JP5991897B2 - Communication apparatus and communication method - Google Patents

Description

  The present invention relates to communication technology.

  As described in Patent Documents 1 and 2, various techniques have been proposed for communication techniques.

JP 2009-141514 A JP 2009-164877 A

  As described in Patent Documents 1 and 2 described above, in a communication device, equalization processing may be performed on a received signal in order to remove distortion in the transmission path included in the received signal. is there. If this equalization process is not performed appropriately, data included in the received signal cannot be acquired properly, and reception performance may be degraded.

  Therefore, the present invention has been made in view of the above points, and an object thereof is to provide a technique capable of appropriately performing equalization processing.

  In order to solve the above-described problem, an aspect of a communication device according to the present invention includes a receiving unit that receives a signal from a communication partner device, and transmission of a frequency domain transmission path based on the received signal received by the receiving unit. A first acquisition unit for obtaining a function, and N first values (N ≧ 2) arranged in the frequency direction with respect to the transmission line transfer function obtained by the first acquisition unit, for each of the front and the rear An addition processing unit for adding L second values (L ≧ 1) and the N first values to which the L second values are added forward and rearward (N + 2L) ) Low-pass filter processing is performed on the third values, and the number of taps is M (M ≧ 2), and N from (N + 2L + M−1) fourth values output from the low-pass filter. And select the N number of the selected fifth values. Based on the value of the transmission line transfer function after the low-pass filter processing and the value of the transmission line transfer function after the low-pass filter processing obtained by the selection unit, the reception unit receives the value. An equalization processing unit that performs equalization processing on the received signal, and the additional processing unit sets L ≧ (M−1) / 2 when the number of taps M is an odd number, When the number of taps M is an even number, L ≧ M / 2 is set, and when the number of taps M is an odd number, the selection unit includes the (N + 2L + M−1) fourth values. The number of taps is selected from the (L + (M−1) / 2 + 1) th value to the (L + (M−1) / 2 + N) th value as the N fifth values with reference to the head, and the number of taps When M is an even number, the top of the (N + 2L + M−1) fourth values is used as a reference. (L + M / 2 + 1) -th from the value up to (L + M / 2 + N) th value is selected as said N fifth value.

  In the communication apparatus according to the aspect of the invention, the additional processing unit may set L = (M−1) / 2 when the tap number M is an odd number, and the tap number M is an even number. In this case, L = M / 2 is set.

  In the communication apparatus according to the aspect of the invention, the addition processing unit may add the N second values to the front of the N first values when adding the L second values. Of the first value of the first value of the first value of the first value of the first value of the first value of L The real part of the L second values is the real value of the Nth first value so that the real part of the L second values is oddly symmetric with respect to the starting value of the part. Of the N first values from the second value from the beginning to the (L + 1) th value from the beginning, among the N imaginary parts of the N first values. , The imaginary part of the L second values so that the imaginary part of the L second values is oddly symmetric with the leading value of the imaginary part of the N first values as the origin. Of the imaginary part of the Nth first values When the L second values are added to the rear of the N first values, of the real parts of the N first values, 2 from the end are added. With respect to L eighth values from the last value to the (L + 1) th value from the end, the L second values with the last value of the real part of the N first values as the origin The real parts of the L second values are added to the rear of the real parts of the Nth first values so that the real parts of the values are oddly symmetric. Of the imaginary part of the N first values with respect to the Lth ninth value from the second value from the end to the (L + 1) th value from the end The imaginary part of the L second values is behind the imaginary part of the Nth first value so that the imaginary part of the L second values is oddly symmetric with respect to the origin value Append to

  Moreover, in one aspect of the communication apparatus according to the present invention, the filter characteristic of the low-pass filter is variable, and a determination unit that determines the filter characteristic of the low-pass filter is further provided. A plurality of filter coefficients for realizing a plurality of types of filter characteristics different from each other are stored, and any one of the plurality of filter coefficients is determined as a use filter coefficient used by the low-pass filter. The low-pass filter process is performed using the use filter coefficient determined by the determination unit.

  In one aspect of the communication apparatus according to the present invention, a second acquisition unit that obtains a time-domain transmission line impulse response based on the transmission line transfer function obtained by the first acquisition unit, and the second acquisition A third acquisition unit for obtaining a maximum delay amount of the transmission line based on the transmission line impulse response obtained by the unit, and the determination unit is configured to obtain the maximum delay amount obtained by the third acquisition unit. Based on the above, the used filter coefficient is determined from the plurality of filter coefficients.

  In one aspect of the communication apparatus according to the present invention, the low-pass filter is a FIR (Finite Impulse Response) filter having a linear phase characteristic.

  Also, one aspect of the communication method according to the present invention is as follows: (a) a step of receiving a signal from a communication partner device; and (b) a frequency domain transmission based on the received signal received in the step (a). A step of obtaining a path transfer function; and (c) front and rear of N first values (N ≧ 2) arranged in the frequency direction for the transmission path transfer function obtained in step (b). A step of adding L second values (L ≧ 1) to (d) the N first values added to the front and rear of the L second values, respectively. A step of performing low pass filter processing with M taps (M ≧ 2) on (N + 2L) third values, and (e) (N + 2L + M−1) pieces of the low pass filter processing obtained as a result of the low pass filter processing. N fifth values are selected from the fourth value, and the selected N fifth values are A step of setting a value of the transmission path transfer function after the pass filter processing; and (f) based on the value of the transmission path transfer function after the low pass filter processing obtained in the step (e). Performing an equalization process on the received signal received, and in the step (c), when the number of taps M is an odd number, L ≧ (M−1) / 2 is set, When the number of taps M is an even number, L ≧ M / 2 is set. In step (e), when the number of taps M is an odd number, the (N + 2L + M−1) fourth number is set. Among the values, the (L + (M−1) / 2 + 1) th value to the (L + (M−1) / 2 + N) th value are selected as the N fifth values with reference to the head. , When the number of taps M is an even number, the (N + 2L + M−1) fourth values Chi, beginning on the basis of the chosen as the (L + M / 2 + 1) -th from the value (L + M / 2 + N) th until the value said N fifth value.

  According to the present invention, equalization processing can be performed appropriately.

It is a figure which shows the structure of a communication apparatus. It is a conceptual diagram which shows a low-pass filter process. It is a figure which shows an example of the data pattern (signal waveform) of original data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a figure which shows an example of the data pattern (signal waveform) of post-pad data. It is a figure which shows an example of the data pattern (signal waveform) of output data. It is a flowchart which shows operation | movement of a communication apparatus.

  FIG. 1 is a diagram illustrating a configuration of a communication apparatus 100 according to an embodiment. The communication apparatus 100 according to the present embodiment has a reception function of receiving an OFDM (Orthogonal Frequency Division Multiplexing) signal in which a plurality of subcarriers orthogonal to each other are combined, for example. Further, the communication device 100 communicates with a communication partner device through, for example, a power line. That is, the communication device 100 performs power line communication (PLC) with the communication partner device. Note that the communication device 100 may perform wireless communication with a communication partner device, or may perform wired communication by a method other than power line communication. Further, the communication device 100 may have a transmission function for transmitting an OFDM signal.

