JP2017143433A - Communication system and communication method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique for reducing PAPR (peak to average power ratio), in band division transmission of a single carrier modulation signal.SOLUTION: A transmitter divides a modulation signal into N subspectra and distributed on the frequency axis, performs the same phase shift at least between M subspectra where the frequency bands are adjoining or between multiple time sections, and selects and transmits such a signal that the temporal fluctuation in the signal power of the transmission signal adding N subspectra after phase shift becomes minimum. A receiver 60 demodulates the reception data by converting the subspectra SS-SS, which are extracted from the reception signal, after distribution or recovered before distribution, into a time-domain signal after correcting the phase and adding.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a communication technique using a band dispersion transmission system.

  In recent years, in order to improve the efficiency of using frequency bands in communication systems, there has been considered a band dispersion transmission method in which a single carrier modulation signal is divided into a plurality of sub-spectrums in the frequency domain and the sub-spectrums are distributed. Yes. In the band dispersion transmission method, since the spectrum is divided and distributed, there is a problem that the peak to average power ratio (PAPR) of the transmission signal increases, and a technique for reducing the PAPR has been studied. (For example, refer nonpatent literature 1). Here, a technique for reducing PAPR in communication using a band dispersion transmission scheme will be described.

FIG. 15 shows an example of a transmission apparatus 150 that uses a band dispersion transmission method. In FIG. 15, a transmission device 150 includes a modulation circuit 101, a waveform shaping filter 102, a DFT (Discrete Fourier Transform) circuit 103, division filters 104-1 to 104-N _D , and phase shifters 105-1-1 to 105-C-. N _D , frequency shifters 106-1-1 to 106-C N _D , adders 107-1 to 107-C, IDFT (Inverse DFT) circuits 108-1 to 108-C, PAPR calculation circuits 109-1 to 109 -C, a minimum PAPR signal selector 110 and a phase sequence controller 151 are provided. Here, N _D and C are positive integers.

  The modulation circuit 101 modulates transmission data by a modulation scheme such as QPSK (Quadrature Phase Shift Keying).

  The waveform shaping filter 102 is a filter for limiting the band of the modulation signal output from the modulation circuit 101.

  The DFT circuit 103 sequentially converts the modulation signal output from the waveform shaping filter 102 into a frequency domain (spectrum) signal for each predetermined DFT section. Here, in order from the earlier processing time, the “first DFT interval”, “second DFT interval”,..., “T-th DFT interval”,..., “T + u-th DFT interval”,. Called. Note that t and u are positive integers.

The division filter 104-k (1 ≦ k ≦ N _D ) is a filter for dividing the band of the modulation signal converted into the frequency domain by the DFT circuit 103 into N _D pieces. For example, the modulation signal of each DFT section transformed into the frequency domain, by splitting filter 104-1 to _{104-N D,} it is divided into _{N D} sub-modulated signal (sub spectrum). Here, “sub-spectrum of the first DFT section”, “sub-spectrum of the second DFT section”,..., “Sub-spectrum of the t-th DFT section”,. The sub-spectrum of the DFT section is called “... The output signal of each division filter 104-k is branched into C-number, respectively to the branch signals from the first to the C th phase shifter 105-q-k (1 ≦ q ≦ C, 1 ≦ k ≦ N D) Is output. Here, q, k, and C are positive integers.

FIG. 16A shows an example of a spectrum before being input to the division filter 104, and FIG. 16B shows an example of a sub-spectrum output from each division filter when N _D = 4.

The phase shifter 105-q-k (1 ≦ q ≦ C, 1 ≦ k ≦ N _D ) adds a predetermined phase to the output signal of the C divided filters 104-k. The phase shifter 105-q-k, the phase sequence theta ₂ represented by the formula _{(1), Θ 3, ...} , Θ q, ..., using a theta _C, each sub-spectrum input to the phase shifter 105-q-k Is shifted (FIG. 16C). The phase sequence controller 151, the phase sequence _{Θ 2, Θ 3, ...,} Θ q, ..., and outputs the theta _C. The phase sequence _{Θ 2, Θ 3, ...,} Θ q, ..., information theta _C is shared with the receiving apparatus 260.

Here, the sub-spectrum of the t-th DFT interval input to the phase shifter 105-q-k is SS _tk . Then, the phase shifter 105-q-k performs the phase shift of the sub-spectrum SS _{tk in} the t-th DFT section (t) shown in Expression (3) based on the phase of the phase sequence Θ _q shown in Expression (2).

The output signal S ^θ _tq of the phase shifter 105-q-k after the phase multiplication is _expressed by Expression (4).

Next, the frequency shifter 106-1-k (1 ≦ q ≦ C, 1 ≦ k ≦ N _D ) disperses each sub-spectrum in a desired band on the frequency axis.

The adder 107-q (1 ≦ q ≦ C) adds the _{N D} sub spectrum respective output signals of the frequency shifters 106-q-k for each set (each equivalent q). Here, FIG. 16D shows an example of the output signal of the adder 107-q when N _D = 4.

  IDFT 108-q (1 ≦ q ≦ C) converts the output signal of adder 107-q from a frequency domain signal to a time domain signal.

  The PAPR calculation circuit 109-q (1 ≦ q ≦ C) calculates the PAPR for the time domain signal converted by the IDFT 108-q. For example, the PAPR calculation circuit 109-q calculates a PAPR by inputting a signal corresponding to the first DFT interval output from the IDFT 108-q and outputs the PAPR to the minimum PAPR signal selector 110. Thereafter, for the second DFT interval,..., The t-th DFT interval,..., The t + u-th DFT interval,..., The PAPR of the corresponding signal is calculated similarly to the first DFT interval, and the minimum PAPR signal selector is selected. To 110.

  The minimum PAPR signal selector 110 selects and transmits the smallest PAPR signal among the PAPRs calculated by the PAPR calculation circuit 109-q for each DFT interval. For example, when the PAPR calculated by the PAPR calculation circuit 109-2 with C = 2 is the smallest, the minimum PAPR signal selector 110 selects and transmits the output signal of the IDFT 108-2 of the same series (C = 2).

FIG. 17 shows an example of a receiving apparatus 260 that uses a band dispersion transmission method. 17, the receiving device 260, DFT circuit 211, extract the filter 212-1 to _{212-N D,} the frequency shifter _{213-1~213-N D,} the phase estimator _{214-2~214-N D,} a phase 215 _{-2~215-N D,} an adder 216, IDFT circuit 217 and the demodulation circuit 218.

  The DFT circuit 211 converts the received signal into a frequency domain signal.

Extraction filter 212-k (1 ≦ k ≦ N D) is the output signal of the DFT circuit 211 which is branched to _{the N D} respectively input, extracts the _{N D} sub spectrum.

Frequency shifter 213-k (1 ≦ k ≦ N D) , the extraction filter 212-1 to _{212-N D} frequency of the transmitter 150 to each sub-spectrum is output shifter 106-q-1~106-q- N D The frequency is shifted to return to the band before the frequency shift.

Phase estimator 214-k (2 ≦ k ≦ N D) estimates the phase difference of the sub-spectrum of a signal that adjacent ones of the frequency shifter _{213-1~213-N D.} In FIG. 17, for example, the phase estimator 214-2 receives the output signals of the frequency shifter 213-1 and the frequency shifter 213-2, and is phase-shifted by the phase shifter 105-q-2 of the transmission device 150. Estimate the phase difference.

Phaser 215-k (2 ≦ k ≦ N D) is the signal phase estimator _{214-2~214-N D} respectively output to compensate for the phase of the sub-spectrum _{SS tqk.} For example, the phase shifter 215-2 compensates for the phase of the sub-spectrum SS _tq2 based on the output signal of the phase estimator 214-2.

The adder 216 adds the respective output signals of the frequency shifter 213-1 and phase shifter 215-2～ phaser _{215-N D,} return to the previous signal waveform into a plurality of sub-spectra by the transmission device 150.

  The IDFT circuit 217 inputs the output signal of the adder 216 and converts it into a time domain signal.

  The demodulation circuit 218 demodulates the modulation signal output from the IDFT circuit 217. For example, when the modulation circuit 101 of the transmission apparatus 150 modulates with QPSK, the received data is demodulated with the same QPSK.

