JPH0128553B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a digital modulator that generates phase modulated signals (including amplitude and phase modulated signals) through digital signal processing.

The phase modulation (amplitude/phase modulation) method is widely used as a highly efficient transmission method. In recent years, with the advancement of integrated circuit (LSI) technology, there has been an increasing demand for applying digital signal processing technology to implement circuits in a form suitable for LSI, even when performing modulation or demodulation. As is well known.

Now, as a conventionally known configuration of a digital phase modulator, the one shown in FIG. 1 is known. In FIG. 1, phase information to be transmitted that enters from a terminal 10a is converted into orthogonal trigonometric functions via a trigonometric function memory 11, and each signal memory 1
The required number of samples are stored in 2 and 12'.
The contents of the signal memories 12, 12' are the same as those of the waveform memory 1.
The impulse response waveform of the baseband filter for waveform shaping stored in 3 and the multiplier 14, 1
4′, and further added (subtracted) by 15, 1
5' and 1-word memories 16 and 16', a convolution operation is performed by cumulative addition (subtraction).
These convolution results are multiplied by the phase rotation information of the carrier wave inputted from the terminal 10b and converted into an orthogonal trigonometric function by the trigonometric function memory 11 by the multipliers 17 and 17', and then added (subtracted). The signals are combined by a calculator 18 to form a phase modulated signal, which is output from an output terminal 19.

By the way, in general, when converting a system into an LSI, it is important to reduce the number of operations in each functional block within the system, especially the number of multiplications, and to efficiently perform time-division multiplexing to reduce the number of operation units in the entire system. This is desirable in order to reduce the amount of In particular, when converting a modulator to an LSI, considering the LSI configuration of the modulator and demodulator at the same time and creating a lean system configuration as a whole will greatly contribute to hardware reduction.

Therefore, in order to consider the hardware scale of the entire system when the phase modulator/demodulator is implemented as an LSI, the configuration of the phase demodulator will also be explained. Conventionally, the following two types of typical configurations of digital phase demodulators are known. First, the received signal sampled at a certain frequency (denoted as _S ) is orthogonally detected using an orthogonal carrier wave, and the detection output is convolved with the impulse response waveform of the baseband filter. Method for obtaining orthogonal demodulated signals by removing generated harmonic components, 2nd method
An orthogonal passband signal is generated by convolving the received signal sampled at _S with the impulse response waveform of the orthogonal passband filter. This method performs demodulation.

The major difference in terms of the scale of demodulator hardware between the two methods above is that the first method needs to store both in-phase and quadrature detection outputs, whereas the second method only stores the received signal. It is a good thing to remember this. On the other hand, the first method requires fewer impulse response waveforms than the second method, which requires in-phase and quadrature-phase responses. However, compared to the memory (ROM) for storing impulse response waveforms, the memory (shift register) for storing the detection output requires more transistors, and according to standard MOS design, the first method is 1500 transistors compared to method 2
A large amount is required.

Furthermore, detection with the first method requires two multiplications every 1/ _S , whereas the second method requires four multiplications every 1/ _B ( _B : baud rate frequency < _S ). Fewer multiplications are required. For example, in the 8-phase phase modulation (4800b/s) recommended by CCITT, _B = 1600
Hz, and _S is usually selected to be 9600Hz or higher, that is, _S ≧6 _B. Therefore, the first method requires (6×2)−4=8 or more multiplications than the second method.

As described above, when two typical conventional digital phase demodulators are compared, the first method is inferior to the second method in terms of hardware scale and number of multiplications (frequency of use of the arithmetic unit).

However, when using the conventional digital phase modulator shown in Figure 1, if you try to share the impulse response waveform (ROM) of the baseband filter necessary for modulation with the demodulator, the above-mentioned Therefore, the first method, which is inferior in both hardware scale and number of multiplications, has no choice but to be adopted. On the other hand, if the second method, which is superior to the first method, is adopted as a demodulator, a ROM that stores the impulse response waveform of the baseband filter is used for modulation, and a ROM that stores the impulse response waveform of the orthogonal passband filter is used for demodulation. Since each device has its own ROM that stores the response waveform, the modulator/demodulator as a whole becomes wasteful.

This invention has been made in view of the above points, and is compatible with the demodulator of the above-mentioned second method, which is superior in terms of hardware scale and number of multiplications, and also requires less hardware than the conventional one. It is an object of the present invention to provide a digital modulator for digital phase modulation or digital amplitude/phase modulation with a reduced scale and number of multiplications.

