ] ^{c (t- [Delta} ] ^{t) + [phi} ] ^{t- [Delta} ] ^t) ] (3) The conversion formula is calculated. The conversion formula obtained from the result of this calculation is used to calculate the included signal components γ _ss , γ _nn , γ
Expressed to organize to _ns, become ^{_{S t × S t-Δt *}} = γ ss + γ nn + γ sn + γ ns ...... (4). Here, the first term γ _ss of the equation (4) is the product of the signal component of the equation (1) and the signal component of the equation (2), and the following equation γ _ss = Ae ^{j ((ωc + Δω) t + φ0)} × Ae ^{− j ((ωc + Δω) (t−Δt) + φ0)} = A ² e ^{j (ωc + Δω) Δt} (5) The second term γ _{nn in} equation (4) is
It is the product of the noise component of equation (1) and the noise component of equation (2), and is expressed by the following equation γ _nn = n _t e ^{j (ωct + φt)} × n _t−Δt e
^{−j (ωc (t−Δt) + φt−Δt)} = n _t n _t−Δt e
^{j (ωcΔt + φt−φt−Δt)} (6) Also, the third term γ _sn of equation (4) is
It is a product of the noise component of the equation (1) and the signal component of the equation (2), and is ^expressed by the following equation γ _sn = Ae ^{j ((ωc + Δω) t + φ0)} × n _t−Δte
^{j (ωc (t−Δt) + φt−Δt)} = An _t−Δt e ^jX (7) X = (ω _c + Δω) t + φ ₀ − (ω _c (+ φ _{t−Δt of t−Δt} ) ...... ( 8), and the fourth term γ _{ns in} equation (4) is
It is the product of the signal component of equation (1) and the noise component of equation (2), and is ^expressed by the following equation: γ _ns = Ae ^{j ((ωc + Δω) (t−Δt) + φ0)} × n _t ej ^{(ωct + φt)} = An _t e ^{jK ...... (9) Y = ω} c t + φ t - (ω c + Δω) (t-Δt) -
It can be arranged like φ ₀ (10).

From the above results, if the reception state in which the C / N of the reception signal is good is obtained, the first and second reception signals S _t and S _{t- Δt} represented by the equations (1) and (2) are Since it is small enough to be ignored, it can be considered that the signal components γ _nn , γ _sn , and γ _ns of the second term, the third term, and the fourth term of the equation (4) are almost 0 (equation (6)). , (7), (9)). Therefore, the multiplication formula of the formula (3) is practically _expressed by the following formula S _t × S _t−Δt * = γ _ss = A ² e ^{j (ωc + Δω) Δt} (11) It will be represented by the product of the signal components of equation (2).

Therefore, by detecting the value of the real part of equation (11),
The signal amplitude A of the equation (1) can be obtained. Also (1
By detecting the value of the imaginary part of the equation (1), the phase (ω _c + Δω) Δt of the equation (1) can be obtained. Since the carrier frequency ω _c and the time Δt are known, the frequency offset Δω which is the transmitted frequency input signal.
Can be calculated.

In this way, when the C / N is in a good reception state (0 [dB] or more) as in FM communication, the frequency Δω of the frequency signal can be detected from the reception signal, and thus the detection result The FM demodulator can be configured based on the above. Because
In the signal transmission by FM modulation, the signal level of the signal component is sufficiently higher than the noise signal component, so C /
Since N is obtained, and hence the conversion formula of the formula (11) can be obtained directly from the multiplication formula of the formula (3), the frequency signal Δω can be immediately detected.

On the other hand, for example, a spread spectrum signal was used.
When receiving a PSK modulated signal, for example, 1
The value of the frequency signal Δω and the value of the signal amplitude A keeps a constant value for about a second. Therefore, even if the C / N of the received signal is 0 [dB] or less, the input frequency It can be converted so that the signal can be detected.