Here, in the phase estimator 214-k of the receiving device 260, sub-spectrums (SS _tqk-1 and SS _tqk ) adjacent in the frequency domain are input, and the phase difference R in the transition region between SS _tqk-1 and SS _{tqk. tqk} is calculated. As shown in FIG. 16A, since the transition areas of the sub-spectrums in the transmission apparatus 150 overlap each other, the transition areas of adjacent sub-spectrums have the same signal component. Here, the overlapping transition area is referred to as an overlapping area. In the receiving apparatus 260, the phase difference (phase shift amount) R _tqk is estimated from the signal component in the superposed region. For example, when the sub-spectrum that does not _perform phase shift on the transmitting apparatus 150 side is the sub-spectrum (SS _tq1 ) of the lowest frequency, the phase estimator 214-2 is input by the frequency shifters 213-1 and 213-2. The phase difference R _tq2 offset by SS _tq2 is _estimated from SS _tq1 and SS _tq2 . Then, the phase estimation value (^ θ _tq2 ) = atan (R _tq2 ) is input to the phase shifter _215-2 , and the phase shifter _215-2 outputs exp (−j (^ θ _tq2) to the output signal of the frequency shifter 213-2. )) To compensate for the phase. Next, the phase estimator _214-3 calculates the phase shift amount R _tq2 from the signal SS _t2 phase-compensated by the phase shifter 215-2 and the output signal SS _t3 of the frequency shifter 213-3. The phase estimation amount (^ θ _tq3 ) = atan (R _tq3 ) is input. The phase shifter 215-3 multiplies SS _t3 output from the frequency shifter 213-3 by exp (−j (^ θ _tq3 )) using the output (^ θ _tq3 ) of the phase estimator 214-3 to _obtain the phase. To compensate.

  FIG. 18A shows an input signal of the phase shifter 215-k when k = 4, and FIG. 18B shows an output signal of the phase shifter 215-k. FIG. 18C shows the state of correlation calculation in the phase estimator 214-k.

Later, phase estimator 214-k estimates the phase difference _{R Tqk} from phaser 215- (k-1) outputted from the _{SS tqk-1} and _{SS Tqk} output from the frequency shifter 213-k, phaser The phase estimation amount (^ θ _tqk ) = atan (R _tqk ) is output to 215-k. The phase shifter 215-k performs phase compensation by multiplying SS _tqk output from the frequency shifter 213-k by exp (−j (^ θ _tqk )).

Note that the signal components of each sub-spectrum are discretized on the frequency axis by DFT processing. Here, as shown in FIG. _18C , the signal components discretized in the transition region on the low frequency side of the sub-spectrum SS _tqk from the low frequency side to the L _tqk1 , L _tqk2 ,..., L _tqkp high frequency side. _Let the signal components discretized in the transition region be H _tqk1 , H _tqk2 ,..., H _tqkp from the low frequency side. p is the number of signal components included in the transition band among the discrete signal points generated in the frequency domain by DFT processing. For example, when the frequency resolution of the DFT processing is r and the bandwidth of the transition band is B _t , p = [B _t / r]. However, the symbol [x] indicates a maximum integer not exceeding x.

The phase difference R _tqk is multiplied by the complex conjugate of the signal component on the high frequency side of the adjacent sub-spectrum SS _tk−1 and the signal component on the low frequency side of SS _tk as shown in the equation (5), and the frequency axis Obtained by smoothing in the direction.

FIG. 18C shows an example when p = 3. At this time, R _tqk is expressed by Equation (6).

  In this way, receiving apparatus 260 sequentially estimates the phase difference between adjacent sub-spectrums for each DFT interval, and compensates for the phase offset caused by the phase shift on the transmitting apparatus 150 side.

Satoshi Miyatake, Junichi Abe, Takatoshi Sugiyama “A Study on PAPR Reduction in Bandwidth Distributed Transmission System”, General Conference, IEICE, March 2015.

In the prior art, the receiving apparatus 260, which estimates the phase sequence is phase shifted by phase shifter 105-q-1~105-q- N D of the transmission device 150 from the phase difference between adjacent sub-spectrum. However, when the number p of discretized signal components existing in a low S / N environment or in the sub-spectrum transition region is not sufficient, the estimation error of the phase difference R _tqk increases. As a result, the receiving apparatus 260 has a problem that the bit error rate (BER) characteristics of the received signal deteriorate. Further, in the conventional technique, when the section to be smoothed includes a section in which power varies temporally or in frequency due to fading, an estimation error of the phase difference R _tqk increases. As a result, there arises a problem that the bit error rate characteristic of the received signal deteriorates.

  The present invention improves the phase estimation accuracy on the receiving side, and deteriorates the BER of a received signal that deteriorates in a low S / N environment or due to a lack of the number p of discrete signal components existing in the transition region of the subspectrum. An object is to provide a communication system and a communication method capable of improving characteristics.

  The communication system according to the present invention divides a modulation unit that modulates transmission data and a modulation signal output from the modulation unit into N (N is an integer of 3 or more) sub-spectra in the frequency domain, and P (P is N sub-spectrums are distributed and arranged on the frequency axis for each of the division unit that generates P transmission signal candidates by duplicating to an integer of 2 or more, and the P transmission signal candidates that are duplicated The first frequency transition unit and M (an integer of 2 ≦ M ≦ N−1) subbands whose frequency bands are adjacent to N subspectrums before or after the first frequency transition unit is distributed. A first phase unit that performs at least one of a process of performing phase shift so that the phase difference between the spectrums is equal and a process of performing the same phase shift in a plurality of time intervals; and N pieces of distributed arrangement and phase shifted Subspectra And a selection unit that selects and transmits a signal of a transmission signal candidate that minimizes the amount of temporal variation in the signal power of the transmission signal, and extracts sub-spectrums that are distributed from the received signal An extraction unit, a second frequency transition unit that returns the sub-spectrum that has been distributed to the frequency band before the distributed allocation, and a sub-spectrum after the distributed allocation extracted by the extraction unit or before the distributed allocation restored by the second frequency transition unit On the other hand, smoothing is performed by applying a predetermined load to at least one of a plurality of estimated phase differences in a superimposed region between M sub-spectrums having adjacent frequency bands and a plurality of estimated phase differences in a plurality of time intervals. The phase smoothing unit for calculating the phase correction value and the phase correction value calculated by the phase smoothing unit to the sub-spectrum after dispersion arrangement extracted from the received signal. Is a second phase unit for correcting the phase of the sub-spectrum returned before the dispersion arrangement, and a demodulating unit for demodulating the received data by adding the N sub-spectrums corrected by the second phase unit to a time domain signal after addition And a receiving device.

  In the communication method according to the present invention, the modulation signal is divided into N (N is an integer of 3 or more) sub-spectrums in the frequency domain, and is copied into P (P is an integer of 2 or more) on the transmission device side. Processing for generating P transmission signal candidates, processing for distributing N sub-spectrums on the frequency axis for each of the P transmission signal candidates, and before or after distributed allocation Process for performing phase shift so that the phase difference between M (an integer of 2 ≦ M ≦ N−1) frequency bands adjacent to each other in N sub-spectrums is equal, and a plurality of time intervals The signal of the transmission signal candidate that minimizes the amount of temporal variation in the signal power of the transmission signal obtained by adding at least one of the processes for performing the same phase shift in the above and the N sub-spectrums that are distributed and phase-shifted is added. Processing to select and transmit, on the receiving device side, processing to extract the sub-spectrum distributed from the received signal, processing to return the sub-spectrum that has been distributed to the frequency band before the distribution, and distributed arrangement Estimation of at least one of a plurality of estimated phase differences in a superimposed region between M sub-spectrums whose frequency bands are adjacent to each other and a plurality of estimated phase differences in a plurality of time intervals with respect to a sub-spectrum before or after distributed reconstruction Processing to calculate the phase correction value by smoothing the phase difference by applying a predetermined load, and correcting the phase of the sub-spectrum after dispersion placement extracted from the received signal or the sub-spectrum returned before dispersion placement by the phase correction value Processing, and adding the N sub-spectrums whose phases have been corrected, and then converting them into time domain signals to demodulate the received data And performing a process.

  In the communication system and the communication method according to the present invention, on the transmission side, phase shift is performed so that a common phase shift amount is obtained in a plurality of DFT sections, and the temporal variation of the power of the transmission signal occurs in the plurality of DFT sections. Send the smallest signal. On the receiving side, the weighted smoothing process of the phase estimation value in the time direction is performed in the same DFT interval as that on the transmitting side or a DFT interval having a length shorter than that, and within the common transition region also in the frequency direction. At least one process of performing a weighted smoothing process according to the received power intensity is performed. This improves the phase estimation accuracy on the receiving side, and improves the BER characteristics of received signals that deteriorate due to a low S / N environment or the lack of the number p of discrete signal components present in the sub-spectrum transition region. can do.