  As illustrated in FIG. 1, the communication device 100 includes a receiving unit 1 having an antenna 1a, an A / D converter 2, an FFT (Fast Fourier Transform) processing unit 3, an equalization processing unit 4, and data acquisition. Part 5. The communication device 100 further includes a transmission path estimation unit 6, an impulse response acquisition unit 7, a maximum delay amount acquisition unit 8, and a filter characteristic determination unit 9.

  The receiving unit 1 receives a signal from the communication partner apparatus using the antenna 1a. The receiving unit 1 performs amplification processing, down-conversion, and the like on the reception signal received by the antenna 1a to generate a baseband reception signal. In the present embodiment, the OFDM signal received by receiving section 1 is composed of J subcarriers. For example, J = 128. Each subcarrier constituting the OFMD signal received by the receiving unit 1 is modulated by, for example, a QPSK (Quadrature Phase Shift Keying) method.

  The A / D converter 2 converts the baseband received signal generated by the receiving unit 1 from an analog format to a digital format and outputs the converted signal. The FFT processing unit 3 performs FFT processing on the baseband received signal output from the A / D converter 2 and outputs the result. Specifically, the FFT processing unit 3 performs DFT (Discrete Fourier Transform) processing on the digital received signal output from the A / D converter 2 and outputs the result. As a result, the FFT processing unit 3 outputs a plurality of complex symbols that respectively modulate a plurality of subcarriers constituting the reception signal received by the receiving unit 1. As types of complex symbols, there are known symbols that are known on the receiving side and data symbols that are unknown on the receiving side. Known symbols are called pilot symbols or preamble symbols. A complex symbol is a complex value.

  The equalization processing unit 4 performs equalization processing on the data symbols output from the FFT processing unit 3 based on the frequency domain transmission path transfer function estimated by the transmission path estimation unit 6. Specifically, the equalization processing unit 4 outputs a data symbol corresponding to a certain subcarrier output from the FFT processing unit 3 to a value (complex number) at the frequency of the subcarrier with respect to the transmission path transfer function in the frequency domain. ), Or by dividing (complex division) by a value (complex number) at a frequency close to the frequency of the subcarrier for the transmission path transfer function (complex division). Thereby, the distortion of the transmission path included in the data symbol is removed.

  The data acquisition unit 5 performs demapping processing, deinterleaving processing, decoding processing, and the like on the output signal from the equalization processing unit 4 to obtain bit data included in the reception signal received by the reception unit 1. Reproduce.

  The transmission path estimation unit 6 includes a transfer function acquisition unit 60, an addition processing unit 61, a low pass filter (LPF) 62, and an output value selection unit 63.

  The transfer function acquisition unit 60 obtains a transmission path transfer function in the frequency domain based on the reception signal received by the reception unit 1. Specifically, the transfer function acquisition unit 60 outputs the frequency based on the plurality of known symbols that respectively modulate the plurality of subcarriers included in the reception signal received by the reception unit 1 and output from the FFT processing unit 3. Find the transmission line transfer function of the region. Hereinafter, the transmission path transfer function in the frequency domain is simply referred to as “transmission path transfer function”. Further, a known symbol used for obtaining a transmission path transfer function is referred to as “estimated known symbol”.

  In the present embodiment, when the communication partner apparatus transmits a known symbol for estimation, N (2 ≦ N ≦ J) subcarriers among J subcarriers constituting the OFDM signal are used. . That is, in an OFDM symbol (an OFDM signal for one symbol) including known estimation symbols, N subcarriers among J subcarriers are modulated with N known estimation symbols. Hereinafter, each of the N subcarriers is referred to as a “known subcarrier”. The N subcarriers are referred to as first to Nth known subcarriers in ascending order of frequency. The frequencies of the first to Nth known subcarriers are continuous in the frequency direction.

  The transfer function acquisition unit 60 performs transmission based on the N known symbols for estimation that respectively modulate the N known subcarriers included in the OFDM symbol received by the receiving unit 1 and output from the FFT processing unit 3. Find the path transfer function. Specifically, the transfer function acquisition unit 60 obtains values (complex numbers) at N different frequencies for the transmission path transfer function based on the N known symbols for estimation. It can be said that the value at a certain frequency of the transmission path transfer function is an estimated value of the transmission path characteristics at the frequency. The transfer function acquisition unit 60 divides the estimation known symbol for each of the N estimation known symbols output from the FFT processing unit 3 by a reference symbol that is an ideal value of the estimation known symbol. The value obtained thereby is taken as the value of the transmission path transfer function at the frequency (center frequency) of the known subcarrier corresponding to the known symbol for estimation. Thereby, the transfer function acquisition unit 60 obtains values at N different frequencies for the transmission path transfer function. Hereinafter, each of the N values is referred to as a “primary estimated value”. The N values arranged in the frequency direction are referred to as “transmission path primary estimated value sequence 200”. Further, the N values arranged in the frequency direction are referred to as first to Nth primary estimated values in order from the lowest frequency.

  The additional processing unit 61 performs L (N) for each of N values arranged in the frequency direction for the transmission line transfer function obtained by the transfer function acquisition unit 60, that is, for each of the front and rear of the transmission line primary estimated value sequence 200. By adding ≧ 2) values (complex numbers), the post-addition estimated value sequence 201 composed of (N + 2L) values is generated. Hereinafter, each of the L values may be referred to as a pad value.

  The low pass filter 62 performs low pass filter processing on the post-addition estimated value sequence 201 generated by the addition processing unit 61. Specifically, the low-pass filter 62 regards the real part of (N + 2L) values (complex numbers) constituting the post-addition estimated value sequence 201 as a plurality of sampling values for the real signal in the time domain, and Low pass filter processing is performed on the real part of (N + 2L) values. Similarly, the low-pass filter 62 regards the imaginary part of the (N + 2L) values as a plurality of sampling values for the real signal in the time domain, and performs a low-pass filter on the imaginary part of the (N + 2L) values. Process. In this way, the low-pass filter process is performed on (N + 2L) values constituting the post-addition estimated value sequence 201. When (N + 2L) values are input to the low-pass filter 62 with the number of taps M (M ≧ 2), the output signal 202 of the low-pass filter 62 is configured with (N + 2L + M−1) values. In the present embodiment, the low-pass filter 62 is composed of, for example, an FIR (Finite Impulse Response) filter having a linear phase characteristic. In the present embodiment, since the low-pass filter 62 is configured by the FIR filter, the filter coefficient (impulse response) used in the low-pass filter 62 is even symmetric. Note that M ≦ N is set.

  Here, the transmission path transfer function (transmission path primary estimated value sequence) obtained by the transfer function acquisition unit 60 includes transmission path noise (noise generated in the transmission path). In the present embodiment, a low-pass filter process is performed on the post-addition processing estimated value sequence 201 generated by the addition processing unit 61, thereby obtaining a transmission path transfer function (transmission path primary) obtained by the transfer function acquisition unit 60. The low-pass filter process is performed on the estimated value sequence 200). Thereby, the transmission line noise contained in the transmission line transfer function (transmission path primary estimated value sequence 200) can be reduced. Hereinafter, the output signal 202 of the low-pass filter 62 may be referred to as a “filter output signal 202”.

