JP2003338779A - MIMO receiver and receiving method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To raise the accuracies of the equalizing start timing, the equalizing range and the channel estimation in the case of many users. <P>SOLUTION: An equalizing start timing t0, an equalizing range te and channel H estimation are made every user signal to generate received signal replicas Re<SB>1</SB>-ReN of each user. These replicas are subtracted from the received signals to take error signals Es. The replicas Re<SB>1</SB>-RnN are added to the Es to take interference-suppressed received signals of users 1-N, respectively, and t0, te, H, Re<SB>1</SB>-ReN are generated about them to further make Es. Similar steps are repeated to increase the accuracies of t0, te, H. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a reception signal in which signals from integer N transmitters of 2 or more, which transmit at the same time using the same frequency band, are received by antennas M of 2 or more integers. On the other hand, the present invention relates to a multi-input multi-output (MIMO) receiver and a receiving method for performing a decoding process after performing an adaptive equalization process.

[0002]

2. Description of the Related Art The problem of the mobile communication business is how to accommodate a large number of users with high quality on a limited frequency. As a means for solving such a problem, there is a multi-input multi-output (MIMO) system. In this system, the 1st to Nth transmitters use the same frequency band, transmit the 1st to Nth symbol sequence signals respectively at the same time, and transmit these transmission signals to a plurality of antennas # 1 to #M. With a MIMO receiver equipped, receive and MI
The MO receiver processes the received signal, estimates the 1st to Nth transmission symbols from the 1st to Nth transmitters, and outputs the output terminals Out1 to Out.
Output to N for each transmitter. The configuration of one transmitter in the MIMO system is shown in FIG. The channel encoder 51 performs error correction coding on the information bits, the interleaved means 52 interleaves the coded bit sequence obtained by the error correction coding, and then the mapping means 53 maps the symbols into a symbol for transmission. . The modulation means 53
, The training side is known before the symbol sequence, and is unique to each transmitter.
The above training symbol sequence is added by the multiplexing means 55 and transmitted. This transmission signal sequence is converted into a high frequency signal and radiated as a radio wave from the antenna.

A MIMO receiver is disclosed in Non-Patent Document 1, for example. The configuration of a conventional MIMO receiver is shown in FIG. Received signals from the antennas # 1 to #M are converted into baseband signals, equalized by the MIMO adaptive equalization unit 35 as received signals r _{1 to} r _M , and transmitted by the MIMO adaptive equalization unit 35 for each transmission sequence. The outputs are decoded by the decoding units 62 _{1 to} 62 _N , respectively. The configuration of the MIMO adaptive equalization unit 35 is shown in FIG. The adaptive equalizer 35 estimates baseband sampled reception signals r _{1 to} r _M , a priori information sequence 2λ 2 _{1 to} λ 2 _N fed back from the SISO decoder 63, and channel state (propagation path characteristic) estimation. The value H is input, the replica generation unit 65 reproduces a replica of the received signal based on the a priori information sequence 2 and the channel estimation value, and the subtraction unit 66 subtracts this from the received signal to obtain ISI (intersymbol interference), MAI (channel interference). Interference). Furthermore, the ISI and MAI remaining after this reduction processing are processed by the MMSE filter 67, and the log likelihood ratio of each symbol is calculated by the LLR generation unit 68 for the processing result, and the prior information series ₁ λ _{l l} ~ λ is calculated.
Output as 1 _N. In the decoding units 62 _{1 to} 62 _N corresponding to each user (transmitter) in FIG. 2, the a priori information sequence 1 of each symbol calculated in the MIMO adaptive equalization unit 35 is processed by a demapping unit not shown in the figure. The bit sequence is converted to a bit sequence, the deinterleaver 69 deinterleaves the bit sequence, and the deinterleaved sequence is converted into a SISO decoder 63.
Decrypt with. The soft-decision output value sequence from the SISO decoder 63 is interleaved by the interleaver 71, and the output of the interleaver 71 is further mapped by a mapping unit (not shown) to form a MIMO as a soft-decision symbol sequence.
It is supplied to the adaptive equalization unit 35 as a priori information sequence 2λ2 _{1 to} λ2 _N. This MIMO adaptive equalization unit 35 and SISO decoding unit 62
The processes in _{1 to} 62 _N are repeated a plurality of times, and the final decoding results are output as the determination bit sequences Bseq1 to BseqN. When the transmission side performs interleaving on the symbol sequence, the signal sequence decoding unit also deinterleaves and interleaves the prior information sequences 1 and 2 of the symbol sequence.

Good equalization in the MIMO adaptive equalization unit 35,
That is, when intersymbol interference and interchannel interference are sufficiently removed, it is important to detect the sampling timing of the received signal, detect the symbol synchronization timing, determine the equalization range, that is, detect the synchronization timing including frame synchronization.
However, this point is not shown in the above-mentioned European Patent Application Publication. FIG. 4 shows a conventional adaptive equalizer when one user (transmitter) monopolizes one channel (transmission path) for transmission. The adaptive equalization unit 6 for generating the equalization start timing signal, the equalization range signal, and the channel state estimation value.
The synchronization channel generation unit 81 is provided in front of 1. Although not shown in the figure, the transmitting side first transmits, for each frame, a long training symbol sequence (synchronization word signal) whose transmitting symbol pattern is known on the receiving side, and then the information content to be transmitted. Is transmitted. Although not shown in the figure on the receiving side, the received radio wave from the transmitting side is amplified and demodulated from the signal received from the antenna, and is converted into a baseband received signal and input to the input terminal 11. The correlation between this received signal and the training symbol sequence TSS from the training sequence generation unit 13 is taken by the sampling signal generation unit 19, and the timing at which the correlation value becomes maximum, that is, the timing of the received signal on the path where the received signal power is strong. Is the sampling timing, and the sampling signal generated by the sampling signal generator 19 at this sampling timing is used to sample the reception signal in the sampling unit 20 to obtain a digital series reception signal. The correlation between the received signal of the digital sequence and the training symbol sequence TSS from the training sequence generation unit 13 is the synchronization timing generation unit 12
Required by. The correlation signal in the synchronization timing generator 12 changes, for example, as shown in FIG. 5A, and the timing ts at which the correlation signal Sigc becomes maximum is detected, and this is set as the symbol synchronization timing ts. A symbol synchronization timing signal generated at this timing ts, a digitized reception signal, a training symbol sequence TSS
Is input to the start timing / range candidate generation unit 21, an equalization start timing signal and an equalization range signal are generated from these, and these signals and the sampling reception signal are input to the channel estimation unit 28 and the channel estimation unit 28. The transmission path characteristics (channel state) are estimated at, and the estimated channel state, sampling reception signal, equalization start timing signal,
The equalization range signal is input to the adaptive equalization unit 61, and the sampling reception signal is adaptively equalized by the adaptive equalization unit 61,
The decision symbol sequence is output.