That is, in this case, based on the equation (4) obtained from the multiplication result of the equation (3), the following equation, MS = ▲ ▼ = _ss + _nn + _sn + _ns (12) Calculate the average value MS. In the equation (12), the condition of the following equation, If the value of Δt is selected so that the above holds, the second term _nn of the equation (10) becomes _nn = 0 (14). In the equation (13), B is the control band ([H
z]) is shown.

Here, in the signal component γ _nn (equation (6)), ω _c Δ
The average value of the signal component of t can be set to 0 by selecting the relationship of 2πn (n is an integer) with respect to the carrier period. The signal components of phi _t and phi _{t-Delta] t,} by the selected Delta] t (13) in function of the type, phi _t
And the components of φ _t-Δt have no correlation with each other, the average value thereof becomes almost zero. As a result, _nn becomes almost zero.

The values of the third and fourth terms _sn and _ns are _sn = 0 ... (15) _ns = 0 ... (16).

Here, the signal components γ _sn and γ _ns (equation (7) and (9)
In the equation), the signal components of ω _c t and (ω _c + Δω) t are
The average value becomes 0 by sampling at an interval of time Δt larger than 1 / 2B. Further, the signal component of (ω _c + Δω) Δt has a value of approximately 2πn, so that the average value is 0.
become. Further, since the noise components of φ _t and φ _t−Δt are random, the average value thereof is 0.

Since φ ₀ is the initial phase, it does not affect the average value. Therefore, the average values _sn and _ns become zero.

The fact that Eqs. (15) and (16) hold can be explained as follows. First, if the probabilistic property of noise represented by the second term of the equation (1) is modeled, the noise usually has an average of 0.
Can be assumed to have a Gaussian distribution of. When such noise is modulated by the carrier whose center frequency is ω _c , the resulting signal component is calculated by the equation of the second term of the equation (1), that is, n _t ej ^{(ωct + φt)} (16-1). Can be expressed. Here, n _t is the square root of the sum of squares of a Gaussian stochastic process, and is called the Rayleigh distribution. Φ _t is a stochastic process independent of n _t and −π and π It has a uniform distribution in the range.

The average of the baseband part of this noise is Is. This means that there is independence between n _t and φ _t ,
Moreover, since the average of cos (φ _t ) and sin (φ _t ) is 0, it can be easily confirmed.

By the way, if we take the average of equations (7) and (8), X is given by the equation (8), and X = (ω _c + Δω) t + φ ₀ − (ω _c (t + Δt) + φ _t−Δt ) (16-4). Here, there is independence between φ _{t −Δt} ) and n _t , and they are uniformly distributed in the range of −π and π.
Therefore, ignoring the integral multiples of 2π contained in X due to the properties of the cos function and sin function, it can be considered that X itself is a stochastic process with uniform distribution in the range of −π and π.

 Therefore, ▲ ▼ ＝ ▲ ▼ ＝ 0 …… (16-5) and eventually ▲ ▼ ＝ ………… (16-6).

Similarly, in equations (9) and (10), Y is −
Since it can be considered that there is a uniform distribution in the range of π and π, ▲ ▼ = 0 (16-7).

After all, for each _Δt selected in the relationship of equation (13), (12)
If the average value MS of the equation is calculated, the first term of equation (12) ▲ _ss
Only ▼ will remain and the average value MS will be become.

Therefore, when the C / N of the received signal is 0 [dB] or less,
By obtaining the average of the outputs of the equation (4) obtained based on the multiplication result of the equation (3), the average value output MS corresponding to the conversion equation of the equation (17) is actually obtained, Therefore, if this average value MS is detected, the signal amplitude A of the received signal can be detected from its real part and the frequency input signal component Δω can be detected from its imaginary part. In this way, as is clear from the fact that the conversion formula (17) does not include a noise component, even if C / N is 0 [dB] or less, it can be prevented from being affected by the noise component. .

The frequency detecting device realizes the above operation principle by the configuration shown in the figure. In the figure, 1 indicates a frequency detecting device as a whole, and a received signal RVS is converted into a digital signal in an analog-digital conversion circuit 2 which performs sampling operation at a sampling time Δt. Here, the sampling time Δt is previously selected to a value that satisfies the condition of the above equation (13) for the reception signal RVS whose frequency band is limited to B [Hz].