  The output value selection unit 63 uses the (N + 2L + M−1) values constituting the filter output signal 202 as the value of the transmission path transfer function after the filter processing used in the equalization processing in the equalization processing unit 4. N values of are selected. Each of the selected N values is referred to as a “secondary estimated value”. The N values arranged in the frequency direction are referred to as “transmission path secondary estimated value sequence 203”. Further, the N values arranged in the frequency direction are referred to as first to Nth secondary estimated values in order from the lowest frequency.

  The first to N-th secondary estimated values correspond to the first to N-th known subcarriers corresponding to the N known symbols for estimation used when the transmission path characteristics are obtained by the transfer function acquisition unit 60. Each corresponds. The output value selection unit 63 outputs the r-th secondary estimated value as a value after filtering the r-th primary estimated value (1 ≦ r ≦ N) obtained by the transfer function acquisition unit 60. The equalization processing unit 4 uses the r-th secondary estimated value as the value of the transmission path transfer function at the frequency of the r-th known subcarrier.

  The processing in the addition processing unit 61, the low-pass filter 62, and the output value selection unit 63 will be described in detail later.

  The impulse response acquisition unit 7 performs an IFFT (Inverse FFT) process on the transmission line transfer function obtained by the transfer function acquisition unit 60 to obtain a transmission impulse response in the time domain. Specifically, the impulse response acquisition unit 7 performs an IDFT (Inverse DFT) process on the transmission channel primary estimated value sequence 200 obtained by the transfer function acquisition unit 60 to obtain a transmission impulse response in the time domain. Ask.

  The maximum delay amount obtaining unit 8 obtains the maximum delay amount of the transmission line, in other words, the maximum spread of the transmission line impulse response, based on the transmission line impulse response obtained by the impulse response obtaining unit 7.

  The filter characteristic determination unit 9 determines the filter characteristic of the low-pass filter 62 based on the maximum delay amount obtained by the maximum delay amount acquisition unit 8. In the present embodiment, the filter characteristics of the low-pass filter 62 are variable. The low-pass filter 62 performs low-pass filter processing having the filter characteristics determined by the filter characteristic determination unit 9 on the post-addition processing estimated value sequence 201 generated by the addition processing unit 61.

  A method for determining the filter characteristics in the communication apparatus 100 will be described later in detail.

<About additional processing>
Next, processing in the additional processing unit 61 will be described in detail.

  Assuming that the impulse response, input signal, and output signal of the low-pass filter 62 are h (n), x (n), and y (n) (n ≧ 0), the output signal y (n) is expressed by the following equation (1). ), (2).

  In the present embodiment, the real part or imaginary part of each value constituting the post-addition processing estimated value sequence 201 generated by the addition processing unit 61 is input to the low-pass filter 62 as the input signal x (n). For the low-pass filter 62 having M taps, in other words, for the low-pass filter 62 having a plurality of element coefficients constituting the filter coefficient, the input signal x (n) having the number of data (the number of samples) is P. When input, an output signal y (n) having (P + M−1) pieces of data is obtained.

  FIG. 2 is a conceptual diagram showing the filter process when the filter process is performed using the formula (1) in the low-pass filter 62. In the expression (1), the output signal is obtained by calculating the sum of products of the filter coefficient and the input signal while fixing the input signal, inverting the filter coefficient, and shifting one by one from left to right. .

  In Formula (1), when n <(M−1), i> n may be satisfied. Since h (n−i) = 0 when i> n, in this case, although there are M element coefficients constituting the filter coefficient, substantially M element coefficients Only a part is used when obtaining the output signal y (n). Therefore, an appropriate value cannot be obtained for the output signals y (0) to y (M-2) indicated by hatching in FIG.

  When the number of data (number of samples) of the input signal x (n) is P, when n> (P−1), the input signal x (n) does not exist, so x (n) = 0. Become. Therefore, in this case, the number of input signals x (n) multiplied by the M element coefficients is substantially smaller than M. Therefore, an appropriate value cannot be obtained for the output signals y (P) to y (M + P−2) indicated by diagonal lines in FIG.

  Thus, if the input signal x (n) to be filtered is directly input to the low-pass filter 62, an appropriate value may not be obtained for some values of the output signal y (n).

  Therefore, in the present embodiment, in the addition processing unit 61, appropriate L pad values are added to the front and rear of the transmission channel primary estimated value sequence 200 that is the filter target, and obtained thereby. The post-addition estimated value sequence 201 composed of (N + 2L) values is input to the low-pass filter 62. This suppresses that some values of the output signal y (n) are not appropriate values. This point will be described in detail below.

<About the pad value data pattern>
Here, an appropriate data pattern of pad values added to the front and rear of the transmission path primary estimated value sequence 200 will be described. In the following description, the case where the real part of the (N + 2L) values constituting the post-addition estimated value sequence 201 is low-pass filtered in the low-pass filter 62 will be described. The same applies to the case where the part is subjected to low-pass filter processing.

  Further, the real part (the real part of the primary estimated value) of the transmission channel primary estimated value sequence 200 obtained by the transfer function acquiring unit 60 is assumed to be original data xr (n) (1 ≦ n ≦ N). The original data xr (n) is the real part of the value of the transmission path transfer function at the frequency of the nth known subcarrier.

  Further, the real part of the L pad values added in front of the transmission path primary estimated value sequence 200 is defined as a front pad value pfr (j) (1 ≦ j ≦ L), and the transmission path primary estimated value sequence 200 The real part of the L pad values added to the rear is assumed to be the rear pad value pbr (j) (1 ≦ j ≦ L). The post-pad data xpr (n) = [pfr (1),... Pfr (L), xr (1), xr ( 2),... Xr (N), pbr (1),... Pbr (L)] (1 ≦ n ≦ (N + 2L)). A real part of (N + 2L + M−1) values constituting the filter output signal 202 is defined as a filter output value for (k) (1 ≦ k ≦ (N + 2L + M−1)). The real part (real part of the secondary estimated value) of the transmission channel secondary estimated value sequence 203 output from the output value selecting unit 63 is set as output data yr (n) (1 ≦ n ≦ N).

  In the following description, as an example, N = 32, L = 10, and M = 21. In such an example, out of 128 subcarriers constituting the OFDM symbol, 32 subcarriers whose frequencies are continuous in the frequency direction are used for transmitting the estimation known symbols.

  FIG. 3 is a diagram illustrating an example of the data pattern 300 of the original data xr (n). The horizontal axis in FIG. 3 indicates the value (sample number) of the index n, and the vertical axis in FIG. 2 indicates the value of the original data xr (n).

  When the original data xr (n) shown in FIG. 3 is acquired by the transfer function acquisition unit 60, the front pad value pfr (j) = xr (1) and the rear pad value pbr (j) = xr (N) Then, the data pattern 301 of the post-pad data xpr (n) is as shown in FIG. Then, the low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 4, and then the output value selection unit 63 outputs the filter output value for (k (k) output from the low-pass filter 62. ) To select N values from k = L + (M−1) / 2 + 1 to k = L + (M−1) / 2 + N, that is, 32 values from k = 21 to k = 52 Is selected, the data pattern 302 of the output data yr (n) is as shown in FIG. In this case, output data yr (n) = filter output value for (n + L + (M−1) / 2). The reason why N values from k = L + (M−1) / 2 + 1 to k = L + (M−1) / 2 + N among the filter output values for (k) will be described later.