  Alternatively, the transmission side is set so that the phase difference between the sub-spectrums becomes the same value in the plurality of sub-spectrums after the phase shift. Thereby, the phase shift amount is set to the same value not in the time direction but also in the frequency direction. On the transmission side and on the reception side, a weighted smoothing process is performed for the phase differences of a plurality of subspectrums. This improves the phase estimation accuracy on the receiving side, and improves the BER characteristics of received signals that deteriorate due to a low S / N environment or the lack of the number p of discrete signal components present in the sub-spectrum transition region. can do.

In addition, when the section to be smoothed due to fading includes a section in which power fluctuates in time or frequency, the load of the corresponding section is reduced or the corresponding section is not used. Thereby, the influence of fading is reduced and the estimation error of the phase difference R _tqk can be reduced. As a result, an effect that the BER characteristic of the received signal can be improved is obtained.

  In addition, since the required sub-spectrum transition area can be reduced by improving the phase estimation accuracy, it is possible to reduce the occupied band of the spectrum and improve the frequency utilization efficiency.

It is a figure which shows an example of the transmitter which concerns on this embodiment. It is a figure which shows an example of the receiver which concerns on this embodiment. It is a figure which shows an example of the process which performs a phase shift so that the phase difference between adjacent subspectrums may become equal. It is a figure which shows an example of the transmission process which concerns on this embodiment. It is a figure which shows an example of the reception process which concerns on this embodiment. It is a figure which shows an example of the simulation result in case there is no hard decision of a phase. It is a figure which shows an example of the simulation result in case there exists a hard decision of a phase. It is a figure which shows an example of the PAPR characteristic of the transmission signal according to the smoothing number. It is a figure which shows an example of the simulation result of the misjudgment probability of the number of smoothing samples, the amount of phase shifts, and PAPR characteristic. It is a diagram illustrating an example of a computer simulation result of the number N _{D =} 16 condition of the sub-spectrum with QPSK modulation. It is a figure which shows an example of the transmitter of an application example. It is a figure which shows an example of the receiver of an application example. It is a figure which shows an example of the transmission process of an application example. It is a figure which shows an example of the reception process of an application example. It is a figure which shows an example of the transmitter of a comparative example. It is a figure which shows an example of the spectrum of the transmitter of a comparative example. It is a figure which shows an example of the receiver of a comparative example. It is a figure which shows an example of the spectrum of the receiver of a comparative example.

Embodiments of a communication system and a communication method according to the present invention will be described below with reference to the drawings.
(First embodiment)
[Transmitter 50]
FIG. 1 shows an example of a transmission device 50 in a communication system 70 according to the present embodiment. In the present embodiment, the communication system 70 includes a transmission device 50 and a reception device 60.

In FIG. 1, a transmission device 50 includes a modulation circuit 1, a waveform shaping filter 2, a DFT circuit 3, divided filters 4-1 to 4 -N _D , phase shifters 5-1-1 to 5 -C N _D , and a frequency shifter. _{6-1-1~6-C-N D,} adders 7-1 to 7-C, IDFT circuits 8-1 to 8-C, PAPR calculation circuit 9-1 to 9-C, the buffer 10-1-1 -Buffer 10-C-1, Buffer 10-1-2-Buffer 10-C-2, and Minimum PAPR signal selector 11. Here, N _D and C are positive integers.

  The modulation circuit 1 modulates transmission data by a modulation method such as QPSK.

  The waveform shaping filter 2 is a filter for limiting the band of the modulation signal output from the modulation circuit 1.

  The DFT circuit 3 sequentially converts the modulation signal output from the waveform shaping filter 2 into a frequency domain (spectrum) signal for each predetermined DFT section. Here, “first DFT interval”, “second DFT interval”,..., “T-th DFT interval”,..., “T + u-th DFT interval”,. Note that t and u are positive integers.

The division filter 4-k (1 ≦ k ≦ N _D ) divides the modulated signal band converted by the DFT circuit 3 into the frequency domain for each DFT section into N _D pieces. Dividing filter 4-1 to 4-N _D, by multiplying the preset filter coefficients to the modulated signal, to generate a N _D sub-modulated signal (sub spectrum). Each output signal of the division filter 4-k is branched into C pieces. Here, k is a positive integer. In order from the earliest processing time of the sub-spectrum, “sub-spectrum of the first DFT section”, “sub-spectrum of the second DFT section”,..., “Sub-spectrum of the t-th DFT section”,. The sub-spectrum of the DFT section is called “... The division filter 4-k operates in the same manner as in FIGS. 16A and 16B described in the related art.

The phase shifter 5-q-k (1 ≦ q ≦ C, 1 ≦ k ≦ N _D ) inputs the outputs of the C divided filters 4-k (1 ≦ k ≦ N _D ), respectively. Here, q is a positive integer. For example, the first _{N D} sub spectrum phaser 5-1-k, 2-th _{N D} sub spectrum phaser 5-2-k of the number C, ..., C-th _{N D} The sub-spectrums are respectively input to the phase shifters 5-C-k. Here, the phase sequence control device 51 outputs phase information to the phase shifter 5-C-k. The phase information output from the phase sequence control device 51 uses, for example, a randomly generated phase value or a combination of phases generated by equally dividing a circle (equally dividing 2π). Note that the phase information is shared with the receiving device 60. Then, the phase shifter 5-Ck adds a phase shift to each sub-spectrum based on the phase information input from the phase sequence control device 51.

For example, in this embodiment, the phase shifter 5-q-k, the phase sequence theta ₂ represented by the formula _{(7), Θ 3, ...} , Θ q, ..., using a theta _C phaser 5-q-k The phase of each subspectrum input to is shifted.

Here, the sub-spectrum of the t-th DFT interval input to the phase shifter 5-qk is SS _tk . Then, the phase shifter 5-q-k shifts the phase of the sub-spectrum SS _{tk in} the t-th DFT section shown in Expression (9) based on the phase of the phase sequence Θ _q shown in Expression (8).

The output signal S ^θ _tq of the phase shifter 5-qk after the phase multiplication is _expressed by Equation (10).

The frequency shifter 6-qk (1 ≦ q ≦ C, 1 ≦ k ≦ N _D ) disperses and arranges each sub-spectrum in a desired band on the frequency axis. Here, each sub-spectrum of the frequency shifter 6-2-k, each sub-spectrum of the frequency shifter 6-3-k,..., Each sub-spectrum of the frequency shifter 6-C-k is the frequency shifter 6-1-k. The frequency is shifted by the same shift amount as that of each sub-spectrum, and is distributed.

The adder 7-q (1 ≦ q ≦ C) adds the _{N D} sub spectrum output signal of the frequency shifter 6-q-k for each set (each equivalent q).

  The IDFT circuit 8-q (1 ≦ q ≦ C) converts the output signal of the adder 7-q from a frequency domain signal to a time domain signal. For example, the IDFT 8-1 converts the output signal of the adder 7-1 from a frequency domain signal to a time domain signal.

  The PAPR calculation circuit 9-q (1 ≦ q ≦ C) calculates the PAPR for the signal converted into the time domain signal by the IDFT8-q.

  Hereinafter, operations of the PAPR calculation circuit 9-q, the buffer 10-q-1 that stores the PAPR calculated by the PAPR calculation circuit 9-q, and the minimum PAPR signal selector 11 will be described. For example, the PAPR calculation circuit 9-q pays attention to the signal in the first DFT section output from the IDFT 8-q, and calculates the temporal fluctuation amount of the power. Here, in this embodiment, the ratio between the maximum power and the average power (PAPR) in a certain time interval of the signal is used as a numerical value representing the temporal variation of power. The calculated PAPR is stored in the buffer 10-q-1. Thereafter, the PAPR calculation circuit 9-q calculates PAPR up to the u-th DFT interval. Here, the PAPR calculation circuit 9-q calculates a representative value from the calculated u PAPRs according to a specific standard. In the present embodiment, as a specific standard, a value obtained by summing u ways of PAPR is used as a representative value. The PAPR calculation process and the representative value calculation are performed for all branched C signals. The minimum PAPR signal selector 11 selects and transmits the signal having the smallest PAPR among the C representative values. Here, the PAPR can be calculated by calculating the maximum value and the average value of the squared values of the input signal and calculating the maximum value / average value (value obtained by dividing the maximum value by the average value). .