As shown in FIG. 5B, the equalization start timing ts and the equalization range TE are the training symbols in the paths (received signal samples) between the Ns symbols before and after the detected symbol synchronization timing ts. When the correlation output Sigc with the sequence exceeds a certain threshold value TH1, the path (sample) is set as the valid path, the leading valid path (sample) position is set as the equalization start timing t0, and the leading valid path (sample) position is set. The equalization range signal TE is output up to the last effective path (sample) position te. Here, as a method for determining whether each path is an effective path, for example, among paths (samples), the maximum correlation power value MAX is multiplied by 1 / Cp (Cp is a preset constant) as a threshold value. It is possible to use a standard in which TH1 is set and a sample of TH1 or more is set as an effective path.

[0006]

[Non-Patent Document 1] European Patent Application Publication EP1233
Publication No. 5655A2 (issued on August 21, 2002)

[0007]

The method of determining the symbol synchronization timing, the equalization start timing, and the equalization range in the MIMO receiver have not been sufficiently studied so far. The purpose of the present invention is a signal sequence (user)
MIMO that efficiently detects at least the equalization start timing and enables good adaptive equalization even when there are many
It is to provide a receiver and a MIMO receiving method.

[0008]

According to one aspect of the present invention, the symbol synchronization timing is detected and the channel is estimated, and the transmission signal from each transmitter is detected from the detected symbol synchronization timing and the estimated channel state. Generate a replica of the training symbol in the received signal,
An interference suppression signal is generated by subtracting a replica of received training symbols from at least one transmitter other than the target transmitter from the received signal, and the interference suppression signal is processed to receive the signal from the target transmitter. The equalization start timing used for the equalization processing of (1) and the equalization range are detected as necessary. With this configuration, even if a plurality of transmission signal sequences are mixed in the same frequency band at the same time, it is possible to accurately detect the equalization start timing and estimate the channel state.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION A MIMO receiver including synchronization timing detection and channel state estimation, that is, a synchronization / channel generation unit 81, proposed in Japanese Patent Application No. 2002-32639 to facilitate understanding of the present invention, will be described first. To do. FIG. 6 shows an example of the configuration of the proposed MIMO receiver. For the received signals (digital value series in which the baseband signal is sampled) received from the input terminals 11 _{1 to} 11 _M received by a plurality of antennas, this received signal for each antenna and the training symbol series generated by the receiver And are used to detect the timing signal indicating the beginning of the frame and the symbol synchronization timing signal synchronized with each symbol in the synchronization timing generation units 12 _{1 to} 12 _M. Hereinafter, both the frame start timing and the symbol timing are simply referred to as synchronization timing.

FIG. 7 shows an antenna A _m (m = 1, ...,
Synchronization timing generating unit for the received signal r _m of M) 12 _m
A configuration example of is shown. The generation of the synchronization timing signal will be described by taking the transmission signal of the transmitter TX1 of the user 1 (hereinafter simply referred to as the signal of the user 1) as an example, the reception signal r _m from the antenna _Am through the input terminal 11 _m , and the training. from sequence generating unit 13 _1, the transmitter TX1 obtained in the correlation unit 14 ₁ to the correlation between the training symbol sequence and the same training symbol sequence TSS ₁ for transmitting the shift register 15 of the energy index generation unit 37 ₁ the correlation output type _1, the weight W ₁ to the output of each shift stages of the shift register 15 _1, ..., by multiplying the W _e multiplication unit 16 _1, the addition section 1
7 ₁ , that is, the weighted addition is performed on the correlation output over a certain period of time, and the received energy index is obtained. Performs the maximum value detection process at the maximum value detecting unit 18 ₁ for the reception energy index, the timing of the reception energy index is maximum, the synchronization timing signal for the user 1 signal in the received signal from the antenna A _m ts1- Output m.

Next, using the synchronization timing and the received signal, the start timing / range candidate generation unit 21 _{1 in} FIG.
A determination of all of the equalization start timing t0 ₁ ~t0 _N equalization range candidate te ₁ ~te _N input signal sequences (the received transmitter transmission signal sequence) in through 21 _M. FIG. 8 shows a configuration example of the start timing / range candidate generation unit 21 _m . Here, taking the signal of the user 1 as an example, the correlation between the received signal r _m from the antenna A _m (input terminal 11 _m ) and the training symbol sequence TSS ₁ from the training sequence generation unit 13 _{1 of the} user 1 is determined by the correlation unit. 14 ₁ and the correlation output is compared with the threshold Tha in the comparison unit 22 ₁ . Comparator 22 _1, the "1" greater than the threshold Tha correlation output, when the correlation output is less than the threshold value Tha outputs "0", the output of the comparator 22 ₁ It is input to the shift register 23 ₁ . Shift register 23
₁ indicates the symbol synchronization timing signal ts by the shift control unit 24 _{1 which} receives the symbol synchronization timing signal.
It is controlled so that the output of the comparing unit 22 ₁ at the time of Ns symbols before and after the timing of 1-m is input to the shift register 23 ₁ . In the start timing equalization range candidate detection unit 25 ₁ , the ts1-m ± N in the shift register 23 ₁
the timing of the in earliest timing in s symbol "1", and outputs the equalization start timing signal t01-m for the user 1 signal in the received signal of the antenna A _m,
A signal representing the length from t01-m to the timing "1" at the latest timing is used as an equalization range candidate signal te.
Output as 1-m. All equalization range candidate signals obtained for each antenna reception signal for the transmission signals of all transmitters are input to the equalization range signal generation means 27 (FIG. 6),
The largest of these N × M equalization range candidate signals is output as the equalization range signal TE.

In the channel estimation unit 28 in FIG. 6, the equalization range signal, the equalization start timing signal and the received signal are input, and the channel state (propagation path characteristic, impulse response) H between each transmitter and each antenna is calculated. Presumed. FIG. 9 shows a configuration example of the channel estimation unit 28. In the channel estimation unit 28, a training symbol sequence TSS ₁ for all transmitters from the training sequence generator 13, ..., with respect to TSS _N, between the transmitter transmits signals using an equalization start timing signal t0 ₁ ~t0 _N The offset compensator 29 compensates for the deviation of the reception timing. A convolutional integrator 31 performs a convolutional integration of the channel state H from the adaptive algorithm unit 33 with respect to the training symbol sequence in which the reception timing deviation is compensated to create a replica Re of the received signal, and the replica of the received signal is used as the received signal. Is subtracted from the subtraction unit 32 to obtain an error signal. In the adaptive algorithm unit 33, the error signal
The channel state estimation value is sequentially updated using the equalization range signal and the previous channel estimation value in the iterative process. As this adaptive algorithm, for example, the RLS algorithm can be used. Returning to the description of FIG. 6, the equalization start timing signals from the start timing / range candidate generation units 21 _{1 to} 21 _M , the equalization range signal from the equalization range generation unit 27, and the estimation from the channel estimation unit 28 are estimated. The channel state and the received signal are input to the MIMO adaptive equalization unit 35, the received signal is subjected to MIMO adaptive equalization processing, and the estimated transmission symbol of each transmitter is output.