The phase information to be transmitted is θ(i), and the impulse response of the baseband filter for waveform shaping is −L≦i≦
Assuming that f(i) is approximated by L, the output waveform b(k) of the baseband filter is basically obtained by the convolution operation shown in the following equation.

b(k)= _L 〓 ^l=-L f(l)e ^j 〓 ^(kl) ...(1) (However, k=1, 2,...) Figure 6 shows b(1), b as an example. (2), b(3) are shown in the generation process, and a in the same figure shows the impulse response f
(i), b in the same figure is the value of e ^j 〓 ^(kl) , c, d, e, f in the same figure,
g, h, and i are respectively e ^j 〓 ^(-1) , e ^j 〓 ⁽⁰⁾ , e ^j 〓 ⁽¹⁾ , e ^j
〓 ⁽²⁾ ,
e ^j 〓 ⁽³⁾ , e ^j 〓 ⁽⁴⁾ , e ^j 〓 ⁽⁵⁾ , i.e. l=3, 2, 1
,
The impulse responses at 0, -1, -2, and -3 are shown, and j in the figure shows the output waveform of the waveform shaping baseband filter, respectively.

Here, if L=2, b(2) is e ^j 〓 ⁽⁰⁾ , e ^j 〓 ⁽¹⁾ ,
Impulse responses d, e, of e ^{j 〓 (2)} , e j 〓 ⁽³⁾ , e ^j 〓 ⁽⁴⁾ ,
f,
By adding the values marked with ● in g and h, it is determined as shown in the ● mark of j. That is, b(2)=f(2)e ^j 〓 ^(2-2) +f(1)e ^j 〓 ^(2-1) +f(0)e ^j 〓 ^(2-0) +f(-1)e ^j 〓 ⁽²⁺¹⁾ +f(-2)e ^j 〓 ⁽²⁺²⁾ = ₂ 〓 ^l=-2 f(l)e ^j 〓 ^(2-l) . In this case, the impulse response f(l) of the waveform shaping baseband filter is approximated to a finite interval between -2≦l≦2. Here, the baseband phase information signal e ^j 〓 ⁽ⁱ⁾ to be transmitted is a complex number, for example, in the case of 4-phase PSK, θ±π/4, ±3π/
2, but for convenience of explanation, (b) is drawn as ejθ(i)→±1, and the impulse responses c to i also follow this.

By the way, when the sampling frequency of the digital phase modulation signal is _fS , when converting this modulation signal into an analog waveform and outputting it as described later, in order to remove harmonic components and facilitate separation of baseband components, , it is necessary to make f _S higher than the generation baud rate frequency f _B of the transmission phase information. For this purpose, if f _S /f _B = n, the waveform shaping baseband filter outputs the output b(k) with a cycle of the transmission phase information generation baud rate 1/f _B , but with the sampling rate 1/f An output b(k+m/n) of _S periods must be generated. That is, b(k) is
It is necessary to interpolate as (k+m/n). The output b(k+m/n) of the waveform shaping baseband filter after this interpolation is given by the following equation (2).

b(k+m/n)= _L 〓 ^l=-L f(l+m/n)e ^j 〓 ^(kl) ...(2) Here, m=0, 1, 2, ..., n-1 n=f _S /f _B f _B : Generation baud rate frequency of transmission phase information.

This equation (2) fixes the value of k (1, 2, ...) in equation (1), and for each value of k, m = 0, 1,
2..., n-1 to obtain b(k+m/n), which means that all calculations for obtaining b(k+m/n) are performed for the value of k. In addition, the impulse response f in equation (2)
(l+m/n) is b(k+
It represents the impulse response necessary to find m/n), and b(k+m/n) is less m/n than b(k).
Since the impulse response is shifted by the phase corresponding to m/n, the impulse response is also f(l+m/
n) must be used.

Next, the reason why formula (2) is valid will be explained using a specific example using FIG. Equation (2) is obtained through the same generation process as Equation (1). Now, in Figure 6, if we consider the case where L = 2, n = 4, and m = 1 as before, b (k + m / n) = b (2 + 1/4) is j
As shown by the △ mark, it is determined by adding the values shown by the △ mark in the impulse responses d, e, f, g, and h of e ^j 〓 ⁽⁰⁾ to e ^j 〓 ⁽⁴⁾ . That is, f(2+1/
4) = 0, so b (2 + 1/4) = f (1 + 1/4) e ^j 〓 ^(2-1) + f (0
+1/4)e ^j 〓 ^(2-0) +f(-1+1/4)e ^j 〓 ⁽²⁺¹⁾ +f(-2+1/4)e ^j 〓 ⁽²⁺²⁾ = ₂ 〓 ^l=-2 f (l+1/4)e ^j 〓 ^(2-l) ...(2a).