In this way, the transmission apparatus 50 modulates the transmission data, divides it into a plurality of sub-spectrums, calculates the PAPR of the added signal by performing the phase shift and the frequency shift, and the phase sequence θ _q that minimizes the PAPR. Is transmitted to the receiving device 60.
[Receiver 60]
FIG. 2 shows an example of the receiving device 60 of the communication system 70 according to the present embodiment. 2, the receiving apparatus 60, DFT circuit 21, extraction filter 22-1 to _{22-N D,} the frequency shifter 23-1 to _{23-N D,} a phase difference estimator _{24-2~24-N D,} with load smoothing circuit _{25-2~25-N D,} a load calculator _{26-2~26-N D,} a buffer _{27-2~27-N D,} a load calculator _{28-2~28-N D,} smoothed with load circuit _{29-2~29-N D,} a buffer _{30-2~30-N D,} phaser _{31-2~31-N D,} adder 32, IDFT circuit 33, the demodulation circuit 34 and the phase smoothing speed controller 61 Have

Here, the receiving device 60 needs to demodulate after restoring the spectrum before the phase shift by the transmitting device 50. Therefore, the transmission device 50 fixes the phase shift amount of a specific subspectrum as a phase reference without phase shifting, and the information on the fixed subspectrum is shared between the transmission device 50 and the reception device 60. In the present embodiment, the phase shift amount of the sub-spectrum (SS _t1 ) having the lowest frequency is set to zero.

  Hereinafter, the receiving device 60 of FIG. 2 will be described.

  The DFT circuit 21 converts the reception signal from the transmission device 50 into a frequency domain signal.

Extraction filter 22-k (1 ≦ k ≦ N D) is the output signal of the DFT circuit 21 is branched into _{the N D} respectively input, extracts the _{N D} sub spectrum.

Frequency shifter 23-k (1 ≦ k ≦ N D) , the extraction filter 22-1 to _{22-N D} frequency shifter of the transmitter 50 of each sub-spectrum is outputted 6-q-1~6-q- N D Is shifted back to the band before the frequency shift.

Phase difference estimator 24-k (2 ≦ k ≦ N D) estimates the phase difference of the sub-spectrum of a signal that adjacent ones of the frequency shifters 23-1 to _{23-N D.} In FIG. 2, for example, the phase difference estimator 24-2 inputs the output signals of the frequency shifter 23-1 and the frequency shifter 23-2, and the phase shifter 5-q-2 of the transmission device 50 performs the phase shift. Estimated phase difference. A method for estimating the phase difference will be described later.

Load with the smoothing circuit 25-k (2 ≦ k ≦ N D) receives the output signal of the phase difference estimator 24-k, smoothing with load.

The load calculator 26-k (2 ≦ k ≦ N _D ) calculates a load value based on the output signal of the phase difference estimator 24-k.

The buffer 27-k (2 ≦ k ≦ N _D ) stores a signal output from the weighted smoothing circuit 25-k.

The load calculator 28-k (2 ≦ k ≦ N _D ) calculates a load value based on u-way DFT section signals stored in the buffer 27-k.

The load smoothing circuit 29-k (2 ≦ k ≦ N _D ) uses the load value calculated by the load calculator 28-k to output u-way DFT section signals stored in the buffer 27-k. Smooth with load.

The buffer 30-k (2 ≦ k ≦ N _D ) stores a signal output from the frequency shifter 23-k (2 ≦ k ≦ N _D ).

The phase shifter 31-k (2 ≦ k ≦ N _D ) is _configured from SS _tqk stored in the buffer 30-k to SS _{(t + u) qk} based on the phase shift amount output from the weighted smoothing circuit 29-k. Compensate the phase of the subspectrum.

Adder 32 restores the spectrum of the pre-divided by adding N _D sub spectrum phaser 31-k is phase compensated output.

  The IDFT circuit 33 receives the spectrum output from the adder 32 and converts it into a time domain signal.

  The demodulation circuit 34 demodulates the modulation signal output from the IDFT circuit 33 and outputs received data. For example, when the modulation circuit 1 of the transmission device 50 modulates with QPSK, the received data is demodulated with the same QPSK.

  The phase smoothing number control device 61 controls the smoothing number when the weighted smoothing circuit 29-k smoothes the phase. The smoothing number will be described later.

  In this way, the reception device 60 can demodulate the signal received from the transmission device 50. In particular, the receiving apparatus 60 according to the present embodiment estimates the phase shift amount from the phase difference of the overlapping region between adjacent subspectrums and smoothes the subspectrum returned to the band before the dispersion arrangement with a load. (Smoothing in the frequency domain). Furthermore, in the u + 1 DFT sections, the phase shift amount smoothed in the frequency domain is smoothed with a load (smoothing in the time domain). As a result, in the present embodiment, the phase estimation accuracy on the receiving side is improved and deteriorated due to a low S / N environment or a lack of the number p of discrete signal components existing in the transition region of the subspectrum. The BER characteristic of the received signal can be improved.

Next, the operation of the phase difference estimator 24-k of the receiving device 60 will be described in detail. The phase difference estimator 24-k calculates the phase difference R _tqk in the transition region of adjacent sub-spectrums (SS _tqk-1 and SS _tqk ) in the frequency domain. Here, in the transmission device 50, since the transition areas of the sub-spectrums overlap each other, the transition areas of the adjacent sub-spectrums include the same signal component. In the receiving device 60, the phase difference R _tqk is estimated from the signal component of the transition region (superimposed region) superimposed by the transmitting device 50.

Note that the phase difference estimator 24-2 is first output from the frequency shifter 23-1 and the frequency shifter 23-2 because the phase shift of the sub-spectrum SS _t1k having the lowest frequency is not performed by the transmission device 50. SS _t1k and SS _t2k are used to estimate the phase shift amount R _t2k of SS _t2k . Then, the estimated value ^ θ _tq2 = atan (R _t2k ) is smoothed by the weighted smoothing circuit 25-2, and the output is accumulated in the buffer 27-2 in u DFT intervals. Thereafter, the weighted smoothing circuit 29-2 performs a weighted smoothing process corresponding to the received power level in u DFT sections. Then, the phase shifter 31-2 receives the phase shift amount subjected to the weighted smoothing process, and multiplies the output signal of the frequency shifter 23-2 by exp (−j (^ θ _tq2 )). To compensate.

The weighted smoothing circuit 29-2 smoothes the u phase estimation values from the phase estimation values R _t2k to R _{(t + u) 2k} in the t-th to t + u-th DFT interval, or the t-th to t + u-th. Smoothing is performed in the following DFT interval, and the smoothed result is output. At this time, in the t-th to t + u-th DFT interval, when the received power intensity is different for each DFT interval, the higher the received power intensity, the better the signal-to-noise power ratio (S (signal) / N (Noise)) The load calculator 28-2 obtains a load corresponding to the received power intensity, and the load smoothing circuit 29-2 performs load smoothing.

  Here, when the received power intensity is constant, the weighted smoothing circuit 29-2 performs the calculation shown in Expression (11).

When the received power intensity is not constant, the weighted smoothing circuit 29-2 sets the load corresponding to the received power as w _tq2 , w _tq2 , w _tq2 , w _{(t + 1) q2} ,..., W _{(t + u) q2.} The calculation shown in the equation (12) is performed.

  Here, w may be a value proportional to the power intensity, for example, or may be a value proportional to the square of the power intensity.

Similarly, the phase difference estimator 24-3 receives the signal SS _tq2 phase-compensated by the phase shifter 31-2 and the output signal SS _tq3 of the frequency shifter 23-3, and estimates the phase shift amount R _tq3 . . Then, the phase shift amount smoothed by the weighted smoothing circuit 25-3 is stored in the buffer 27-3 in the DFT interval from the tth to the t + uth. The weighted smoothing circuit 29-3 smoothes the phase shift amount of the DFT section from the t-th to the t + u-th, and outputs the smoothed phase shift amount (^ θ _tq3 ) to the phase shifter 21-3. The phase shifter 31-3 uses the output of the weighted smoothing circuit 29-3 to multiply SS _tq3 output from the frequency shifter 23-3 by exp (−j (^ θ _tq3 )) to compensate the phase. .

