JP2009232079A - Interpolation filter - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method of preventing an increase in circuit scale and computational complexity by solving the problem that, in n-fold oversampling, a larger number of taps of an LPF are required as n is increased, accordingly, the circuit scale is increased, that is to say, the number of samplings per unit time is increased, therefore, product-sum operation in the LPF is increased. <P>SOLUTION: A sampling-rate converter frequency-converts a sampling frequency of an input signal into f Hz and a sampling frequency of an output signal to nf Hz. The sampling-rate converter is provided with: a low-pass-filter device which updates taps by an output signal of a counter operating at nf Hz so as to operate a delay device at f Hz; a tap coefficient control device for outputting different tap coefficients to a low pass filter at the frequency of nf Hz; and a counter device for controlling the tap coefficient control device. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

The present invention relates to a sampling rate conversion apparatus in a digital signal receiving apparatus.

In recent years, communication systems are required to be designed so as to be able to flexibly cope with radio wave usage conditions and frequency allocation regulations in each country. As an example, in WiMAX, the channel bandwidth is variable between 1.25 MHz and 20 MHz.

In order to process the received signal, it is necessary to convert the sampling rate of the signal to the rate of the processing system inside the system. An interpolation filter (interpolation filter) is used for this purpose.

FIG. 3 shows the basic structure of a sampling rate conversion apparatus using an interpolation filter. When an input signal having a sampling frequency f1 is oversampled to nf1 and output, first, the sampling rate increasing circuit 301 inserts (n-1) 0s between discrete data of the input signal. Next, the sampling rate can be converted by passing through the LPF 303 that removes other than the fundamental frequency at the operating frequency nf1 (see, for example, Patent Document 1).

FIG. 4 shows the state of the zero insertion. In the figure, two 0 points 403 are inserted between each sampling point 401 with respect to the original frequency, and the sampling frequency is tripled. However, if this is left, the connection between the zero insertion point and the original sampling point becomes a discontinuous point, and harmonics are generated, so the band is limited by the LPF.

JP 2006-222824 A

However, when the filter having the basic structure as described above is used as it is, the larger the coefficient n, the more taps of the LPF are required, and the circuit scale increases. That is, since the number of samplings per unit time increases, the product-sum operation in the LPF increases.

Further, when the rate of the input signal is not constant, it is necessary to prepare a rate amplification circuit and an LPF for each rate, which also leads to an increase in circuit scale. In other words, when a certain communication device is installed in the A region and the B region, if the frequency used in each region is different, the sampling rate conversion device in the device must be adapted to both regions in advance. As a result, the circuit scale is increased as compared with a device for a single frequency.

In order to solve the above-mentioned problems, the present invention provides a sampling rate conversion device that converts the sampling frequency of an input signal to f hertz and the sampling frequency of an output signal to nf hertz.
a low-pass filter device in which a delay device is operated by an output signal of a counter operating at nf hertz and is adapted to a plurality of sampling frequencies;
A tap controller that outputs different tap coefficients at a frequency nf to a low-pass filter;
A counter device for controlling the tap coefficient control device according to an output value;
A sampling rate conversion device comprising:

In the present invention, the tap coefficient control device changes the tap coefficient of the filter in synchronization with the sampling frequency of the output signal of the sampling rate conversion device, so that an arbitrary sampling value and the sampling value at the next sampling point can be changed. A sampling rate conversion device is characterized in that interpolation processing can be performed without inserting zeros into.

According to the present invention, the tap coefficient control device holds n values for one tap if the sampling rate conversion device aims at oversampling n times. A rate converter is assumed.

In the present invention, since the tap coefficient control device changes the tap coefficient in accordance with the output of the counter device, even when the sampling frequency is converted to an arbitrary magnification, the tap coefficient control device does not increase or decrease the number of taps. A sampling rate conversion device characterized by being capable of supporting a plurality of sampling frequencies by the number of taps.

Further, according to the present invention, the tap coefficient control device holds n values for one tap as long as the sampling rate converter enables up to n-times oversampling, and sets the tap coefficient to other values. A sampling rate converting apparatus is also used for oversampling of magnification.