The functional configuration of this MIMO adaptive equalization unit 35 is shown in FIG. 3, for example. Although not described in the above with respect to FIG. 3, the symbol offset compensating unit 64 compensates for the deviation between the user received signals by the equalization start timing signal for the a priori information sequence 2 as shown by the broken line in FIG. And supplies the equalization range signal TE to the replica generation unit 65 and the MMSE filter 67. Each user received signal equalized and separated by the adaptive equalization unit 35 is decoded by the signal sequence decoding unit 62 having the functional configuration shown in FIG. 2, for example. Further, the symbol synchronization detection method and the equalization range determination method proposed in the Japanese patent application will be described. FIG. 10 shows another configuration example of the MIMO receiver. In this configuration example, the synchronization timing between the received signal of the antenna A ₁ to A _M, equalization start timing, in a case where the equalization range candidates can be regarded as the same, they were common among the antennas of the received signal, all Obtained using the signal received by the antenna. FIG. 11 shows the configuration of the synchronization timing generator 12 for the user 1 signal.
First, the correlation outputs of the signals r _{1 to} r _M received by the respective antennas from the input terminals 11 _{1 to} 11 _M and the training symbol sequence TSS ₁ from the training sequence generation unit 13 _{1 of the} user 1 are correlated units 14 ₁ to. 14 respectively determined by _M, for these correlation output to produce an energy index performs weighting respectively added over a period of time the energy index generating unit 37 ₁ to 37 _M. Next, the sum of the energy indexes obtained for each antenna reception signal at the sampling time point of each received signal is obtained by the addition section 38, and the maximum value detection section 18 detects the timing at which the sum of the energy indexes becomes maximum, At the timing, the synchronization timing signal ts1 for the signal of the user 1 is output.

Next, the start timing / range candidate generation unit 21 (FIG. 10) starts equalization of all input signal sequences by using the synchronization timing and the received signal for each user signal thus obtained. The timing and the equalization range candidates are determined. FIG. 12 shows the configuration of the start timing / range candidate generation unit 21 for the user 1 signal. Input terminal 1
Respectively determined the correlation between the user 1 of the training symbol sequence from the received signal r ₁ ~r _M and training sequence generator 13 ₁ of each antenna from 1 ₁ to 11 _M in the correlation unit 14 ₁ to 14 _M, each sample time point At, the adder 39 obtains the sum of the correlation outputs of all the antenna reception signals. The comparison unit 22 compares the total sum of the correlation outputs with the threshold value Tha. The comparison unit 22 outputs “1” when the correlation output is larger than the threshold value Tha, and when the correlation output is smaller than the threshold value Tha. "0" is output, and the output of the comparison unit 22 is input to the shift register 23. In the shift register 23, the output of the comparison unit 22 of the times of the Ns symbols before and after the synchronization timing ts1 is output by the shift control unit 24 that receives the synchronization timing signal.
Controlled to be entered in. The start timing signal / equalization range candidate detection unit 25 sets the equalization start timing signal t0 for the signal of the user 1 at the timing "1" which is the earliest timing in the ts1 ± Ns symbol period.
A signal representing the range from t01 to the timing "1" at the latest timing is output as the equalization range candidate signal te1.

The equalization range candidate signals for the received signals of all the antennas for each transmitter are the range generator 27 (see FIG. 1).
0), and the largest one of all the input equalization range candidate signals is output as the equalization range signal.
After that, as in the case where the equalization start timing signal and the equalization range signal are obtained for each reception signal of the antenna, the channel estimation unit 28 uses the equalization range signal from the range generation unit 27 and the start timing / range candidate generation unit. The channel state between each transmitter and each antenna is estimated by using the equalization start timing signal and the received signal from 21 and the channel state estimated by these equalization start timing signal and the equalization range signal is used. MI
The MO adaptive equalization unit 35 performs MIMO adaptive equalization processing on the received signal, estimates the transmission symbols of each transmitter, outputs them separately, and further decodes these outputs by the signal sequence decoding unit 62. Even with the proposed synchronization timing detection method and equalization range determination method described above, there is a possibility that the synchronization timing and the equalization range may be wrong. If the equalization range becomes unnecessarily long, it causes useless increase of the calculation amount, and if the equalization range becomes too short, it causes the equalization capability to decrease. In particular, when the number N of transmission signal sequences increases, it is considered that deterioration of synchronization establishment accuracy, equalization range determination accuracy, and channel estimation accuracy due to interference between input signal sequences increases. For this reason,
In the MIMO system, the number of training symbols required to obtain a constant signal characteristic must be increased as compared with the case where only one transmission signal sequence is received. Examples of the present invention will be described below.

First Embodiment On the transmitting side, each of the N transmitters TX1 to TXN transmits signals at the same frequency and the same time. In addition, each transmitter transmits a training symbol sequence unique to each transmitter whose transmission symbol pattern is known on the receiving side. FIG. 13 shows a first embodiment of the present invention. Here, it is assumed that signals that are demodulated as reception signals to be baseband signals and that are digital series are input to the input terminals 11 _{1 to} 11 _M. For the reception signals received by the plurality of antennas, the synchronization / channel generation unit 81 uses the antenna A
_The iterative processing unit 41 _m provided for each _m (m = 1, ..., M) performs generation of an equalization start timing signal, generation of an equalization range signal, and channel estimation. Each iterative processing unit 41 _m is composed of a plurality of processing stages 42 _s (s = 1, ..., S) connected in cascade. In each processing stage 42 _s , the received signal replicas generated in the immediately preceding processing stage 42 _s-1 and the error signal E _s obtained by subtracting these received signal replicas from the received signal are input, and received signals from all transmitters are input. Equalization start timing signal, equalization range signal,
The channel state and the reception replica are generated, the equalization range signal is output, and the generated reception signal replicas and the error obtained by subtracting these reception signal replicas from the reception signal are supplied to the next processing stage 42 _{s + 1} . Output the signal E _s . Also,
Since it is desirable that the equalization range be the same for the reception signals of all the antennas, the equalization range signal generation unit 43 common to the iterative processing units 41 _{1 to} 41 _M for each processing stage 42 _s.
s, and each processing stage 4 of each iterative processing unit 41 _{1 to} 41 _M
The equalization range candidate signals from the 2 _s from the equalized range signal generating unit is input to the 43 _s the equalization range signal generator 43 _s repeated to determine the equalization range signal the longest in these candidate processing unit 4
It returns to each processing stage 42 _s of 1 _{1 to} 41 _M.