Thereafter, the phase difference estimator 24-k (4 ≦ k ≦ N D) is a phase shifter 31- (k-1) outputted from the _{SS tq (k-1)} and SS output from the frequency shifter 23-k _The phase shift amount R _tqk is estimated from _tqk . Then, the phase shift amount smoothed by the weighted smoothing circuit 25-k is stored in the buffer 27-k in the D-th section from the t-th to the t + u-th. The weighted smoothing circuit 29-k smoothes the phase shift amount of the DFT section from the t-th to the t + u-th, and outputs the smoothed phase shift amount (^ θ _tqk ) to the phase shifter 21-k. Then, the phase shifter 31-k outputs exp (−j (^ θ _tqk ) to SS _tqk output from the frequency shifter 23-k using the output of the weighted smoothing circuit 29-k and accumulated in the buffer 30-k. )) To compensate for the phase.

Here, the signal components of each sub-spectrum are discretized on the frequency axis by DFT processing. Further, as described in FIG. _18C of the prior art, the signal components discretized in the transition region on the low frequency side of the sub-spectrum SS _tqk are set to L _tqk1 , L _tqk2 ,..., L _tqkp from the low frequency side. , H _tqk1 , H _tqk2 ,..., H _tqkp are signal components discretized in the transition region on the high frequency side from the low frequency side. However, p is the number of signal components included in the band of the transition region among the discretized signal points generated by the DFT processing. For example, when the frequency resolution of the DFT is r and the bandwidth of the transition region is B _t , p = [B _t / r]. Here, the symbol [x] indicates a maximum integer not exceeding x.

The phase difference R _tqk is obtained by multiplying the complex component of the signal component on the high frequency side and the signal component on the low frequency side of the adjacent sub-spectrums SS _tqk−1 and SS _tqk and smoothing in the frequency axis direction. Here, when there is a difference in power intensity of discretized signal components (L _tqk1 , L _tqk2 ,..., L _tqkp, etc.) existing in the same transition region, the signal-to-noise power ratio (S / N) is considered to have a high probability, and a load corresponding to the received power intensity is set to perform smoothing with a load.

  Equation (13) shows the smoothing process when the weighted average is not used.

In the case of p = 3, R _tqk is expressed by Expression (14).

Further, when the load value is v _tqki , the smoothing process in the case of using the weighted average is expressed by Expression (15).

Here, v _tqki may be a value proportional to the power intensity, for example, or may be a value proportional to the square of the power intensity.

  In this way, the receiving apparatus 60 can sequentially estimate the phase difference between adjacent sub-spectrums for each DFT section, and can compensate for the phase offset generated by the phase shift process on the transmitting apparatus 50 side.

  Next, an example will be described in which the phase shift is performed so that the phase differences between a plurality of adjacent sub-spectrums are equal in the transmission device 50 according to the present embodiment.

FIG. 3 shows an example in which the phase shift is performed so that the phase difference between two adjacent sub-spectrums becomes equal. Here, it is assumed that the number of spectrum divisions N _D = 8 and the number of adjacent sub-spectrums M = 2. The process of performing the phase shift so that the phase differences between adjacent sub-spectrums are equal is performed in each DFT section of a plurality of DFT sections (for example, eight DFT sections). In the example of FIG. Processing is performed on eight sub-spectrums in the section. Here, M represents the number of “sub-spectrums to which the same phase shift (phase rotation) is added”, and M is an integer of 2 ≦ M ≦ N _D −1. For example, in FIG. 3, the phase shift is performed so that the phase difference between the two sub-spectrum of the sub-spectrum 3 and the sub-spectrum 4 is the same [Delta] [theta] _1. Similarly, the phase difference between the two sub-spectrums of sub-spectrum 1 and sub-spectrum 2 is 0, the phase difference between the two sub-spectrums of sub-spectrum 5 and sub-spectrum 6 is Δθ ₂ , sub-spectrum 7 and sub-spectrum The phase shift is performed so that the phase difference between the two sub-spectrums of 8 is Δθ ₃ .

  As described above, the transmission device 50 according to the present embodiment performs the phase shift so that the phase differences between the M sub-spectrums having adjacent frequency bands are equal.

Further, in the same DFT section, the transmission apparatus 50 has equal phase differences between M (2 ≦ M ≦ N _D −1) adjacent sub-spectrums with the amount of phase shift given to the phase shifter 5-q−k. Thus, the receiving device 60 simply calculates the phase difference in the superposed region of M subspectrums based on the value of the number M of subspectrums to be smoothed output from the phase smoothing number control device 61. Smoothing with average or load. Thereby, the phase estimation accuracy on the receiving side can be improved. Here, the load when performing the smoothing process with a load can be set according to the received power intensity, for example. In this case, the load is increased as the received power intensity is increased, and conversely, the load is decreased as the received power intensity is decreased.

Further, all phase shift amount values θ _qk used on the transmission side are shared in advance between the transmission device 50 and the reception device 60, and the phase shift closest to the phase estimation value on the reception side is determined by a hard phase determination. By performing phase compensation using the quantity value θ _qk , the phase estimation accuracy in the receiving device 60 can be improved.

  Further, in the configuration of the transmission device 50 shown in FIG. 1, the frequency shifter 6-qk is not provided between the phase shifter 5-qk and the adder 7-q, but is divided between the division filter 4-k and the phase shifter 5. The phase shifter 5-q-k may be phase-shifted with respect to each sub-spectrum after the sub-spectrum is arranged in a distributed manner by the frequency shifter 6-q-k.

  Further, the information indicating the length of the DFT section for which the transmission device 50 calculates the PAPR needs to be known on the reception device 60 side. For example, when the transmission device 50 and the reception device 60 start communication, the information is transmitted. The information may be notified from the device 50 to the receiving device 60, or the receiving device 60 may estimate from the received signal.

In addition, in the configuration of the receiving device 60 illustrated in FIG. 2, the frequency shifter 23-k is not between the extraction filter 22-k and the phase difference estimator 24-k but the phase shifter 31-k and the adder 32-k. You may arrange | position between.
[Transmission process]
FIG. 4 shows an example of transmission processing in the present embodiment. Note that the transmission processing in FIG. 4 is performed by the transmission device 50.

  In step S101, a variable t (t is a positive integer) indicating the time interval of the time domain signal is initialized to t = 1. The time interval t can be handled in the same manner as the variable t indicating the order of the DFT intervals processed by the DFT circuit 3. For example, in the time interval t, the order of DFT intervals for DFT processing is t.

  In step S <b> 102, transmission data in the time interval t is input to the modulation circuit 1.

  In step S103, the modulation circuit 1 modulates the input transmission data in the time interval t by a modulation method such as QPSK.

  In step S104, the waveform shaping filter 2 adjusts the waveform of the modulation signal output from the modulation circuit 1 using a window function or the like, and the DFT circuit 3 converts the waveform into a frequency domain signal.

In step S105, the _{N D} divided filter 4-k divides the spectrum of the modulated signal DFT circuit 3 is converted to the frequency domain to the _{N D} sub spectrum.

  In step S106, a variable q indicating C phase sequences from 1 to C is initialized to q = 1.

In step S107, the phase shifter 5-q-k makes the phase difference between the adjacent M sub-spectrums equal based on the phase sequence control value (phase sequence θ _q ) given from the phase sequence control device 51. Phase shift is performed. Note that the phase sequence control device 51 shares information of the phase sequence θ _q with the receiving device 60.

  In step S108, the frequency shifter 6-qk shifts the frequency of each subspectrum after the phase shift output from the phase shifter 5-qk. Note that the frequency to be shifted is determined in advance.

In step S109, the adder 7-q adds the _{N D} sub _spectrum the _{N D} frequency shifter 6-q-k respectively output the spectrum after the addition, the time domain by IDFT circuit 8-q Is converted into a signal. The time domain signal converted by the IDFT circuit 8-q is input to the PAPR calculation circuit 9-q, and the time domain signal for each phase sequence θ _{q and for} each time interval t is buffer 10-. stored in q-2.

  In step S110, the PAPR calculation circuit 9-q calculates the PAPR from the signal in the time domain of the DFT section t output from the IDFT circuit 8-q.

In step S111, it stores the PAPR at the time interval t of the phase sequence theta _q in the buffer 10-q-1. For example, the PAPR in the time interval t = 1 of the phase sequence θ ₁ is stored in the buffer 10-1-1. Similarly the phase sequence theta ₂ for phase sequence theta _C, PAPR in the time interval t = 1 are stored respectively from the buffer 10-2-1 in the buffer 10-C-1. Similarly, the time interval t = 2 from the time interval t = u + 1, each of the PAPR from the phase sequence theta ₁ with respect to the phase sequence theta _C are respectively stored from the buffer 10-1-1 in the buffer 10-C-1 . In this way, each PAPR calculated for each phase series θ _{q and for} each time interval t is stored in the buffer 10-q-1.