  In FIG. 5, in addition to the data pattern 302 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 5, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 5, when the front pad value pfr (j) = xr (1) and the rear pad value pbr (j) = xr (N), the data pattern of the output data yr (n) (signal (Waveform) is substantially coincident with the data pattern (signal waveform) of the original data xr (n) at the center, but slightly deviates at both ends. Therefore, even if such a real part of the pad value is added to the real part of the transmission line primary estimated value sequence 200, some values of the output data yr (n) are not sufficiently appropriate values. .

  6 shows that when the original data xr (n) shown in FIG. 3 is acquired by the transfer function acquisition unit 60, the front pad value pfr (j) = xr (L + 2−j) and the rear pad value pbr (j). A data pattern 311 of post-pad data xpr (n) when = xr (N−i) is shown.

  In the example of FIG. 6, among the N values of the original data xr (n), from the second value from the top (xr (2)) to the (L + 1) th value from the top (xr (L + 1)). For the L values, the L values of the front pad value pfr (j) are evenly symmetric with the leading value (xr (1)) of the N values of the original data xr (n) as the origin. Thus, L values of the front pad value pfr (j) are added in front of the N values of the original data xr (n). That is, among the 32 values of the original data xr (n), for the 10 values from the second value from the beginning to the eleventh value from the beginning, the leading value of the 32 values is set. As the origin, 10 pad values of the front pad value pfr (j) are forward of 32 values of the original data xr (n) so that the 10 values of the front pad value pfr (j) are even-symmetrical. Has been added. As a result, the data pattern of the data composed of the front pad value pfr (j) and the original data d (1) becomes a data pattern in which the data pattern of the original data xr (1) to xr (L + 1) is folded to the left side. .

  In the example of FIG. 6, among the N values of the original data xr (n), the second value (xr (N−1)) from the end to the (L + 1) th value (xr (N−)) from the end. L)), the L values of the rear pad value pbr (j) with the end value (xr (N)) of the N values of the original data xr (n) as the origin. L are added to the rear of the N values of the original data xr (n) so that is even symmetric. Thus, the data pattern of the original data xr (N) and the rear pad value pbr (j) is such that the data pattern of the original data xr (N−L) to xr (N−1) is folded to the right side. It becomes a data pattern.

  The low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 6, and then the output value selection unit 63 outputs k = L + (M− of the filter output value for (k). 1) When N values from / 2 + 1 to k = L + (M−1) / 2 + N are selected, that is, when 32 values from k = 21 to k = 52 are selected, the output data yr (n) The data pattern 312 is as shown in FIG.

  In FIG. 7, in addition to the data pattern 312 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 7, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 7, when the front pad value pfr (j) = xr (L + 2−j) and the rear pad value pbr (j) = xr (N−i), the front pad value pfr (j) = Xr (1) and rear pad value pbr (j) = xr (N), the data pattern (signal waveform) 312 of the output data yr (n) is the data pattern of the original data xr (n). (Signal waveform) 300 is almost the same at the center, but slightly deviated at both ends. Therefore, even if such a real part of the pad value is added to the real part of the transmission line primary estimated value sequence 200, some values of the output data yr (n) are not sufficiently appropriate values. .

  FIG. 8 shows that when the original data xr (n) shown in FIG. 3 is obtained in the transfer function acquisition unit 60, the front pad value pfr (j) = 2xr (1) −xr (L + 2−j), the rear pad A data pattern 321 of post-pad data xpr (n) when the value pbr (j) = 2xr (N) −xr (Ni) is shown.

  In the example of FIG. 8, of the N values of the original data xr (n), the second value from the top (xr (2)) to the (L + 1) th value (xr (L + 1)) from the top. For the L values, the L values of the front pad value pfr (j) are oddly symmetric with the leading value (xr (1)) of the N values of the original data xr (n) as the origin. Thus, L values of the front pad value pfr (j) are added in front of the N values of the original data xr (n). That is, among the 32 values of the original data xr (n), for the 10 values from the second value from the beginning to the eleventh value from the beginning, the leading value of the 32 values is set. As the origin, 10 pad values of the front pad value pfr (j) are forward of the 32 values of the original data xr (n) so that the 10 values of the front pad value pfr (j) are oddly symmetric. Has been added. As a result, the data pattern of the data composed of the front pad value pfr (j) and the original data xr (1) is obtained by folding the data pattern of the original data xr (1) to xr (L + 1) to the left side and then further on the vertical direction side. The data pattern is as if it was folded back.

  In the example of FIG. 8, among the N values of the original data xr (n), the second value (xr (N−1)) from the end to the (L + 1) th value (xr (N− L)), the L values of the rear pad value pbr (j) with the end value (xr (N)) of the N values of the original data xr (n) as the origin. Are oddly symmetric, L values of the backward pad value pbr (j) are added behind the N values of the original data xr (n). As a result, the data pattern of the original data xr (N) and the rear pad value pbr (j) is obtained by folding the data patterns of the original data xr (N−L) to xr (N−1) to the right side. Further, the data pattern is such that it is folded back in the vertical direction.

  The low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 8, and then the output value selection unit 63 outputs k = L + (M− of the filter output value for (k). 1) When N values from / 2 + 1 to k = L + (M−1) / 2 + N are selected, that is, when 32 values from k = 21 to k = 52 are selected, the output data yr (n) The data pattern 322 is as shown in FIG.

  In FIG. 9, in addition to the data pattern 322 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 9, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 9, when the front pad value pfr (j) = 2xr (1) −xr (L + 2−j) and the rear pad value pbr (j) = 2xr (N) −xr (N−i) In this case, the data pattern (signal waveform) 322 of the output data yr (n) substantially matches the data pattern (signal waveform) 300 of the original data xr (n) not only at the center but also at both ends. It becomes like this. Therefore, when such a real part of the pad value is added to the real part of the transmission path primary estimated value sequence 200, each value of the output data yr (n) is an appropriate value.

  Thus, when the front pad value pfr (j) = 2xr (1) −xr (L + 2−j) and the rear pad value pbr (j) = 2xr (N) −xr (N−i), the output is performed. No distortion occurs in the data pattern of the data yr (n).

  Therefore, the addition processing unit 61 according to the present embodiment is pfr (j), which is a real part of L pad values to be added in front of the transmission channel primary estimated value sequence 200 obtained by the transfer function acquisition unit 60. Is pfr (j) = 2xr (1) -xr (L + 2-j). Further, the addition processing unit 61 sets pbr (j), which is the real part of the L pad values to be added behind the transmission channel primary estimated value sequence 200, to pbr (j) = 2xr (N) −xr (N -I).