FIG. 14 shows an example of the functional configuration of the sth processing stage 42 _s in the iterative processing unit 41 _m for the received signal of the antenna A _m . Here, the signal of the user 1 (transmitter TX1
The received signal will be described as an example. First, the user generated in the pre-processing stage 42 _s-1 is added to the error signal obtained by subtracting the received signal replicas of all the user received signals generated in the s-1 th processing stage (pre-processing stage) 42 _s-1 from the received signal. 1
Signal reception signal replica (hereinafter referred to as reception signal replica of user 1; the same is true for other signals) Re ₁
Is added by the adder unit 44 ₁ . This addition output signal is a reception signal replica Re _{2 to} Re _N other than the user 1 from the reception signal.
Then, the interference suppression (user 1) reception signal r _IC1 is obtained by subtracting from the interference suppression reception signal r _IC1 , and the synchronization timing generation unit 12 ₁ generates a synchronization timing signal using this interference suppression reception signal r _IC1 . Furthermore, the generated synchronization timing signal and the addition unit 4
The interference suppression received signal from 4 ₁ is input, and the start timing / range candidate generation unit 21 ₁ generates the equalization start timing signal t 0 ₁ and the equalization range candidate signal te ₁ . The synchronization timing generation unit 12 ₁ and the start timing / range candidate generation unit 21 ₁ respectively detect the synchronization timing signal ts1-m in FIG. 7 described above and the equalization start timing signal in FIG. 8, respectively. The configuration can be the same as that shown in the part for detecting t01-m and the equalization range candidate signal te1-m. The equalization range candidate signal generated as described above is the interference suppression reception signal r of another user (2 to N).
Start timing / range candidate generator 2 for _IC2- r _ICN
It is input to the equalization range generation unit 43 _s together with the equalization range candidate signals obtained from 1 _{2 to} 21 _N and the equalization range candidate signals obtained in the s-th processing stage 42 _s for other antenna reception signals in the same manner. The equalization range generation unit 43 _s outputs, as an equalization range signal, the one having the longest equalization range candidate among all the input equalization range candidate signals. The equalization range signal is input to the channel estimation unit 28 together with the equalization start timing and the received signal, and the channel state is estimated. Channel estimation unit 2
The configuration of 8 can also be the same as that shown in FIG.

The equalization start timing signal t 0 ₁ , the equalization range signal te ₁ , the estimated channel state H and the training symbol sequence TSS ₁ of the user ₁ are input to the replica generator 45 ₁ and the replica generator 45 ₁ uses the user 1. A received signal replica Re ₁ (from transmitter TX1) is created. FIG. 15 shows a configuration example of the replica generator 45. The channel state estimation value from the channel estimation unit 28 and the training symbol sequence TS from the training sequence generation unit 13
S _n (n = 1, ..., N) is convolved and integrated by a convolution integrator 31, and an equalization start timing signal t0n-m is used for this output to shift the synchronization timing between user received signals. The offset compensator 29 compensates for the replica of the received signal of the user n. In this way, the replicas of the received signals of all the users 1 to N (from the transmitters TX1 to TXN) generated by the replica generators 45 _{1 to} 45 _N in the error signal generation unit 46 are transmitted to the antenna A.
_An error signal is generated by subtracting from the received signal of _m .
Channel state values generated as described above, replicas Re _{1 to} R for the training symbols in each user signal
e _N, and the error signal Es obtained by subtracting the replicas Re ₁ to Re _N from the received signal is supplied to the next processing stage 42 _{s + 1.}
By repeating the above process also in the processing stage 42 _{s + 1} to update the replica of the training symbol in the received signal, the equalization start timing signal, the equalization range detection accuracy, and the channel state estimation value estimation accuracy. Improve. The equalization start timing signal of each user received signal generated in the final processing stage 42 _S , the channel state estimation value, the equalization range signal, and the received signal are the MIMO adaptive equalization unit 35 (FIG. 13).
, The transmission symbols of each user (transmitter) are estimated and output. The first processing stage (first stage) 42 ₁ is shown in FIG.
The adders 44 _{1 to} 44 _{N in the} middle are omitted, and the reception signal from the input terminal 11 _m is represented by the broken line, and the iterative processor 41 _{m is used.}
Are supplied to the synchronization timing generators 12 _{1 to} 12 _N and the start timing / range candidate generators 21 _{1 to} 21 _N , respectively, and the S-th processing stage 42 (final processing stage) receives the equalization start timing signal and the channel state estimated value H. , Output only the equalization range signal. At least two processing stages 42 are provided.

Second Embodiment In the first embodiment, the received signal from the input terminal 11 _m is used for the channel state estimation, and the received signal for all users is collectively evaluated. In the second embodiment of the invention, the second processing stage 42 ₂
In the subsequent channel estimation section, the channel state is estimated for each user received signal using the interference suppression received signal r _IC obtained by subtracting the received signal replica other than the desired user received signal from the received signal. By estimating the channel state for each user received signal in this way, the number of parameters to be estimated can be reduced to 1 / the number of user signals, so that the convergence speed of the channel state estimation value can be increased, and Channel state estimation is possible with a small number of training symbols. FIG. 16 shows the second processing stage 42 ₂ in the second embodiment.
The functional configuration of the subsequent processing stages is shown by giving the same reference numerals to the parts corresponding to those in FIG. The point different from the first embodiment is that the channel estimation unit 28 ₁ -2 for each user signal
Only in that provided as 8 _N, other configurations 14
Is the same as that shown in. FIG. 17 shows the configuration of the channel estimation unit 28 _n for each user signal. The difference from the channel estimation unit that estimates the channel states of all user signals shown in FIG. 9 is that the input equalization start timing signal, the generated training symbol sequence, the channel state estimation value, and the received signal replica are the desired users n. It is only for the signal of. Other configurations and operations are the same as the case of performing channel state estimation of received signals of all users.