This analog-to-digital conversion is, as described above with respect to the equation (1), the real part data R which is the combined value of the signal component and the noise component included in the reception signal RVS, which is the value I ₀ on the real axis of the complex plane. Output _eo and imaginary part data I _mo having a value Q _{0 on the} imaginary axis. Thus, the analog-to-digital conversion circuit 2 performs the first multiplication of the real part data _Reo and the imaginary part data _Imo, which are vector representations of the reception signal RVS in the same manner as described above for the expressions (1) and (2). Supply to the circuit 3.

The first multiplication circuit 3 receives a frequency detection signal F _OUT obtained from a voltage controlled oscillator (VCO) 4. Here, the frequency detection signal F _OUT is supplied to the multiplication circuit 3 as a vector signal which is a subtraction input for the frequency signal component Δω described in the equations (11) and (17), and the following equation I ₁ = cosφ is used as the value of the real number axis. ₁・ ・ ・ The value I ₁ represented by (18) is used as the data R _eo and I on the received signal side.
_{In addition} to multiplying _mo , the data R _eo and I _mo are multiplied by a value Q ₁ represented by the following equation Q ₁ = sinφ ₁ (19) as a value on the imaginary axis.

The multiplication circuit 3 is composed of a ROM, receives the data of the reception signal RVS and the data of the frequency detection signal F _OUT , and multiplies both of them as a complex number. The following expression R _e1 = I ₀ I _Q + I ₁ Q ₀ . ... (20) The real part data R _e1 and the imaginary part data I _m1 represented by the following formula I _m1 = −I ₀ I ₁ + Q ₀ Q ₁ (21) are transmitted.

The multiplication calculation force represented by the equations (20) and (21) means that the vector representing the reception signal RVS is rotated on the complex plane by an angle φ ₁ corresponding to the frequency detection signal F _OUT, and The frequency detection signal F _OUT is fed back and combined with the reception signal RVS to enable automatic frequency control (AF
C) You have configured a loop.

The real part data R _e1 of the first multiplication circuit 3 is directly supplied to the second multiplication circuit 5 as the first real part data I _1k , and delayed by the delay circuit 6 to obtain the second real number. The partial data I _{1 (k−1)} is supplied to the second multiplication circuit 5.
Further, the first multiplication circuit 3 outputs the imaginary part output data I _m1 as the first
Of the imaginary part data Q _1k of the second imaginary part data Q _1k
It is supplied to the second multiplication circuit 5 as _{1 (k-1)} .

Here, the delay circuits 6 and 7 have a delay amount corresponding to a time of a predetermined number of samples (for example, one sample time), and thus, as described above with respect to the equation (3), the delay circuits 6 and 7 with respect to the signal at the time point t. Time corresponding to the amount of delay Δ
Performs the function of forming an input signal for multiplication by the complex conjugate component of the signal at a time point t earlier. As a result, at the output terminal of the second multiplication circuit 5, the real part data R represented by the following equation R _e2 = I _1k I _{1 (k-1)} + Q _1k Q _{1 (k-1)} (22) _e2 and the imaginary part data I _m2 represented by the following equation I _m2 = −I _1k Q _{1 (k-1)} + I _{1 (k-1)} Q _k (23) are obtained, and these real parts are obtained. The data R _e2 and the imaginary part data I _m2 are respectively integrated circuits 8
And 9 to the arithmetic processing circuit (CPU) 10.
Here, the integrating circuits 8 and 9 are composed of an adder circuit and a latch circuit, and are provided so as to reduce the frequencies of the real part data _Re2 and the imaginary part data _{Im2 to} a frequency that the CPU 10 can easily process.

In the case of this embodiment, the CPU 10 executes the functions described above with respect to the expressions (12) to (17) by taking the average value of the supplied real part data R _e3 and imaginary part data I _m3. Therefore, while obtaining the real part and imaginary part data for the frequency signal component Δω in equation (17), this data is
It is converted into a frequency signal by performing t _an ^-1 conversion.