  The same applies to the imaginary part of the L pad values added in front of the transmission channel primary estimated value sequence 200 obtained by the transfer function acquisition unit 60. Here, the imaginary part of the transmission path primary estimated value sequence 200 (imaginary part of the primary estimated value) is the original data xi (n) (1 ≦ n ≦ N), and the transmission path primary estimated value sequence 200 is placed in front of it. The imaginary part of the L pad values to be added is defined as the front pad value pfi (j) (1 ≦ j ≦ L), and the imaginary part of the L pad values added to the rear of the transmission path primary estimated value sequence 200. Is the rear pad value pbi (j) (1 ≦ j ≦ L), and the imaginary part of the transmission line secondary estimated value sequence 203 is output data yi (n) (1 ≦ n ≦ N). The additional processing unit 61 according to the present embodiment sets pfi (j) = 2xi (1) −xi (L + 2−j) and pbi (j) = 2xi (N) −xi (N−i). Thereby, each value of the output data yi (n) becomes an appropriate value.

  In the above example, N = 32, L = 10, and M = 21, but the same applies to other values of N, L, and M.

<About the number (length) of pad values>
Here, an appropriate number (length) of L pad values, that is, an appropriate value of L will be described. In the following description, the number of taps M of the low-pass filter 62 is an odd number, for example, M = 21 as described above. Similarly to the above, N = 32.

  FIG. 10 shows that when L = M−1 = 20 and the original data xr (n) shown in FIG. 3 is acquired by the transfer function acquisition unit 60, the front pad value pfr (j) = 2xr (1 ) -Xr (L + 2-j) and the rear pad value pbr (j) = 2xr (N) -xr (Ni), the data pattern 331 of post-pad data xpr (n) is shown.

  The low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 10, and then the output value selection unit 63 outputs k = L + (M− of the filter output value for (k). 1) When N values from / 2 + 1 to k = L + (M−1) / 2 + N are selected, that is, when 32 values from k = 31 to k = 62 are selected, the output data yr (n) The data pattern 332 is as shown in FIG.

  In FIG. 11, in addition to the data pattern 332 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 11, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 11, L = M−1, front pad value pfr (j) = 2xr (1) −xr (L + 2−j), rear pad value pbr (j) = 2xr (N) −xr ( In the case of (N−i), the data pattern 332 of the output data yr (n) comes to substantially match the data pattern 300 of the original data xr (n). Therefore, when such a real part of the pad value is added to the real part of the transmission path primary estimated value sequence 200, each value of the output data yr (n) is an appropriate value.

  FIG. 12 shows a case where L = (M−1) / 4 = 5 and when the original data xr (n) shown in FIG. 3 is acquired by the transfer function acquisition unit 60, the front pad value pfr (j) = 2xr (1) −xr (L + 2−j), rear pad value pbr (j) = 2xr (N) −xr (N−i), and a data pattern 341 of post-pad data xpr (n) is shown. Yes.

  The low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 12, and then the output value selection unit 63 outputs k = L + (M− of the filter output value for (k). 1) When N values from / 2 + 1 to k = L + (M−1) / 2 + N are selected, that is, when 32 values from k = 16 to k = 47 are selected, the output data yr (n) The data pattern 342 is as shown in FIG.

  In FIG. 13, in addition to the data pattern 342 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 13, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 13, L = (M−1) / 4, front pad value pfr (j) = 2xr (1) −xr (L + 2−j), rear pad value pbr (j) = 2xr (N ) −xr (N−i), the data pattern 342 of the output data yr (n) is slightly at the rear end (right end) with respect to the data pattern 300 of the original data xr (n). It will shift. That is, in the case of L = (M−1) / 4, the data pattern 342 of the output data yr (n) is slightly distorted compared to the case of L = M−1.

  On the other hand, L = (M−1) / 2 = 10, the front pad value pfr (j) = 2xr (1) −xr (L + 2−j), and the rear pad value pbr (j) = 2xr (N) −xr. In the case of (Ni), the data pattern of the output data yr (n) is the data pattern 322 shown in FIG. Therefore, the data pattern of the output data yr (n) in this case almost matches the data pattern 300 of the original data xr (n) as in the case of L = M−1.

  Thus, when L ≧ (M−1) / 2, the distortion of the data pattern of the output data yr (n) is reduced.

  Here, for the output signal y (n) represented by the above formulas (1) and (2), the output signals y (0) to y (M−2) and the output signals y (P) to y (M + P). -2), an appropriate value cannot be obtained. Therefore, considering equations (1) and (2), if (M−1) pad values are added to the front and rear of the transmission channel primary estimated value sequence 200, the output data yr ( It is considered that the distortion of the data pattern of n) is reduced.

  However, in the present embodiment, since the filter coefficients used in the low-pass filter 62 have symmetry (even symmetry), the transmission line primary estimated value sequence 200 has (( Even when half the pad values of M-1) are added, it is considered that the distortion of the data pattern of the output data yr (n) is sufficiently small.

  Therefore, the addition processing unit 61 according to the present embodiment adds each of the front pad value pfr (j) and the rear pad value pbr (j) to be added to the front and rear of the real part of the transmission path primary estimated value sequence 200, respectively. The number (L) of a plurality of values for is set to one half or more of (M−1). That is, L ≧ (M−1) / 2.

  However, when M is an even number, (M−1) / 2 is not an integer. Therefore, when M is an even number, L ≧ M / 2, and when M is an odd number, L ≧ (M−1) / 2. Thereby, the value of the real part (the real part of the secondary estimated value) of the transmission channel secondary estimated value sequence 203 output from the output value selecting unit 63 becomes an appropriate value.

  The same applies to the imaginary part of the L pad values added in front of the transmission channel primary estimated value sequence 200 obtained by the transfer function acquisition unit 60. When M is an odd number, L ≧ (M−1 ) / 2, and if M is an even number, L ≧ M / 2. Thereby, the value of the imaginary part (imaginary part of the secondary estimated value) of the transmission channel secondary estimated value sequence 203 output from the output value selecting unit 63 becomes an appropriate value.

  In this example, the front pad value pfr (j) = 2xr (1) −xr (L + 2−j) and the rear pad value pbr (j) = 2xr (N) −xr (N−i) are used. Even if the data pattern of the pad value pfr (j) and the rear pad value pbr (j) is another data pattern, L ≧ M / 2 when M is an even number, and L ≧ M when M is an odd number. By setting (M-1) / 2, the values of the real part and the imaginary part of the transmission channel secondary estimated value sequence 203 output from the output value selection unit 63 become appropriate values.

<About the value selection method in the output value selection unit>
Next, processing in the output value selection unit 63 will be described in detail.

  A signal that has passed through a digital filter always has a delay due to the group delay characteristics of the digital filter. A group delay amount GD in an FIR filter having a linear phase characteristic, such as the low-pass filter 62 according to the present embodiment, is expressed by the following equation (3) using the number of taps M of the filter.

  On the other hand, in the present embodiment, L pad values are added to the front and rear of the transmission channel primary estimated value sequence 200 to be filtered.

  Therefore, the output value selection unit 63 according to the present embodiment considers the number of pad values added to the transmission channel primary estimated value sequence 200 in order to remove the influence of the group delay in the low-pass filter 62. Among the (N + 2L + M−1) values constituting the filter output signal 202 of the low-pass filter 62, N values from the (L + GD + 1) th value to the (L + GD + N) th value are selected with reference to the head. And output. That is, out of the (N + 2L + M−1) values constituting the filter output signal 202, the (L + (M−1) / 2 + 1) th value from the (L + (M−1) / 2 + 1) th value with respect to the head is used as a reference. Select and output N values up to the first value.