Third Embodiment In the third embodiment of the present invention, in the channel state estimation of the second embodiment, the channel state estimation is performed using the equalization range candidate signal obtained for each user signal. FIG. 18 shows the configuration in this case. The configuration differs from the second embodiment is omitted equalization range generation unit 43 _s for each processing stage 42 _s, configuration of the s processing stage 42 _s, as shown by the broken line in FIG. 16, for each user Start timing / range candidate generation unit 21 _{1 to} 21 _N
The equalization range candidate signals from the respective
Input to _{1 to} 28 _N. Other configurations are the same as those in the second embodiment. Only the S-th processing stage (final processing stage) 42 _{S of} each equalization range candidate signal is supplied to the equalization range generation unit 43 in FIG.

Fourth Embodiment Regarding the same user received signal in the first embodiment, when it can be considered that the symbol synchronization timing signal, the equalization start timing signal, and the equalization range candidate signal are substantially the same between the antennas. It is possible to make these signals common to the same user received signal between the antennas, and to process the signals received by all the antennas by one iteration processing unit 41. This is the fourth aspect of the present invention.
This is an example, and FIG. 19 shows a configuration example thereof. The difference from the configuration of the first embodiment is that the iterative processing unit 41 is not provided for each received signal of the antenna, but one iterative processing unit 41 collectively processes the received signals from all the antennas. is there. FIG. 20 shows the sth processing stage 42 _s in this case.
A configuration example of is shown. The entire configuration is the s-th in the first embodiment.
It is similar to the processing stage 42 _s, in the first embodiment is only the received signal of one antenna A _m has been input, the received signal from the antenna A ₁ to A _M in this fourth embodiment, i.e., the input All the received signals from the terminals 11 _{1 to} 11 _M are input. The synchronization timing generators 121 to ₁ used here
2 _N , start timing / range candidate generation unit 21 _{1 to} 21
Regarding _N with respect to the received signal of the user 1, _N can have the same configuration as that shown in FIGS. 11 and 12, respectively. Also, the replica generator
The processing shown in FIG. 15 in the first embodiment may be performed for all antennas for each antenna. The batch processing of the received signals from all the antennas can be applied to the case where the channel estimation unit 28 is provided for each user as in the second embodiment shown in FIG.

Fifth Embodiment In the above embodiments, the first processing stage 42 ₁ of the iterative processing unit 41 performs synchronization detection without subtracting the received signal replica from the received signal. In this case, the synchronization detection processing including the frame synchronization in the synchronization detection unit 82 _n of the user signal having a weak reception power due to the influence of the user signal having a high reception power malfunctions and the synchronization timing largely shifts. This may lead to deterioration in the equalization start timing of the user signal and the equalization range detection accuracy. Especially when the difference in received power between user (transmitter) signals is large, the accuracy of equalization start timing, equalization range detection, and channel estimation obtained by the user (transmitter) signal with higher received power is higher. It is expected that the equalization start timing and the equalization range can be obtained more accurately by performing the equalization start timing, the equalization range detection, and the channel estimation in order from the user signal with the largest received power. The fifth embodiment of the present invention is based on this idea. The overall configuration of the fifth embodiment is almost the same as that shown in FIG. 18, but in each processing stage 42, the processing for each user is performed in series instead of in parallel. In order to add the order of the serial processing, FIG.
As indicated by a broken line therein, a processing order detection unit 48 is provided in at least the first processing stage 42 ₁ and performs synchronization detection, that is, symbol synchronization timing signal, equalization start timing signal, equalization range signal generation and channel estimation. Previously, for example, using the functional configuration shown in FIG. 7, the maximum value of the received energy index of each user signal is detected, and the user signals are ranked in descending order of the maximum value of the received energy index. For example, in FIG.
As shown in A, the correlation / index detection units 48a ₁ , ..., 48a _{N including} the correlation unit 14 and the energy index generation unit 37 in FIG. 7 or FIG. 11 and a unit that detects the maximum value of the generated energy index. , Each user 1 training symbol sequence TSS ₁ , ..., User N training symbol sequence TSS _N and the received signal are respectively processed, and the maximum value of the energy index of each user received signal is detected. The order determining unit 48b determines the order of processing for user signals in order of magnitude.

Here, the maximum value of the received energy index is
The largest order is user 1, user 2, ... user N
It FIG. 22 shows the processing stage 42._sShows the configuration of. Processing stage 42_sso
Is the descending order of the maximum received energy index, in this case
Is the synchronization timing signal from the user 1 signal and the equalization start
Imming signal t0₁And equalization range signal te₁And the generation of
Synchronization detection and channel status H₁Is estimated. Beginning
Therefore, the s−1th processing stage 42_s-1Of all users generated by
Received signal after subtracting received signal replica, that is, error signal
No. Es pretreatment stage 42_s-1User 1 received message generated by
No. Replica Re₁Adder 44₁Add with. This added
The received signal is the received signal response other than user 1 from the received signal.
Rica Re₂~ Re_NInterference suppression user 1 received signal minus
And this interference suppression user 1 received signal r_IC1Using
The synchronization timing generator 12₁With sync timing signal
To generate. In addition, this symbol sync timing signal
And interference suppression user 1 received signal r_IC1Enter and start
Imming / range candidate generator 21₁Start of equalization in Thailand
A minging signal and an equalization range candidate signal are generated. Synchronous timing
Generation unit 12₁And start timing / range candidate generation unit 2
1₁Shows, for example, the configuration for the signal of the user 1.
7 so that it has the same configuration as shown in FIG. 8 respectively.
can do. These equalization start timing signals, etc.
Range candidate signal, addition unit 44₁Interference suppression user 1
The channel estimation unit 28 receives the received signal.₁Type in channel form
Estimated state value H₁And these channel state estimation values H₁,etc
Start timing signal t0₁And User 1 training
Symbol series TSS₁And using the replica generator 45₁so
Received signal replica Re of user 1₁To generate. This first
s processing stage 42_sUser 1 received signal replication generated in
The addition unit 44₁Interference suppression from user 1 received signal
Subtraction unit 49₁Subtract with. This signal is
s processing stage 42_sUser 1 received signal replication generated in
Mosquitoes and s-1 processing stage 42_s-1Other than user 1 generated by
The error signal is obtained by subtracting the received signal replica. This mistake
The maximum value of the received energy index is the next largest for the difference signal.
S-1 processing stage 42 for the received signal_s-1Received from
Signal replica, in this example User 2 received signal replica
Adder 44₂And the interference suppression user 2 received signal is added
Is generated, and this is the synchronization timing generation unit 12₂, Start ta
Imming / range candidate generator 21₂, Channel estimation unit 28₂
To the replica generator 45.
₂Will generate a replica of the user 2 received signal, which will be
Subtraction unit 2 from the reception signal of the suppression user 2 ₂Deducted at
An error signal is generated as a result, and this error signal is added to the addition unit 44.₃What
Supplied. For signal processing for users 3 and later
The same is done.