Thus, the frequency data F _D obtained by the CPU 10 is VCO
The frequency detection signal F _OUT corresponding to the frequency is converted into a frequency detection signal F _OUT and fed back to the first multiplication circuit 3.

In this way, the frequency detection device 1 forms an AFC loop as a whole, and feeds the frequency detection signal F _OUT obtained from the VCO 4 to the multiplication circuit 3 to feed back the CPU 10
Is the frequency detection signal F _OUT is the frequency signal Δω of the received signal RVS
It is possible to obtain a frequency detection signal F _OUT whose frequency changes in accordance with the frequency signal Δω of the reception signal RVS at the output end of the VCO 4,
This is sent as the detection output of the frequency detection device 1.

According to the frequency detection device 1 shown in the figure, by performing the arithmetic processing with each signal component of the received signal RVS as the value of the real number axis and the value of the imaginary number axis on the complex number plane, After executing the calculation of the above in the second multiplication circuit 5, the CPU 10 executes the calculation of the above-mentioned average values for the formulas (12) to (17) to obtain the frequency included in the received signal RVS. A frequency detection signal F _OUT representing the signal Δω can be obtained.

For this reason, as described above with respect to the equations (12) to (17), the value of the sampling time Δt is selected to satisfy the condition of the frequency limit width ((13)).
Even if the received signal RVS having C / N of 0 [dB] or less is received, by practically eliminating the signal component including the noise component (Equation (14)) to (15) It is possible to reliably detect only the received frequency signal component without being affected by the noise component. Therefore, when a signal obtained by PSK-modulating a carrier as a frequency signal arrives, PSK demodulation can be surely performed even when the C / N is bad.

In the above-mentioned embodiment, the conversion formula of formula (17) is obtained by taking the average value based on the multiplication formula of formula (3), but the reception signal RVS with good C / N is received. When it is possible to do so, the frequency signal component Δω may be detected directly by using the conversion equation of the equation (11) without executing the calculation of the average value in the CPU 10. This example is
It is suitable to be applied when receiving and demodulating an FM signal having a good C / N.

In the above-described embodiment, the explanation has been given assuming that only the angle signal component of the vector signal component is processed, but the CPU 10 detects the signal amplitude component A of the equations (11) and (17). According to this structure, it is possible to obtain the same envelope detection signal as if the square-law detector is used. Therefore, in the receiver in the communication using the spread spectrum signal by the frequency detecting device according to the present invention, the envelope signal used for the search mode when the PN synchronization is obtained can be obtained at the same time as the frequency signal, and thus the overall configuration is simple. A spell tram spread signal receiver can be realized.

〔The invention's effect〕

As described above, according to the present invention, by taking the average value of the multiplication output of the conjugate complex number, even if the frequency signal with C / N of 0 [dB] or less arrives, the noise component is not affected and the reliability is ensured. Further, it is possible to easily obtain a frequency detecting device capable of detecting the frequency of the frequency signal.

[Brief description of drawings]

The figure is a block diagram showing a frequency detection signal according to the present invention. 1 ... Frequency detection device, 2 ... Analog-digital conversion circuit, 3, 5 ... Multiplication circuit, 4 ... VCO, 6, 7 ...
Delay circuit, 8, 9 ... Integrating circuit, 10 ... CPU ,.

Claims (1)

[Claims] 1. A sampling time Δt (Δt>) for an input signal.
(1 / 2B), B is a limiting band), a predetermined complex number signal representing the input signal at the time point t and a predetermined complex number signal for the first complex number signal based on the output of the sampling unit. Multiplication means for obtaining a multiplication output of a second complex number signal representing a conjugate complex number of the input signal at a time point different by time Δt; and signal processing means for taking an average value for each time Δt based on the multiplication output of the multiplication means. And detecting the frequency of the input frequency signal based on the processing output of the signal processing means.