  However, when the number of taps M is an even number, the group delay amount GD is not an integer. Therefore, in this case, the group delay amount GD = M / 2, and the output value selection unit 63 sets (L + M / M) with the head as a reference among (N + 2L + M−1) values constituting the filter output signal 202. N values from the (2 + 1) th value to the (L + M / 2 + N) th value are selected and output.

  Considering the above contents separately for the real part and the imaginary part, the output value selection unit 63, when the number of taps M is an odd number, out of the filter output value for (k), k = L + (M−1). N values from / 2 + 1 to k = L + (M−1) / 2 + N are selected and output. Further, if the imaginary part of (N + 2L + M−1) values constituting the filter output signal 202 is a filter output value foi (k) (1 ≦ k ≦ (N + 2L + M−1)), the output value selection unit 63 taps. When the number M is an odd number, N values from k = L + (M−1) / 2 + 1 to k = L + (M−1) / 2 + N are selected from the filter output values foi (k). Output.

  On the other hand, when the number of taps M is an even number, the output value selection unit 63 selects N values from k = L + M / 2 + 1 to k = L + M / 2 + N among the filter output values for (k). Output. Further, when the tap number M is an even number, the output value selection unit 63 selects N values from k = L + M / 2 + 1 to k = L + M / 2 + N among the filter output values foi (k). Output.

  FIG. 14 shows a case where the number of taps M is an even number, for example, M = 22, N = 32, and L = M = 22, and the original data xr (n) shown in FIG. In this case, the post-pad data when the front pad value pfr (j) = 2xr (1) −xr (L + 2−j) and the rear pad value pbr (j) = 2xr (N) −xr (N−i) A data pattern 351 of xpr (n) is shown.

  The low-pass filter 62 performs low-pass filter processing on the post-pad data xpr (n) shown in FIG. 14, and then the output value selection unit 63 starts from k = L + M / 2 + 1 of the filter output value for (k). When N values from k = L + M / 2 + N are selected, that is, when 32 values from k = 23 to k = 54 are selected, the data pattern 352 of the output data yr (n) is as shown in FIG. Become.

  In FIG. 15, in addition to the data pattern 352 of the output data yr (n), a data pattern 300 of the original data xr (n) is also shown. In FIG. 15, the Δ mark indicates the value of the output data yr (n), and the X mark indicates the value of the original data xr (n). The post-pad data xpr (n) does not include a frequency component (noise component) higher than the cutoff frequency of the low-pass filter 62.

  As shown in FIG. 15, when the number of taps M is an even number, the group delay amount GD is set to M / 2 instead of the original value. Therefore, the data pattern 352 of the output data yr (n) is the original data. The data pattern 300 of xr (n) is advanced by 1/2 sample.

  On the other hand, when the number of taps M is an odd number, the group delay amount GD is set to the original value (M−1) / 2, and therefore, as shown in FIG. 11 described above corresponding to FIG. In addition, the data pattern 332 of the output data yr (n) substantially matches the data pattern 300 of the original data xr (n).

<How to determine filter characteristics>
The filter characteristic determination unit 9 according to the present embodiment stores in advance a plurality of filter coefficients for realizing a plurality of types of filter characteristics having different cutoff frequencies. The filter characteristic determination unit 9 determines one of the stored filter coefficients as a use filter coefficient used by the low-pass filter 62. In this way, the filter characteristic determination unit 9 determines the cutoff frequency of the low-pass filter 62. The low pass filter 62 performs a low pass filter process on the post-addition estimated value sequence 201 using the used filter coefficient determined by the filter characteristic determination unit 9.

  Next, a method for determining the used filter coefficient will be described.

  An OFDM symbol received by the receiving unit 1 is composed of an effective symbol (original signal for one symbol) and a guard interval. In the present embodiment, the guard interval time length is 1/8 of the effective symbol time length. Since the time length of the effective symbol is 1 / fo, the time length of the guard interval is 1 / (8 × f0).

  Further, when a signal passing through a transmission line having a delay amount T is received by the receiving unit 1, a signal waveform regarding the value of the transmission line transfer function acquired by the transfer function acquiring unit 60, that is, the above-described original data The signal waveform (see FIG. 2) of xr (j) and original data xi (j) has a frequency component having a frequency Ft corresponding to a delay amount (delay spread) T expressed by the following equation (4). To be included.

  Here, the delay amount T is a relative time length when the time length of the effective symbol is J, that is, “128”. The frequency Ft is a normalized angular frequency.

  Therefore, the cutoff frequency of the low-pass filter 62 is smaller than the frequency Ft, and the low-pass filter 62 has a frequency Ft corresponding to the delay amount T from the signal waveforms of the original data xr (j) and the original data xi (j). When the component is removed, the transmission path transfer function (transmission path secondary estimated value sequence) after the filtering process does not include information regarding the characteristics of the transmission path of the delay amount T necessary for the equalization process. Will end up.

  The normalized angular frequency is a value obtained by normalizing the angular frequency with the sampling frequency fs of the input data of the low-pass filter 62. Assuming that the frequency interval for a plurality of subcarriers constituting the OFDM signal received by the receiving unit 1 is f0, the sampling frequency fs = J × f0 is obtained using the number J of the plurality of subcarriers constituting the OFDM signal. expressed. In this embodiment, since J = 128, fs = 128 × f0.

  On the other hand, when the delay amount T of the transmission path between communication apparatuses that perform communication using the OFDM method is equal to or less than the time length of the guard interval, intersymbol interference and intercarrier interference do not occur. In view of this point, in the OFDM communication system in which the communication apparatus 100 according to the present embodiment is used, the transmission path delay amount T is approximately equal to or less than the guard interval time length. Accordingly, the signal waveforms of the original data xr (j) and the original data xi (j) hardly contain frequency components having a frequency Ft corresponding to the delay amount T larger than the time length of the guard interval. Therefore, the cut-off frequency of the low-pass filter 62 may be set to be equal to or lower than the frequency Ft corresponding to the delay amount T having the same size as the guard interval time length.

  Therefore, the maximum value Fcmax that can be set as the cut-off frequency of the low-pass filter 62 is expressed by the following equation (5) using the time length Lg of the guard interval.

  Here, the cut-off frequency is represented by a normalized angular frequency. Similarly to the delay amount T, the guard interval time length Lg is a relative time length when the time length of the effective symbol is J, that is, “128”. In the present embodiment, since the guard interval time length Lg is 1/8 of the effective symbol time length, Lg = 128/8 = 16 as shown in equation (5).

  In the present embodiment, the filter characteristic determination unit 9 stores a first filter coefficient for realizing a filter characteristic with a cutoff frequency = 0.25π, that is, a filter characteristic with the cutoff frequency set to the maximum value Fcmax. Has been. Further, the filter characteristic determination unit 9 realizes a filter coefficient for realizing a filter characteristic in which the cutoff frequency is set to a value smaller than the maximum value Fcmax, for example, a filter characteristic with a cutoff frequency = 0.2π. And a third filter coefficient for realizing a filter characteristic with a cutoff frequency = 0.1π is stored. The filter characteristic determination unit 9 determines one of the first to third filter coefficients as a use filter coefficient used by the low-pass filter 62.