In the first processing stage 42 ₁ , the process for the received signal of the user 1 having the largest received energy index maximum value, that is, the synchronization timing generation unit 12 ₁ , the start timing / range candidate generation unit 21 ₁ , the channel estimation Part 2
8 ₁ , the reception signal is directly supplied to the subtraction unit 49 ₁ as shown by a broken line, the channel state of the user 1 signal is estimated, the user 1 replica is generated, and the reception signal of the user 2 having the next largest received energy index maximum value is received. On the other hand, the reception signal replica of the user 1 generated by the replica generator 45 ₁ is subtracted from the reception signal by the subtraction unit 49 ₁ , and the synchronization timing generation unit 12 ₂ , start timing / range candidate generation unit 21 ₂ , channel It is supplied to the estimation unit 28 ₂ and the subtraction unit 49 ₂ . The same processing is performed thereafter. Further, in the processing order detection unit 48 (FIG. 18) in the first processing stage 42 ₁ , the reception signal from the terminal 11m is input to the switching unit 48c as shown in FIG. 21A, and the synchronization timing generation unit 12 ₁ , ..., It is output to the synchronization timing generation unit 12 ₁ for the signal processing of the first user in the processing order determined by the order determination unit 48b among the 12 _N , in this example, the signal processing of the user 1. Further, each output of the subtraction units 49 ₁ , ..., 49 _N is input to the serial connection unit 48 d, and the processing for each user signal is sequentially performed in the processing order determined by the order determination unit 48 b, so that the synchronization timing generation unit 12 _1, ..., is supplied to the 12 _N, the subtraction unit 49 ₁ in this example, ..., the outputs of 49 _N-1 is, the synchronization timing generator 12 _2, ..., will be supplied respectively to 12 _N.

After the second processing stage 42 _{2 as} well, the first processing stage 42 ₁
The processing order may be determined by the second processing stage 42 ₂
After that, the processing order detection unit 48 may be provided to determine the processing order again. In this case, not the received signal but the output signal of the addition unit 44 _n is input to each correlation / index detection unit 48a _n (n = 1, ..., N) in FIG. 21A. In addition, the processing order detection unit 4 after the second processing stage 42 ₂
In FIG. 21, as shown in FIG. 21B, each user n channel state estimation value H _n generated in the preprocessing stage 42 _s-1 is supplied to the received energy calculation unit 48 e _n, and each channel state estimation value (impulse response) is supplied. It is also possible to obtain the received energy by obtaining the sum of squares of the respective coefficients, and the order determining unit 48b may determine the processing order for the user signals in descending order of the received energy. Alternatively, in FIG.
Index detection unit 48a _1, ..., not the maximum value of the energy index by 48a _N, correlation unit 1 in the correlation output, i.e. 7
4 _1, ..., may be supplied to the order determination section 48b detects the respective maximum values of the outputs of the 14 _N. Performing the processing for each user signal in series in this way can also be applied to the case of collectively processing received signals of all antennas shown in FIGS. 19 and 20. In that case, in FIG. 20, the channel estimation unit 28 is provided for each user, and the processing order detection unit 48 is provided in at least the first processing stage 42 ₁ in FIG. 19 as shown by the broken line.

Sixth Embodiment In the fifth embodiment, the iterative processing means 41 _{m has} a configuration in which the synchronization timing, the equalization start timing, the equalization range detection accuracy, and the channel state are particularly determined when the difference in received power between users is large. As a method expected to improve the estimation accuracy, a method of serially subtracting received signal replicas of sequentially generated training symbol sequences from the received signal was mentioned. When this method is used, the interference-suppressed reception signal (reception signal from which the replica is subtracted) that is detected in the first processing stage 42 _{1 at} an early stage is a reception signal replica of another user generated at a later stage. Is processed without being subtracted, there arises a problem that errors easily occur in the synchronization timing, the equalization start timing, and the equalization range. In particular, when the difference between the first and second largest received powers is small and a relatively large error occurs in the equalization start timing and the equalization range detected for the first user signal, the received signal replica of that user is generated. However, the received signal is significantly different from the received signal of the user, and this greatly affects the processing for the received signals of other users. Sixth embodiment of the present invention is to solve this problem, using the synchronous detection unit and the channel estimation unit of the first processing stage 42 ₁ only parallel as shown in FIG. 23, in the second processing stage 42 ₂ and later The influence is reduced by using the serial type synchronization detection unit and the channel estimation unit. The configuration shown in FIG. 19 can be used as the configuration of the parallel type processing means, and the configuration shown in FIG. 21 can be used as the configuration of the serial type processing stage. Also in the sixth embodiment, it is possible to collectively process the reception signals of all the antennas as in the fourth embodiment shown in FIGS.

Seventh Embodiment In the fifth and sixth embodiments, the iterative processing means 41 _m is configured so that the equalization start timing and the equalization range detection accuracy and the channel state are particularly high when the difference in received power between users is large. From the point that the estimation accuracy is expected to be improved, the method of subtracting the received signal replica of the training symbol sequence from the received signal is adopted in the serial configuration. However, when this method is used, processing for each user cannot be performed in parallel, so
There is a problem that the processing delay becomes large. In the seventh embodiment of the present invention, as a method for solving this problem, a composite type structure having both the advantage of the parallel type and the advantage of the series type is adopted. The entire structure of the seventh embodiment can be the same as that shown in FIG. Before performing the equalization start timing signal, the equalization range signal generation, and the channel estimation, the processing order detection unit 48 uses, for example, the energy index generation unit 37 illustrated in FIG. The maximum value is detected, and N _G1 , N _G2 , N _G3 , ..., N _GX (N _G1 , N _G2 , N
_G3 , ..., N _GX are natural numbers).
Here, each group is G1, G2, G3, ..., GX. This group is divided into, for example, as shown in FIG. 24, the maximum values I _{Emax1 to} I of the received energy index of the user n from the correlation / index detection units 48a _{1 to} 48a _N in FIG. 21A.
The maximum value I _E Max in _EmaxN detected by the maximum value detecting unit 48f, a group selecting section 48 g, when the number of groups and _{X, I E Max × (X} -1) / X or more of the maximum value I _Emaxn
To belong to the user group G1, with the exception of those users, to belong to user _{I E Max × (X-2} ) / X more I _Emaxn group G2, performs grouping in the same manner. The processing stages 42 _{1 to} 42 _S are processed in parallel for each group, and the parallel processing is performed in the order of groups G1 to G2, G3, ..., GX. Figure 25
FIG. 11 shows the structure of the repeating processing means 41 _m including the composite type, that is, the parallel type and the serial type. In the s-th processing stage 42 _s , the s-th
1 and the user received signal replica belonging to the error signal and the group G1 from the processing stage 42 _s-1 is input to the parallel processing unit 51 _G1, each user receives signals of the group G1 to the error signal in the parallel processing unit 51 _G1 group G1 The replicas are respectively added to generate N _G1 interference suppression signals, and for each of them, that is, for each user reception signal belonging to group G1, each synchronization timing signal equalization start timing signal, equalization range candidate signal, and channel state estimation. Each value is detected, and each received signal replica is further generated. The processing for the N _G1 user signals belonging to the group G1 is thus performed in parallel. These N _G1 received signal replicas are
The signal (error signal) subtracted from the signal added with the generated N _G1 interference-suppressed reception signals is supplied to the parallel processing unit 51 _G2 of the next group G2. This parallel processing unit 51 _G2 user received signal replica belonging to the group G2 from the first s-1 processing stage 42 _s-1 is input, the parallel processor 51 _G2
The sync timing signal of a user belonging to the group G2 and an error signal by processing similar to the parallel processing unit 51 _G1 from the parallel processing unit 51 _G1 with these user received signal replica, the equalization start timing signal, the equalization range candidate signal , Channel state estimate, received signal replica detection and generation in parallel,
The parallel processing unit 51 _G2 subtracts the received signal replica belonging to _G2 from the signal obtained by adding the N _G2 interference suppression received signals generated by the parallel processing unit 51 _G2 to generate an error signal, and and outputs to the parallel processing unit 51 _G3 group G3. The same processing is performed thereafter.