Similarly, in Fig. 6, b(k
+m/n)=b(2+2 ^{/ 4} ) is ^the impulse response ^d , e,
It is determined by adding the values indicated by ■ marks in f, g, and h. In other words, since f (2 + 2/4) = 0, b (2 + 2/4) = f (1 + 2/4) e ^j 〓 ^(2-1) + f (0 + 2
/4)e ^j 〓 ^(2-0) +f(-1+2/4)e ^j 〓 ⁽²⁺¹⁾ +f(-2+2/4)e ^j 〓 ⁽²⁺²⁾ = ₂ 〓 ^l=-2 f( l+2/4)e ^j 〓 ^(2-l) ...(2b)

Similarly, when m=3, b(k+m/n)=
b(2+3/4) is as shown by the x mark in j in the same figure.
Impulse responses d, e, f, g, h of e ^j 〓 ⁽⁰⁾ ~e ^j 〓 ⁽⁴⁾
It is found by adding the values indicated by the x marks in .

When the above equations (2a), (2b), etc. are expressed as a general equation using k, m, n, and L, it becomes as shown in equation (2). The digital phase modulation signal carrier wave obtained by multiplying the carrier wave by this equation (2) is as follows: y(k+m/n)=ej{ν(k+m/n)+θ0} b(k+m/n)=ej{ν(k+m/n) )+θ0} _L 〓 ^l=-L f(l+m/n)e ^j 〓 ^(kl) ......It is given by the real part or imaginary part of (3).

Here, ν2π・f _C /f _B ... (4) θ0: Initial phase of carrier wave (arbitrary) f _C : Carrier wave frequency.

Equation (3) converts the phase information θ(i) to be transmitted into orthogonal trigonometric functions, and combines them with the baseband and waveform shaping
This represents the process of performing a convolution operation with the impulse response waveform f(i) of the filter and further shifting it to the carrier frequency band.The conventional circuit shown in Figure 1 realizes this directly using a circuit. It is a phase modulator.

Note that the digital phase modulation signal obtained as described above is further converted into a smooth analog waveform by a D/A converter and a reduced pass filter. In this case, if the digital phase modulation signal is output at a baud rate at which the phase information to be transmitted is generated, harmonics included in the D/A converter output of the modulation signal waveform cannot be removed, so as mentioned above, _{S is}
chosen higher. For example, _B = 1600Hz, _S = 9600
Hz, then n=6.

Now, equation (3) can be transformed as follows.

y(k+m/n)= _L 〓 〓 ^l=-L {(l+m/n)・e ^j 〓 ^(l+m/n) }・ej{θ(
k−l)+ν(k−l)+θ0)} ...(5) This is an equation expressing the process of generating a digital phase modulation signal in the present invention. In other words, the composite phase θ(i)+ν・i+ of the phase information θ(i) to be transmitted and the carrier wave phase rotation information ν・i+θ ₀ for each baud rate.
Generate θ ₀ and convert it into an orthogonal trigonometric function, ei{θ(i)+ν・i+θ0} ei{θ(i)+ν・i+θ0} = cos{θ(i)+ν・i+θ ₀ }+jsin{ θ(i)+ν・i
+θ ₀ }...(6), and then use this orthogonal trigonometric function and the passband
Orthogonal impulse response waveform of filter: (i)e ^j 〓

Claims (1)

[Claims] 1. In a modulator that generates a phase modulated signal or an amplitude/phase modulated signal by digital signal processing, a modulator that generates and outputs composite phase information of phase information to be transmitted and phase rotation information of a carrier wave. 1, a second means for converting the synthesized phase information into an orthogonal trigonometric function, a third means for pre-storing the impulse response waveform of the orthogonal passband filter, and a means for converting the impulse response waveform and the orthogonal trigonometric function into an orthogonal trigonometric function. and fourth means for multiplying the respective functions, cumulatively adding (subtracting) the multiplication results, and combining them with each other to generate a modulation signal. 2. The digital modulator according to claim 1, wherein the first means outputs the generated composite phase information to the second means via a memory that stores a plurality of pieces of this information. 3. The first means outputs the generated composite phase information to the second means via a memory that stores a plurality of pieces of this information and a code conversion circuit that converts the code of the output of this memory. Claims characterized in that the adder (subtractor) for cumulative addition (subtraction) in the means is controlled whether to perform addition or subtraction depending on the output code of the code conversion circuit. The digital modulator according to item 1.