  In step S112, it is determined whether or not a series of processing from step S107 to step S111 has been completed for all the phase sequences q from 1 to C. If the series of processes has been completed up to the Cth of the phase series q, the process proceeds to step S113, and if not, the process proceeds to step S114.

  In step S113, it is determined whether or not the processing from step S102 to step S112 has been completed for each time interval (DFT interval) t from 1 to (u + 1) th, and a series of processing has been completed to (u + 1) th. If this is the case, the process proceeds to step S115, and if not completed to the (u + 1) th, the process proceeds to step S116.

  In step S114, the phase sequence (q + 1) to be processed next is set.

When the phase sequence θ _C and the time interval t = u + 1 are completed in step S115, the minimum PAPR signal selector 11 finishes storing the PAPRs up to q = C and t = u + 1 in the buffer 10-q−1. , it extracts the largest PAPR in u + 1 pieces of PAPR which is phase-shifted by the phase sequence θ _{q (q} = 1~C) as a representative value of PAPR in the phase sequence θ _{q (q} = 1~C). For example, the PAPR of the phase sequence theta ₂ of the time interval t = 3 if the maximum, to extract the PAPR of phase sequence theta ₂ of the time interval t = 3 as the representative value. Similarly, the largest PAPR among u + 1 PAPRs for each phase sequence from the phase sequence θ ₁ to θ _C is extracted as a representative value of the PAPR in the phase sequence θ _q . Then, in the time interval t corresponding to the phase sequence θ _q corresponding to the smallest PAPR among the representative values of the PAPR for each phase sequence from the phase sequence θ ₁ to θ _C held in the buffer 10-q-1. The signal is read from the buffer 10-1-2 and output.

  In step S116, the next time interval (t + 1) to be processed is set.

In this way, the transmission apparatus 50 modulates the transmission data, divides it into a plurality of sub-spectrums, calculates the PAPR of the added signal by performing the phase shift and the frequency shift, and the phase sequence θ _q that minimizes the PAPR. Is transmitted to the receiving device 60.

  FIG. 5 shows an example of reception processing in the present embodiment. 5 is performed by the receiving device 60.

In step S201, DFT circuit 21, a reception signal received from the transmitting apparatus 50 into a signal in the frequency domain, by the _{N D} extraction filter 22-k, the _{N D} sub spectrum divided by the sender Extract.

  In step S202, the frequency shifter 23-k returns each sub-spectrum output by the extraction filter 22-k to a band before the frequency shifter 6-q-k of the transmission device 50 performs frequency shift and disperses the sub-spectrum.

  In step S <b> 203, the phase difference estimator 24-k performs the phase difference of M−1 transition regions (superimposed regions) between adjacent M sub-spectrums with respect to the sub-spectrum returned to the band before the dispersion arrangement. From this, the phase shift amount is estimated. The value of M is shared by the transmission device 50 and the reception device 60 as a phase sequence control value and is set in advance.

  In step S204, the load calculator 26-k calculates a load when the phase shift amount is smoothed.

  In step S205, the weighted smoothing circuit 25-k smoothes the phase shift amount in the overlapping region with the weight.

  In step S206, the phase shift amount smoothed by the weighted smoothing circuit 25-k is stored in the buffer 27-k.

  In step S207, the load calculator 28-k calculates the phase shift amount load in the u + 1 DFT sections stored in the buffer 27-k.

  In step S208, the weighted smoothing circuit 29-k smoothes the phase shift amount in the u + 1 DFT interval with the weight based on the load output from the load calculator 28-k.

  In step S209, the phase shifter 31-k compensates the phase of each subspectrum held in the buffer 30-k by the smoothed phase shift amount output from the weighted smoothing circuit 29-k.

In step S210, the adder 32 are synthesized by adding the _{N D} sub _spectrum the _{N D} phase shifters 31-k outputs. After that, the IDFT circuit 33 converts the frequency domain spectrum output from the adder 32 into a time domain signal.

  In step S211, the demodulation circuit 34 demodulates the time domain signal converted by the IDFT circuit 33 into reception data.

  In this way, the receiving device 60 can demodulate the received data. In particular, the receiving apparatus 60 according to the present embodiment estimates the phase shift amount from the phase difference of the superposed region between adjacent subspectrums of the subspectrums returned to the band before the dispersion arrangement, and smoothes with a load ( Smoothing in the frequency domain). Furthermore, in the u + 1 DFT sections, the phase shift amount smoothed in the frequency domain is smoothed with a load (smoothing in the time domain). Thereby, in this embodiment, the receiving device 60 improves the estimation accuracy of the phase shift amount performed by the transmitting device 50, and is discretized in a low S / N environment or in a sub-spectrum transition region. It is possible to improve the BER characteristic of a received signal that deteriorates due to an insufficient number p of signal components.

  Next, the effect in this embodiment is demonstrated using FIGS. 6-10.

FIG. 6 shows an example of a phase estimation error (FIG. 6 (a)) and a BER characteristic (FIG. 6 (b)) when the receiving side does not make a hard phase determination. Here, in FIG. 6, E _b / N _o represents the ratio of energy per bit to noise power density. BER (Bit Error Rate) represents the ratio of the number of bits received in error to the number of transmitted bits. In the computer simulation shown in FIG. 6, when the number of sub-spectrums N _D = 8 in QPSK modulation, the same phase shift is added in u + 1 DFT intervals on the transmission side, and u + 1 DFT intervals on the reception side. Shows the result when the phase estimation value is smoothed.

In FIG. 6A, it can be seen that the phase estimation error at E _b / N _o = 0 dB decreases as the number of smoothed DFT sections (smoothing number) increases. Also, in FIG. 6B, it can be seen that the BER characteristics become lower as the number of DFT sections to be smoothed increases. That is, the BER characteristic is improved due to the reduction of the phase estimation error. Here, the phase estimation error is the difference between the phase estimation value calculated in each sub-spectrum (sub-spectrum 1 to sub-spectrum 8 in the example of FIG. 3) and the phase shift amount given on the transmission side in all DFT sections. Represents the average difference.

  Next, simulation results when the phase shift is performed so that the phase differences between the sub-spectrums with M = 2 are equal will be described with reference to FIGS.

FIG. 7 shows a simulation result when a hard decision of the phase is performed on the receiving side. Note that all phase shift amounts θ _qk used on the transmission side are shared in advance with the reception side, and phase compensation is performed using the phase shift amount θ _qk closest to the phase estimation value on the reception side by hard phase determination. . FIG. 7A shows the erroneous determination probability of the phase shift amount, FIG. 7B shows the phase estimation error, and FIG. 7C shows the BER characteristic. Here, in FIG. 7, when performing N _D -1 times of the phase estimated by the N _D sub spectrum 1DFT section, erroneous determination probability of the phase shift amount in the N D _-1 amino phase estimates It represents the probability of occurrence of misjudgment in a plurality of DFT sections when it is considered that there has been a misjudgment for a DFT section with one or more errors. When FIG. 6A and FIG. 7B are compared, when (the number of smoothed DFT sections) = 1, the phase estimation error is reduced to about 1 / 1.07 by performing a hard decision. I understand. It can also be seen that when (the number of smoothed DFT sections) = 16, the hard decision is reduced to about 1/240. That is, the effect of the hard decision of a phase becomes high, so that the smoothing number of a DFT area increases. Regarding the erroneous determination probability of the phase shift amount shown in FIG. 7A as well as the phase estimation error, the erroneous determination probability of the phase shift amount decreases as the smoothing number in the DFT interval increases. Further, as shown in FIG. 7C, it can be seen that the BER characteristic is improved by increasing the smoothing number in the DFT section and improving the phase estimation accuracy.

  FIG. 8 shows an example of the PAPR characteristic of the transmission signal according to the smoothing number, as in FIGS. 6 and 7. Here, CCDF (Complementary Cumulative Distribution Function) represents the probability that the PAPR will be a predetermined value or more. In FIG. 8, it can be seen that as the number of smoothing in the DFT interval increases, the PAPR also increases, and asymptotically approaches the characteristics when no phase shift is performed.