The configuration of each parallel processing unit 51 _{G1 to} 51 _GX is shown in FIG.
4, the same as the first to third embodiments shown in FIGS. 16 and 18. In this case, the first processing stage 42 ₁ also has the configuration shown in FIG. 14, and the parallel processing unit 51 _G1 of the first group _G1 has the error signal and the preceding group G1.
The received signal is input to the received signal replica without being input to the parallel processing unit 51 _{G2 by} subtracting the generated user received signal replica belonging to G1 from the received signal. Supply as a signal. Thereafter, the parallel processing units 51 _{G2 to} 51 _Gx sequentially perform similar processing. The received signal input to the switching unit 48c is transferred to the synchronization timing generation units 12 ₁ , ..., 12 _N shown in FIG.
The processing order detection unit 48 shown in FIG. 4 supplies the determined group G ₁ to each user, and the output group of the subtraction units 49 ₁ , ..., 49 _N input to the serial / parallel connection unit 48h. The user for G1 is supplied to each user for the next group G2 determined by the processing order detector 48 in the synchronization timing generators 12 ₁ , ..., 12 _N , and each of the groups G2 is supplied. The output of the subtraction unit 49 for the user is supplied to the synchronization timing generation unit 12 for each user of the next group G3, and the processing for each group is sequentially performed in the same manner. In the second processing stage 42 ₂ after, correlation-index detection unit 48a ₁ if a composite process, ..., an adder 44 ₁ instead of the received signal 48a _N, ..., to supply the respective outputs of the 44 _N Good Alternatively, instead of the correlation / index detection unit, as shown in FIG.
1B, the received energy calculation unit 48e ₁ , ..., 48
e _N may be used. Alternatively, in FIG. 24, in the correlation / index detection units 48a ₁ , ..., 48a _N , the correlation output is changed to the maximum value of the energy index, that is, the correlation unit 1 in FIG.
4 _1, ..., a maximum value of the output of 14 _N may be supplied to the maximum value detecting unit 48f and the group selection section 48 g.

In the fifth to seventh embodiments, the received signals of the respective antennas A _{1 to} A _M are processed by the repetitive processing means 41 _{1 to} 41 _M , respectively, but like the fourth embodiment, the reception from all the antennas is performed. Even when signals are collectively processed by the single repetitive processing unit 41, a serial type configuration or a combined configuration of a serial type and a parallel type may be used. Furthermore, the _{first to} Sth processing stages 42 _{1 to} 42 _S
Does not need to be provided as a hardware configuration, respectively. For example, FIG. 14, FIG. 16, FIG. 20, FIG.
One of the processing stages 42 shown in FIG. 5 may be provided in the synchronization / channel generation unit 81 as shown in FIG. 26, and this may be used repeatedly. However, except for the one shown in FIG. 20, one processing stage 42 is properly provided for each iterative processing unit 41 _m . The repetitive control unit 91 controls the input switching unit 92 to first detect the antenna reception signal by the synchronization detection unit 82 ₁ of the processing stage 42.
To 82 _N (only the synchronization detection unit 82 _{1 in} FIG. 24, only the parallel processing unit 51 _{G1 in} FIG. 25) to process the received signal,
The processing result, that is, each of the received signal replicas Re _{1 to} Re _N is caused to act as the first processing stage 42 ₁ and the addition units 44 _{1 to} 44 _N are added.
To the adders 44 _{1 to} 44 _N (only the adder 44 _{1 in} FIG. 22, only the parallel processor 51 _{G1 in} FIG. 25) to supply the error signal Es. 2 processing stages 42 ₂ , and so on 1
One processing stage 42 is repeatedly used, and at the S-th time, the processing stage 42 acts as the S-th processing stage 42 _S , and the equalization start timing signal, the equalization range signal, and the channel state estimation value in the processing result are MIMO. It may be output to the adaptive equalization unit 35. In each of the above-described embodiments, the channel estimation value obtained in the previous processing stage 42 _s-1 may be input to and used by the corresponding channel estimation unit of the processing stage 42 _s . In this way, the channel state converges quickly.

In each of the above-mentioned embodiments, when the reliability of the detected timing is low, it is desirable not to output the replica generated using the timing. In this case, for example, as indicated by a broken line in FIG. 14, the energy index of the received training symbol sequence signal of the user n from the synchronization timing generation unit 12 _n (n = 1, ..., N) (FIG. 7).
Output of the energy index generating unit 37 _n ) or user n
The correlation signal with the training symbol sequence (output of the correlation unit 14 _{n in} FIG. 7) is compared with the threshold Tr from the threshold storage unit 76 _n in the reliability determination unit 75 _n , If there is, it is determined that there is no reliability, and the reliability determination unit 75 _n
Output of the switch 77 _n is turned off so that the replica from the replica generator 45 _n is not output. When it is determined that the timing detected by the synchronization timing generation unit 12 _n is not reliable, the equalization start timing for the signal of the user n, the equalization range is detected, channel estimation is performed, and replica generation is not performed. Should not be output. Whether or not the reliability of the detected timing is low is judged by the channel state (impulse response) H _n estimated by the channel estimation unit 28 _n in the pre-processing unit 42 _{s-1 as} indicated by the broken line in FIG. 2 for each coefficient
The sum of multiplications is calculated by the reception energy calculation unit 78 _n , and if this reception energy is less than or equal to the threshold value Tr, the reliability determination unit 7
The switch unit 77 _n may be turned off by the output of 5 _n . In any case, a fixed value may be used as the threshold value Tr, or a constant X (0 <X <1, relatively high reliability is added to the square sum of impulse response coefficients of the previous channel state estimation value. To 0.5 or more) or the maximum value in each energy index or correlation output of the entire received signal sequence is multiplied by a constant X (0 <X <1, preferably 0.5 or less in this case). It can be a value. Not outputting the replica corresponding to the above case where the reliability of the timing is low can be applied to other embodiments as well, and its functional configuration is indicated by a broken line in the corresponding figure, and the description thereof will be omitted.