According to the present invention, the filter device controls the operation of the shift register in the LPF, that is, the delay device, by the counter output. In the case of 2 times oversampling, the delay unit operates once every 2 clocks. In the case of 4 times oversampling, the delay unit operates once every 4 clocks. The tap coefficient is variable depending on the operation clock of the filter. In this way, the same output as that can be obtained even though 0 insertion is not performed, and it is not necessary to change the number of taps according to the rate.

Thus, according to the present invention, insertion of data 0 becomes unnecessary, a single filter can be used for a plurality of rates, the circuit scale can be reduced, and the oversampling factor can be increased. It is possible to cope without increasing the number of taps, and the effect that the increase of the multiplier can be suppressed appears.

Preferred embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a diagram showing an embodiment of the present invention. The sampling rate conversion apparatus according to this embodiment includes a filter apparatus 101, a tap coefficient control apparatus 102, and an n-bit counter 103. FIG. 5 shows a configuration of a general FIR filter in the filter device 101, in which 501 is a delay device and 503 is a tap.

As an example, the operation of this apparatus will be described with respect to the case where the sampling rate is converted by 4 times oversampling and the number of filter taps of the LPF is 8 taps. Here, when the frequency of the input signal is f1 and the frequency of the output signal of the filter device 101 is f2, as shown in FIG. 1, the operation clock c of this sampling rate conversion device is f2.

The delay unit 501 in the filter device 101, that is, the shift register, delays data only once every four clocks when the lower 2 bits of the output of the n-bit counter 103 are 0. That is, the data update timing is the frequency f1 of the input signal.

On the other hand, the tap control apparatus 102 outputs eight tap coefficients 503 to the filter 101. Each coefficient repeats four kinds of values in synchronization with the clock c, that is, f2. The value of tap n at this time is set to tapn0 to tapn3 (n = 0 to 7). That is, for example, the value of tap 0 repeats four types of values of tap00, tap01, tap02, and tap03 in synchronization with f2.

Since it is known that the result of multiplying the inserted zero data by the tap coefficient is zero when three zeros are inserted into the value f1 and the rate is four times the frequency f2. This does not need to be calculated.

Since the data input to the filter 101 of this apparatus is shifted by the delay unit 501 once every four clocks of the clock c, that is, the frequency f2, the data is updated once at each tap 503 as described above. Until then, four multiplications are performed with four types of tap coefficients from tapn0 to tapn3. This is equivalent to passing through the filter having the impulse response of FIG. 2 obtained by sampling the data whose rate is increased by inserting 0 at the frequency f2.

This will be described more specifically. The input data string to the sampling rate converter is {a0, a1, a2,. . . , A7,. . . }, In order to perform oversampling by four times, if three 0s are inserted, the data string becomes {a0,0,0,0, a1,0,0,0,. . . . , A7, 0, 0, 0,. . . . }.

In the case of the above example, each tap has four values, and the delay device of the FIR filter is operated only once every four clocks. Therefore, at time 0, the output of the filter is a0 * tap00. At time 1, the output of the filter is a0 * tap01.

In the case of the above example, the present invention performs the same processing as a filter having 8 taps and 32 taps. For example, tap01, tap02, and tap03 at time 0 are set to 0 with 0 inserted. Does not appear in the formula. Similarly, at time 1, tap00, tap02, and tap03 do not appear in the equation because 0 is being applied.

At time 28, a0 * tap 70 + a1 * tap 60+,. . . . . . , + A7 * tap00,
At time 29, a0 * tap 71 + a1 * tap 61+,. . . . . . , + A7 * tap01,
At time 30, a0 * tap72 + a1 * tap62 +,. . . . . . , + A7 * tap02,
At time 31, a0 * tap73 + a1 * tap63 +,. . . . . . , + A7 * tap03,
For example, at time 31, multiplication of 0 inserted with 0 and tap 70, tap 71, and tap 72 can be omitted because the result is 0, and thus does not appear in the equation.

That is, the same process as oversampling with zero insertion is realized without zero insertion. If this is n-times oversampling, the tap control device 102 holds n values for one tap, and only one value that does not perform multiplication with 0 among n values. This can be used according to the present invention.

Thus, in the case of four times oversampling, paying attention to the lower 2 bits of the counter 103, data is shifted by the delay unit 501 only once every four clocks, and tapn0, tapn1, tapn2, and tapn3 are used as the nth tap coefficients. By sequentially transferring the four from the tap control device 102 to the filter 101, a result equivalent to performing a 32-tap filter operation with eight taps can be obtained as a result of the present invention.