  FIG. 9 shows changes in the PAPR characteristic and the misjudgment probability of the phase shift amount for each number of averaged samples. In FIG. 9, the phase estimation accuracy and the BER characteristics are significantly improved when smoothing is performed with 8 DFT sections as compared with the case of smoothing with one DFT section as described in the prior art. Recognize. Furthermore, compared with the case where only smoothing is performed in 8 DFT sections, the phase difference between M sub-spectrums which are smoothed in 8 DFT sections and which are adjacent (M = 2 in the example of FIG. 9) is equal. When phase shifting is performed, the PAPR slightly increases (about 0.3 dB), but the erroneous determination probability of the phase is greatly reduced (1/5 times), and the BER characteristics are greatly improved. Can do.

FIG. 10 shows an example of a computer simulation result under the condition of QPSK modulation and the number of subspectrums N _D = 16. Note that FIG. 10 illustrates a case where the phase is smoothed in two DFT sections and the smoothing between the sub-spectrums is not performed, and the phase smoothing (M = 2) is performed in the two sub-spectrums. In addition, a result of comparing the PAPR characteristic (FIG. 10A) and the BER characteristic (FIG. 10B) when smoothing is not performed in a plurality of DFT sections is shown. In addition, in FIG. 10, each characteristic, such as a prior art and a theoretical value, is also shown for the comparison.

  From the result of FIG. 10A, smoothing in a plurality of DFT sections is not performed, and the characteristic D of the prior art that does not perform smoothing between sub-spectrums and smoothing in a plurality of DFT sections are not performed. Further, comparing with the characteristic B that smoothes the phase (M = 2) in the two sub-spectrums, it can be seen that the PAPR degradation amount due to the smoothing in the sub-spectrum direction is about 0.1 dB. Also, smoothing between sub-spectrums is not performed, and comparing the characteristics A and characteristics B smoothed in two DFT sections, the PAPR of the characteristics B is reduced by about 0.2 dB more than the characteristics A Has been.

  From the result of FIG. 10 (b), the characteristic D of the prior art which does not perform smoothing in a plurality of DFT sections and does not perform smoothing between sub-spectrums is deteriorated by 3 dB or more compared to the theoretical BER characteristic. However, it can be seen that the BER characteristics can be improved by applying the phase smoothing method of the characteristics A and B. In particular, in the result of FIG. 10B, the characteristic B has a good characteristic that is asymptotic to the characteristic C of the theoretical value.

From the simulation results of FIG. 10A and FIG. 10B described above, when smoothing in the DFT direction is performed (characteristic A), and when smoothing between a plurality of sub-spectrums is performed (characteristic B). In the PAPR and PAPR characteristics, it is understood that the performance of the phase smoothing method of characteristic B is superior to that of the phase smoothing method of characteristic A.
(Application examples)
As an application example of the first embodiment, a configuration in which a preamble is added by the transmission device 50 illustrated in FIG. 1 and channel estimation based on the preamble is performed by the reception device 60 will be described.

  FIG. 11 shows an example of the transmission device 50a of this application example. In FIG. 11, blocks having the same symbols as those of the transmission device 50 illustrated in FIG. 1 have the same or similar functions as the transmission device 50. 1 differs from FIG. 1 in that a preamble addition circuit 12-q is added between the adder 7-q and the IDFT circuit 8-q, and that the preamble addition circuit is provided after the minimum PAPR signal selector 11. 13 is added.

  The preamble adding circuit 12-q adds a preamble in the frequency domain to the spectrum signal output from the adder 7-q. For example, in the preamble adding circuit 12-q, the amplitude and phase information are known on the receiving device 60a side in the spectrum output from the adder 7-q (the transmitting device 50a and the receiving device 60a share in advance). A frequency domain waveform is added as a preamble. Then, the receiving device 60a compares the amplitude and phase information as known information with the amplitude and phase information of the received preamble signal, and multiplies the received signal by the inverse characteristic of the distortion characteristic obtained from the comparison result. Thus, it is possible to compensate for the amplitude and phase distortion in the frequency domain in the transmission path.

  The preamble adding circuit 13 adds a preamble to the output signal of the minimum PAPR signal selector 11 in the time domain. For example, the preamble adding circuit 13 adds a time-domain waveform whose amplitude and phase information in the time domain is known on the receiving apparatus 60a side (shared in advance between the transmitting apparatus 50a and the receiving apparatus 60a) as a preamble. . Then, the receiving device 60a compares the amplitude and phase information, which is known information, with the amplitude and phase information of the received preamble signal, and obtains a temporal change amount of the amplitude and phase obtained from the comparison result. By multiplying the reception signal by the inverse characteristic, it is possible to compensate for time domain amplitude and phase distortion in the transmission path.

  In this way, the transmission device 50a according to this application example adds a preamble to both the frequency domain and the time domain and transmits the preamble to the reception device 60a, and the reception device 60a transmits the time domain in the transmission path. Amplitude and phase distortion can be compensated.

  FIG. 12 shows an example of the receiving device 60a of this application example. In FIG. 12, blocks having the same symbols as those of the receiving device 60 illustrated in FIG. 2 have the same or similar functions as the receiving device 60. 2 is different from FIG. 2 in that a channel estimation circuit 35 and a channel compensation circuit 36 are added before the DFT circuit 21, and a channel estimation circuit 37 and a channel between the DFT circuit 21 and the extraction filter 22. The compensation circuit 38 is added.

  Based on the preamble added in the time domain by the preamble addition circuit 13 of the transmission device 50a, the channel estimation circuit 35 estimates the amount of time-selective amplitude / phase variation that occurs in the propagation path, and outputs it as a channel estimation value. The channel estimation circuit 35 also outputs the channel estimation value to the load calculator 28-k, and the load calculator 28-k calculates a load according to the amount of time selective amplitude / phase variation. For example, the load calculator 28-k decreases the load when the time-selective amplitude / phase variation is large, and conversely increases the load when the variation is small.

  The channel compensation circuit 36 compensates for the time-selective amplitude / phase variation estimated by the channel estimation circuit 35. The time domain signal compensated by the channel compensation circuit 36 for amplitude and phase fluctuations is converted by the DFT circuit 21 into a frequency domain spectrum.

  The channel estimation circuit 37 estimates the amplitude / phase fluctuation amount of the frequency selectivity generated in the propagation path based on the preamble added in the frequency domain by the preamble addition circuit 12-q of the transmission device 50a, and outputs it as a channel estimation value. To do. The channel estimation circuit 35 also outputs the channel estimation value to the load calculator 26-k, and the load calculator 26-k calculates a load corresponding to the frequency-selective amplitude / phase variation. For example, the load calculator 26-k decreases the load when the amplitude / phase variation amount of the frequency selectivity is large, and conversely increases the load when the variation amount is small.

The channel compensation circuit 38 compensates for the amplitude and phase fluctuation amount of the frequency selectivity estimated by the channel estimation circuit 37. The spectrum of the frequency domain channel compensation circuit 38 compensates for the amount of variation of amplitude and phase, N _D sub spectrum is extracted by the extracting filter 22-k.

  In this way, the receiving device 60a according to this application example, based on both the frequency domain and time domain preambles added by the transmitting device 50a, the time selectivity generated in the propagation path and the frequency selectivity amplitude · The amount of phase variation can be estimated and compensated. Furthermore, since the receiving device 60a adjusts the load for smoothing the phase difference based on both the frequency domain and time domain preambles added by the transmitting device 50a, the time selectivity generated in the propagation path and It is possible to reduce the load when the amplitude / phase variation amount of the frequency selectivity is large, and to reduce the error of the calculated phase difference.

  FIG. 13 shows an example of the transmission process of this application example. In FIG. 13, a process having the same reference sign as the transmission process shown in FIG. 4 indicates the same or similar process as in FIG. 4. Further, the difference from FIG. 4 is that step 117 is added after step 115.

In step S117, the transmission device 50a transmits the signal having the minimum PAPR in the phase sequence θ _q (q = 1 to C) with the preamble added in the time domain and the frequency domain. Specifically, as described with reference to FIG. 11, the preamble adding circuit 12-q adds a preamble in the frequency domain, and the preamble adding circuit 13 adds a preamble in the time domain.

  In this way, the transmission device 50a according to this application example adds the preamble to both the frequency domain and the time domain of the transmission signal of the transmission device 50 and transmits the result to the reception device 60a.

  FIG. 14 shows an example of reception processing of this application example. In FIG. 14, the process having the same reference sign as the reception process shown in FIG. 5 indicates the same or similar process as in FIG. 5. Further, the difference from FIG. 5 is that step 200 is added before step 201.