  FIG. 16 is a flowchart showing the operation of the communication apparatus 100 regarding the determination of the used filter coefficient. As shown in FIG. 16, when the power supply to the communication device 100 is started or the operation of the communication device 100 is reset and the communication device 100 starts the operation in step s1, the filter characteristics in step s2. The determination unit 9 first determines the first filter coefficient as the use filter coefficient. That is, the first filter coefficient is used as the initial coefficient of the used filter coefficient. Thereby, the cut-off frequency of the low-pass filter 62 is set to the frequency Ft corresponding to the time length Lg of the guard interval.

  Thereafter, in step s3, the communication device 100 starts communication. At this time, in the low-pass filter 62, the first filter coefficient is used and the post-addition estimated value sequence 201 is filtered. After the communication apparatus 100 starts communication, in step s4, the maximum delay amount acquisition unit 8 obtains the maximum delay amount of the transmission path (maximum spread of the transmission path impulse response). The maximum delay amount obtained by the maximum delay amount acquiring unit 8 is a relative time length when the effective symbol time length is J, that is, “128”.

  When the maximum delay amount is obtained by the maximum delay amount acquisition unit 8, in step s5, the filter characteristic determination unit 9 determines a use cutoff frequency based on the maximum delay amount.

  Specifically, the filter characteristic determination unit 9 obtains the frequency Ft when the delay amount T is the maximum delay amount using Expression (4). And the filter characteristic determination part 9 determines the filter coefficient whose cutoff frequency is more than the calculated | required frequency Ft among the 1st-3rd filter coefficients, and is the nearest to the said frequency Ft as a use filter coefficient. . Thereby, the cut-off frequency of the low-pass filter 62 can be made as small as possible while suppressing the information on the transmission line characteristics necessary for the equalization processing from being removed from the transmission line transfer function. Therefore, transmission line noise included in the transmission path transfer function can be removed as much as possible while suppressing the removal of information necessary for equalization processing from the transmission path transfer function. As a result, equalization processing can be performed appropriately.

  For example, if the obtained frequency Ft is 0.15π, the filter characteristic determination unit 9 determines the second filter coefficient having a cutoff frequency of 0.2π as the use filter coefficient. Further, if the obtained frequency Ft is 0.23π, the filter characteristic determination unit 9 determines the first filter coefficient having a cutoff frequency of 0.25π as the use filter coefficient. Further, if the obtained frequency Ft is 0.1π, the filter characteristic determination unit 9 determines the third filter coefficient having a cutoff frequency of 0.1π as the use filter coefficient.

  The filter characteristic determination unit 9 determines the maximum delay amount obtained by the maximum delay amount acquisition unit 8 when the obtained frequency Ft is larger than the maximum value Fcmax that can be set as the cutoff frequency, that is, 0.25π. Is used, the first filter coefficient for realizing the filter characteristic with the cutoff frequency set to the maximum value Fcmax, that is, the initial coefficient of the filter coefficient is used as the use filter coefficient. Thereby, even if there is an error in the maximum delay amount obtained by the maximum delay amount acquisition unit 8, the cutoff frequency of the low-pass filter 62 is greater than the frequency Ft corresponding to the guard interval time length Lg. Can be suppressed. Therefore, it can be suppressed that the cut-off frequency of the low-pass filter 62 becomes unnecessarily large and the effect of removing the transmission line noise included in the transmission line transfer function is reduced.

  When the used filter coefficient is determined in step s5, the low-pass filter 62 performs a filtering process on the post-addition estimated value sequence 201 using the used filter coefficient. Thereafter, every time the maximum delay amount is obtained in step s4, step s5 is executed, and the filter characteristic determining unit 9 determines the filter coefficient to be used and the filter characteristic of the low-pass filter 62.

  As described above, in this embodiment, since the low-pass filter processing is performed on the transmission line transfer function, transmission line noise included in the transmission line transfer characteristic can be removed. Therefore, equalization processing can be performed appropriately by performing equalization processing using the transmission path transfer characteristics after low-pass filter processing. As a result, data included in the received signal can be appropriately acquired, and the reception performance of the communication device 100 is improved.

  Further, in the present embodiment, in the additional processing unit 61, the transmission channel primary estimation value sequence 200 (original data xr (n) and original data xi (n) for the transmission channel transfer function acquired by the transfer function acquisition unit 60 L pad values are added to the front and rear of)). When the number of taps M of the low-pass filter 62 is an odd number, L ≧ (M−1) / 2 is set, and the top of (N + 2L + M−1) values constituting the filter output signal 202 is set to the top. As a reference, N values from the (L + (M−1) / 2 + 1) th value to the (L + (M−1) / 2 + N) th value are selected and output to the equalization processing unit 4 Yes. Therefore, as shown in FIGS. 9 and 11 described above, the values of the transmission channel secondary estimated value sequence 203 (output data yr (n) and output data yi (n)) used in the equalization process are appropriate. . Therefore, equalization processing can be performed appropriately. As a result, the reception performance of the communication device 100 is improved.

  Further, when the number of taps M of the low-pass filter 62 is an even number, L ≧ M / 2 is set, and among the (N + 2L + M−1) values constituting the filter output signal 202, the head is used as a reference. N values from the (L + M / 2 + 1) th value to the (L + M / 2 + N) th value are selected and output to the equalization processing unit 4. Therefore, as in the case where the number of taps M is an odd number, the values of the transmission channel secondary estimated value sequence 203 (output data yr (n) and output data yi (n)) used in the equalization process are appropriate. Therefore, equalization processing can be performed appropriately. As a result, the reception performance of the communication device 100 is improved.

  When the number of taps M is an odd number, it is desirable to set L = (M−1) / 2. As the value of L increases, the amount of processing in the low-pass filter 62 increases. On the other hand, when L ≧ (M−1) / 2, even if the value of L increases, the transmission channel secondary estimated value sequence 203 used in the equalization processing hardly changes. Therefore, when the number of taps M is an odd number, by setting L = (M−1) / 2, the transmission path secondary used in the equalization process while reducing the processing amount in the low-pass filter 62. The value of the estimated value sequence 203 can be made appropriate. Similarly, when the number of taps M is an even number, by setting L = M / 2, the processing amount in the low-pass filter 62 is reduced, and the transmission channel secondary estimated value used in the equalization processing is reduced. The values in column 203 can be appropriate.

  In the present embodiment, the second value from the beginning (xr (n) of the N values of the original data xr (n), which is the real part of the transmission line primary estimated value sequence 200 for the transmission line transfer function. 2)) to the L values from the top to the (L + 1) th value (xr (L + 1)), the leading value (xr (1)) of the N values of the original data xr (n) , The L values of the front pad value pfr (j) are the N values of the original data xr (n) so that the L values of the front pad value pfr (j) are oddly symmetric. It is added to the front. Of the N values of the original data xr (n), the second value (xr (N−1)) from the end to the (L + 1) th value (xr (N−L)) from the end. The L values of the rear pad value pbr (j) are oddly symmetric with respect to the L values with the end value (xr (N)) of the N values of the original data xr (n) as the origin. In this way, L values of the rear pad value pbr (j) are added behind the N values of the original data xr (n). Therefore, as shown in FIG. 9 and the like described above, the value of the transmission channel secondary estimated value sequence 203 used in the equalization process is appropriate. Therefore, equalization processing can be performed appropriately. As a result, the reception performance of the communication device 100 is improved.