As an example, the case of 4 times oversampling was taken up, but in the case of 2 times oversampling, the frequency of data update of the shift register is once every 2 clocks, and in the case of 8 times oversampling, once every 8 clocks. You can do that. Since the update frequency of the tap coefficient is fixed, the tap coefficient is updated twice for one data in the case of 2 times and 8 times in the case of 8 times.

For example, in the case of double oversampling, attention is paid to the lower 1 bit of the counter 103, and the delay unit 501 is operated only when this is 0.

Furthermore, according to the present invention, as described above, the tap control device 102 controls the tap coefficient input to the tap in the filter 101 based on the output of the counter 103. If sampling is possible, the number of tap coefficients held by one tap is n as described above.

Here, by controlling which one of the tap coefficient tables in the tap control device 102 having n values is used according to the oversampling magnification, it is not necessary to recalculate the tap coefficient even if the magnification changes. Means to do this are also conceivable.

In other words, the value of the n tap coefficients indicates the impulse response of a filter having a bandwidth of m / n, for example, where m is the frequency of the output signal of the apparatus. All n tap coefficients, n / 2 times oversampling is the 0, 2, 4, ... th tap coefficient, n / 4 times oversampling is 0, 4, 8, ... Use the th tap coefficient.

For example, for a tap coefficient having four values, tap 0, tap 1, tap 2, and tap 3 for 4 times oversampling, tap 0 and tap 2 for 2 times oversampling, and tap 0 for no oversampling. It is to use.

As described above, the bandwidth of the filter can be adjusted to the magnification according to the way of referring to the value from the tap coefficient table. Once the table is defined, it is not necessary to recalculate or re-enter tap coefficients even if the magnification changes.

FIG. 6 is a diagram for explaining FIG. 5 in more detail. As shown here, when the input signal is output by n-times oversampling, the tap coefficient to be used is selected according to the bit value of the counter operating at the output frequency, and the product-sum operation that is the basic operation of the FIR filter is performed. Is. For example, in the case of quadruple oversampling, a 2-bit counter is used, and the tap coefficients used for the counter outputs 00, 01, 10, and 11 are changed.

As described above, according to the present invention, oversampling can be performed with one filter for a plurality of rates without performing zero insertion.

Configuration diagram according to the present invention Filter impulse response General sampling rate converter State of 0 insertion FIR filter configuration Detailed configuration of FIR filter

Explanation of symbols

101 ... Filter, 102 ... Tap control device, 103 ... Counter,
301 ... Sampling rate increase, 303 ... LPF,
401 ... sampling point, 403 ... 0 insertion point,
501 ... Delay device, 503 ... Tap.

Claims (5)

In a sampling rate converter for converting the sampling frequency of an input signal to f hertz and the sampling frequency of an output signal to nf hertz,
a low-pass filter device in which a delay device is operated by an output signal of a counter operating at nf hertz and is adapted to a plurality of sampling frequencies;
A tap control device that outputs different tap coefficients at a frequency nf to a low-pass filter;
A counter device intended to control the tap coefficient control device according to an output value;
A sampling rate conversion device comprising: The tap coefficient control device inserts 0 between an arbitrary sampling value and the sampling value at the next sampling point by changing the tap coefficient of the filter in synchronization with the sampling frequency of the output signal of the sampling rate conversion device. The sampling rate conversion apparatus according to claim 1, wherein the interpolation process can be performed without any problem. 2. The tap coefficient control device according to claim 1, wherein the tap coefficient control device holds n values for one tap if the sampling rate conversion device aims at oversampling n times. Sampling rate converter. Since the tap coefficient control device changes the tap coefficient according to the output of the counter device, even when the sampling frequency is converted to an arbitrary magnification, the tap coefficient control device does not increase or decrease the number of taps, and a plurality of tap coefficients are controlled. The sampling rate conversion apparatus according to claim 1, wherein the sampling rate conversion apparatus can cope with a sampling frequency. The tap coefficient control device holds n values for one tap if the sampling rate conversion device allows up to n times oversampling, and the tap coefficient can be used for oversampling at other magnifications. 2. The sampling rate conversion device according to claim 1, wherein the sampling rate conversion device is used.