  In step S200, the receiving apparatus 60a estimates and compensates for the time-selective or frequency-selective amplitude / phase channel fluctuation amount generated in the propagation path before extracting the spectrum divided in step S201. The receiving device 60a outputs the channel fluctuation amount estimated in step S200 as a reference value for the load calculation processing performed in steps S204 and S207, and calculates a load based on the channel fluctuation amount in steps S204 and S207. .

  In this way, the receiving device 60a according to this application example, based on both the frequency domain and time domain preambles added by the transmitting device 50a, the time selectivity generated in the propagation path and the frequency selectivity amplitude · The amount of phase variation can be estimated and compensated. Further, the load for smoothing the phase difference is adjusted based on both the frequency domain preamble and the time domain preamble added by the transmission device 50a. For example, the receiving device 60a can reduce the error of the calculated phase difference by reducing the load when the amplitude / phase variation amount of the time selectivity and frequency selectivity generated in the propagation path is large.

  As described above, the communication system and the communication method according to the present invention improve the phase estimation accuracy on the receiving side, and are made discrete in a low S / N environment or in a sub-spectrum transition region. It is possible to improve the BER characteristic of a received signal that deteriorates due to an insufficient number p of signal components.

50, 50a, 150 ... transmitting device; 60, 60a, 260 ... receiving device; 1, 101 ... modulation circuit; 2, 102 ... waveform shaping filter; 3, 21, 103, 211 ... DFT circuit; 4,104: division filter; 5, 31, 105, 215 ... phase shifter; 6, 23, 106, 213 ... frequency shifter; 7, 32, 107, 216 ... addition 8, 33, 108, 217 ... IDFT circuit; 9, 109 ... PAPR calculation circuit; 10, 27, 30 ... buffer; 11, 110 ... minimum PAPR signal selector; ... Preamble addition unit; 22, 212 ... Extraction filter; 24 ... Phase difference estimator; 25, 29 ... Loaded smoothing circuit; 26, 28 ... Load calculator; ... Demodulation circuits; 35, 3 ... channel estimation circuit; 36,38 ... channel compensation circuit; 51 ... phase sequence controller; 61 ... phase smoothing speed controller; 214 ... phase estimator

Claims (8)

A modulator for modulating transmission data;
The modulation signal output from the modulation unit is divided into N (N is an integer of 3 or more) sub-spectrums in the frequency domain, and duplicated into P (P is an integer of 2 or more) to obtain P transmission signal candidates. A dividing unit to be generated;
A first frequency transition unit that disperses and arranges the N sub-spectrums on the frequency axis for each of the P transmission signal candidates replicated;
A phase difference between M (integers of 2 ≦ M ≦ N−1) frequency bands adjacent to the N sub-spectrums before or after the first frequency transition unit is distributed. A first phase unit that performs at least one of a process of performing phase shift so as to be equal and a process of performing the same phase shift in a plurality of time intervals;
And a selection unit that selects and transmits the signal of the transmission signal candidate that minimizes the amount of temporal variation in the signal power of the transmission signal obtained by adding the N sub-spectrums that are dispersedly arranged and phase-shifted. When,
An extractor for extracting sub-spectrums distributed from the received signal;
A second frequency transition unit for returning the dispersed sub-spectrum to the frequency band before the dispersed arrangement;
A plurality of estimated phase differences in a superposed region between M sub-spectrums whose frequency bands are adjacent to each other after the dispersed arrangement extracted by the extraction unit or before the distributed arrangement restored by the second frequency transition unit, and A phase smoothing unit that calculates a phase correction value by smoothing at least one estimated phase difference of a plurality of estimated phase differences in a plurality of time intervals by applying a predetermined load;
A second phase unit that corrects the phase of the sub-spectrum after dispersion placement extracted from the received signal or the sub-spectrum returned before dispersion placement, based on the phase correction value calculated by the phase smoothing unit;
And a demodulator that demodulates received data by adding the N sub-spectrums corrected by the second phase unit to a signal in the time domain. The communication system according to claim 1,
The selection unit of the transmission device, when the division unit performs the process of sequentially dividing into sub-spectrums for each fixed time interval, temporal variation of power in u + 1 (u is a positive integer) time interval Calculating a transmission signal candidate with the smallest amount of temporal fluctuation of the power from the P replica signals, and transmitting,
The phase smoothing unit of the receiving device includes a plurality of signals in a superposed region of adjacent sub-spectrums with respect to the sub-spectrums after the dispersed placement extracted by the extracting unit or before the dispersed placement restored by the second frequency transition unit. A process of calculating a plurality of phase differences of components, applying a load corresponding to the received power intensity within the same overlapping region to smooth the frequency direction, and converting the received phase intensity into u + 1 time intervals A communication system, wherein the phase correction value is calculated by performing at least one process of performing a smoothing in a time direction by applying a corresponding load. The communication system according to claim 1 or 2,
The phase smoothing unit of the receiving device smoothes a phase difference between adjacent M sub-spectrum overlapping regions by a simple average, or increases a signal load higher than a signal load with low received power intensity. A communication system, characterized in that the phase correction value is calculated by performing a process of setting and smoothing the phase difference of the superposed areas of the M subspectrums under a load. The communication system according to any one of claims 1 to 3,
The receiving apparatus further includes a channel estimation unit that estimates the amount of variation in amplitude and phase of frequency selectivity or time selectivity generated in a communication channel of a propagation path,
The said phase smoothing part performs the process which changes and smoothes the said load according to the said variation | change_quantity estimated by the said channel estimation part. The communication system characterized by the above-mentioned. On the transmitting device side, the modulation signal is divided into N (N is an integer of 3 or more) sub-spectrums in the frequency domain, and duplicated into P (P is an integer of 2 or more) to generate P transmission signal candidates. Processing to
A process of distributing and arranging the N sub-spectrums on the frequency axis for each of the P transmitted signal candidates,
Phase shift so that the phase difference between M (an integer of 2 ≦ M ≦ N−1) adjacent frequency bands is equal to the N sub-spectrums before or after the distributed arrangement And at least one of a process for performing the same phase shift in a plurality of time intervals, and
Performing a process of selecting and transmitting the signal of the transmission signal candidate that minimizes the amount of temporal variation in the signal power of the transmission signal obtained by adding the N sub-spectrums that are dispersed and phase-shifted;
On the receiving device side, a process of extracting the sub-spectrum distributed from the received signal;
Processing to return the sub-spectrum that has been distributed to the frequency band before the distribution,
At least one of a plurality of estimated phase differences in a superposed region between M sub-spectrums whose frequency bands are adjacent to each other and a plurality of estimated phase differences in a plurality of time intervals with respect to a sub-spectrum after distributed arrangement or before restored distributed arrangement A process of calculating a phase correction value by applying a predetermined load to the estimated phase difference and smoothing,
The phase correction value corrects the phase of the subspectrum after dispersion placement extracted from the received signal or the phase of the subspectrum returned before dispersion placement;
And a process of demodulating received data by adding N sub-spectrums that have undergone phase correction to a signal in a time domain. The communication method according to claim 5, wherein
When the transmission unit performs the process of sequentially dividing the sub-spectrum into sub-spectrums for each fixed time interval, the transmission device calculates a temporal variation amount of power in u + 1 (u is a positive integer) time interval, Performing a process of selecting and transmitting a transmission signal candidate having the smallest amount of temporal variation of the power from the P replica signals;
The receiving apparatus calculates a plurality of phase differences of a plurality of signal components in a superimposed region of adjacent sub-spectrums after the dispersed arrangement or before the restored dispersed arrangement, and the plurality of phase differences are superimposed on the same At least one of processing for smoothing in the frequency direction by applying a load corresponding to the received power intensity in the region and processing for smoothing in the time direction by applying a load corresponding to the received power intensity in u + 1 time intervals A communication method comprising: performing a process to calculate the phase correction value. The communication method according to claim 5 or 6,
The receiving apparatus performs processing for smoothing the phase difference between the overlapping regions of the M sub-spectrums adjacent to each other by a simple average, or sets the signal load higher than the signal load with low received power intensity to M A communication method characterized by calculating a phase correction value by performing a process of applying a load to the phase difference in the superspected region of the sub-spectrum and smoothing it. The communication method according to any one of claims 5 to 7,
The receiving apparatus estimates a frequency-selective or time-selective amplitude and phase fluctuation amount that occurs in a communication channel of a propagation path, and performs a process of changing and smoothing the load according to the fluctuation amount. A communication method characterized by the above.

Images