  In the present embodiment, the filter characteristic determination unit 9 stores a plurality of filter coefficients for realizing a plurality of types of filter characteristics having mutually different cutoff frequencies, and any one of the plurality of filter coefficients. Is determined as a use filter coefficient used by the low-pass filter 62. Therefore, each time the filter characteristic of the low-pass filter 62 is to be changed, the used filter coefficient in the low-pass filter 62 is easily changed as compared with the case of calculating the filter coefficient for realizing the changed filter characteristic. be able to.

  Although the amount of processing when changing the filter characteristic increases, the filter characteristic determination unit 9 calculates the filter coefficient for realizing the changed filter characteristic every time the filter characteristic of the low-pass filter 62 is changed. It may be calculated. That is, after determining the cutoff frequency based on the maximum delay amount acquired by the maximum delay amount acquisition unit 8, the filter characteristic determination unit 9 calculates a filter coefficient for realizing a filter characteristic having the cutoff frequency. You may do it.

DESCRIPTION OF SYMBOLS 1 Reception part 4 Equalization process part 7 Impulse response acquisition part 8 Maximum delay amount acquisition part 9 Filter characteristic determination part 60 Transfer function acquisition part 61 Additional process part 62 Low pass filter 63 Output value selection part 100 Communication apparatus

Claims (7)

A receiving unit for receiving a signal from a communication partner device;
A first acquisition unit that obtains a transmission path transfer function in a frequency domain based on a reception signal received by the reception unit;
The L second values (L) for the front and rear of the N first values (N ≧ 2) arranged in the frequency direction for the transmission line transfer function obtained by the first acquisition unit. An additional processing unit for adding ≧ 1);
The number M of taps (M) for performing low-pass filter processing on the (N + 2L) third values composed of the N first values added to the front and rear of the L second values, respectively. A low-pass filter with M ≧ 2);
N fifth values are selected from (N + 2L + M−1) fourth values output from the low-pass filter, and the selected fifth values are used as the transmission after the low-pass filter processing. A selection unit for the value of the path transfer function;
An equalization processing unit that performs an equalization process on the received signal received by the reception unit based on the value of the transmission path transfer function after the low-pass filter processing, which is obtained by the selection unit;
The additional processor is
If the number of taps M is an odd number, set L ≧ (M−1) / 2,
If the number of taps M is an even number, set L ≧ M / 2,
The selection unit includes:
When the number of taps M is an odd number, out of the (N + 2L + M−1) fourth values, the (L + (M−1) / 2 + 1) th value from the (L + (M−1) / 2 + 1) th value is used as a reference. -1) / 2 + N) up to the Nth value is selected as the N fifth values;
When the number of taps M is an even number, from the (L + M / 2 + 1) th value to the (L + M / 2 + N) th value of the (N + 2L + M-1) fourth values with reference to the head. Is selected as the N fifth values. The communication device according to claim 1,
The additional processor is
If the number of taps M is an odd number, set L = (M−1) / 2,
A communication device that sets L = M / 2 when the number of taps M is an even number. The communication device according to any one of claims 1 and 2,
The additional processor is
When adding the L second values to the front of the N first values,
Of the N real parts of the first values, the N first values are compared with the L sixth values from the second value from the beginning to the (L + 1) th value from the beginning. The real part of the L second values is the Nth first value so that the real part of the L second values is oddly symmetric with the leading value of the real part of the value as the origin. L number of seventh values from the second value from the beginning to the (L + 1) th value from the beginning of the N imaginary parts of the first value, and added to the front of the real part of the value On the other hand, the L second values are set such that the imaginary parts of the L second values are oddly symmetric with the first value of the imaginary part of the N first values as the origin. Adding an imaginary part in front of the imaginary part of the Nth first values;
When adding the L second values behind the N first values,
Of the real parts of the N first values, the N first values for the L eighth values from the second value from the end to the (L + 1) th value from the end The real part of the L second values is the N-th first value so that the real part of the L second values is odd-symmetric with the end value of the real part of the value as the origin. L-th ninth value from the second value from the end to the (L + 1) -th value from the end among the imaginary parts of the N first values and added to the rear of the real part of the value On the other hand, the L second values are set so that the imaginary part of the L second values is oddly symmetric with the end value of the imaginary part of the N first values as the origin. A communication device that adds an imaginary part to the rear of the imaginary part of the Nth first values. A communication device according to any one of claims 1 to 3,
The filter characteristic of the low-pass filter is variable,
A decision unit for determining a filter characteristic of the low-pass filter;
The determining unit stores a plurality of filter coefficients for realizing a plurality of types of filter characteristics having mutually different cutoff frequencies, and one of the plurality of filter coefficients is used by the low-pass filter. Determined as a coefficient,
The communication apparatus, wherein the low-pass filter performs the low-pass filter process using the use filter coefficient determined by the determination unit. The communication device according to claim 4,
A second acquisition unit for determining a transmission impulse response in a time domain based on the transmission line transfer function obtained by the first acquisition unit;
A third acquisition unit for determining a maximum delay amount of the transmission line based on the transmission line impulse response obtained by the second acquisition unit;
The determination unit is a communication device that determines the use filter coefficient from the plurality of filter coefficients based on the maximum delay amount obtained by the third acquisition unit. A communication device according to any one of claims 1 to 5,
The low-pass filter is a communication device that is a FIR (Finite Impulse Response) filter having a linear phase characteristic. (A) receiving a signal from the communication partner device;
(B) obtaining a transmission transfer function in the frequency domain based on the received signal received in the step (a);
(C) L second values for the front and rear of the N first values (N ≧ 2) arranged in the frequency direction with respect to the transmission line transfer function obtained in the step (b). Adding a value (L ≧ 1);
(D) For the (N + 2L) third values consisting of the N first values with the L second values added to the front and rear, respectively, the number of taps M (M ≧ 2) performing a low-pass filter process;
(E) N fifth values are selected from the (N + 2L + M−1) fourth values obtained as a result of the low-pass filter processing, and the selected N fifth values are used as the low-pass filter. A step of setting the value of the transmission line transfer function after processing;
(F) performing an equalization process on the received signal received by the receiving unit based on the value of the transmission path transfer function after the low-pass filter process obtained in the step (e). Prepared,
In the step (c),
When the number of taps M is an odd number, L ≧ (M−1) / 2 is set,
When the number of taps M is an even number, L ≧ M / 2 is set,
In the step (e),
When the number of taps M is an odd number, out of the (N + 2L + M−1) fourth values, the (L + (M−1) / 2 + 1) th value from the (L + (M−1) / 2 + 1) th value is used as a reference. -1) / 2 + N) up to the Nth value is selected as the N fifth values;
When the number of taps M is an even number, from the (L + M / 2 + 1) th value to the (L + M / 2 + N) th value of the (N + 2L + M-1) fourth values with reference to the head. Is selected as the N fifth values.

Images