Eighth Embodiment In the eighth embodiment of the present invention, when an error is detected in the decoding decision bit string, this detection is fed back to the synchronization detector of the synchronization / channel generator to obtain the synchronization timing determined in advance. The following first candidate of is used. FIG. 27 shows the functional configuration of the eighth embodiment.
Shown in. The synchronization / channel generation unit 81, for example, the synchronization timing in the final processing stage 42 _s in the synchronization detection unit 82 _n (n = 1, ..., N) in any one of FIG. 13, FIG. 18, and FIG. In the generation unit 12 _n , the energy index (corresponding to the output of the energy index generation unit 37 _n in FIG. 7) or the correlation output (correlation unit 14 _{n in} FIG. 7)
(Corresponding to the output of tsn-m1, tsn-m2, tsn-m)
3, and the second and subsequent ones are stored as the second candidate, the third candidate, ... In the timing candidate storage unit 83 _n , and the equalization start timing signal is initially stored using the maximum value tsn-m1. The equalization range signal and the channel state estimation value are generated, the MIMO adaptive equalization unit 35 performs adaptive equalization processing on the received signal, and the signal sequence decoding unit 62 of the users 1 to N of the first time in the iterative decoding. A decision bit sequence is generated, and error detection processing is performed on the decision bit sequences of these users 1 to N by the error detection units 84 ₁ , ..., 84 _N. If an error is detected in the decoding determination bit sequence of user n in this error detection process, the detected output is fed back to the synchronization detection unit 82 _n for the user n signal of the synchronization / channel generation unit 81, and the timing candidate storage unit 83 is provided. The next timing candidate from _n, tsn-m2 in this case, is taken out, and the equalization start timing signal is used using the signal at this timing.
An equalization range signal and a channel state estimation value are generated, and these are used to perform adaptive equalization processing on the received signal by the user n.
For signals, an equalizer 69 _n for signal series _n (see FIG. 3)
And the equalized output is decoded by the signal sequence n decoding unit 62 _n (see FIG. 2) of the signal sequence decoding unit 62 to obtain the user n.
The decision bit sequence of is determined. An error detection process is also performed on this determination bit sequence, and when an error is detected again, the same process is performed using the next timing candidate tsn-m3. The same applies hereinafter. Note that the error detection process is, for example, CRC.
The method (Cyclic Redundancy Check) can be used. In FIG. 27, the processing stage 42 of the iterative processing unit 41 in the synchronization / channel generation unit 81 is assumed to be the one shown in FIG. 13, but it may be one shown in FIGS. 16, 20, 22, and 25.

Ninth Embodiment Next, the method of the present invention will be described. FIG. 28 shows the first to fourth
The processing procedure of the ninth embodiment of the method corresponding to the embodiment is shown.
In step S1, for each transmitter signal from the received signal, using the corresponding training symbol sequence, the equalization start timing t0 is detected, the equalization range candidate te is detected, the channel state H is estimated, and the received signal replica Re _n ( n =
1, ..., N) are generated in parallel, and these received signal replicas are subtracted from the received signal to obtain an error signal E
produces s. In step S ₂ , the processing number parameter s is initialized to 2, and in step S 3, the received signal replicas Re _{1 to} Re _N and the error signal Es generated in the s−1th processing are fetched, and in step S 4, the fetched error signal Es is fetched. Are added to the replicas Re _{1 to} Re _N to generate interference suppression reception signals r _{IC1 to} r _ICN , and the equalization start timing t0 is detected and equalized using the training symbol sequence corresponding to these interference suppression reception signals. The range candidate te is detected and the channel state H is estimated. In step S5, it is checked whether s is equal to S, and if not equal, in step S6, the received signal replicas Re _{1 to} Re _N are generated using t0, te, H and each training symbol sequence generated in step S4.
And the error signal Es is generated using these.
After that, s is incremented by 1 in step S7 and the process returns to step S3. If s is equal to S in step S5, t0, te, and H at that time are output to the MIMO adaptive equalization unit 35 in step S8. In step S9, adaptive equalization processing is performed on the received signal using t0, te, and H, and in step S10, decoding processing is performed on each equalized signal series (transmitter signal).

When performing error detection in the eighth embodiment, a plurality of timing candidates ts detected in step S4
n-m2, tsn-m3, etc. are stored, error detection processing is performed on each signal sequence decoded in step S11 as indicated by the broken line, and when an error is detected, the process returns to step S4 and is stored. Oita timing candidate tsn-m
2 is used to detect the equalization start timing t0 and the equalization range te. Further, when the replica corresponding to the detection timing having low reliability is not used, as shown by the broken line in FIG. 28, at step S1-1, t0, te is set for each transmitter signal.
Is detected, H is estimated, and the reliability of the detection timing is determined. In the above-described embodiment, the correlation output between the transmitter signal and the corresponding received signal (or interference suppression signal) with the energy index or the training symbol sequence, or the sum of squares of the channel state coefficient estimated in the previous process is less than or equal to a predetermined value. If so, it is determined that the reliability is low. Step S1
At -2, corresponding ones other than those determined to have low reliability of the detection timing in the replicas Re _{1 to} Re _N are generated,
These are subtracted from the received signal to obtain the error signal Es, and the process proceeds to step S2. Instead of step S6, step S
The reliability of the detection timing is determined in 6-1 and step S
It is also possible to generate only a replica whose reliability is not low in 6-2, subtract this from the received signal to generate Es, and move to step S7. If the steps S1 and S4 correspond to the first and fourth embodiments, the channel state H
Is collectively performed, and when corresponding to the second and third embodiments, is performed for each transmitter training symbol sequence. The timing offset compensation in the adaptive equalization process is based on the earliest one in the equalization start timing, and the equalization range TE used in the adaptive equalization process is the longest one in each equalization range candidate te. It is similar to the case described in.

Tenth Embodiment A tenth embodiment of the method corresponding to the fifth embodiment will be described with reference to FIG. In step S1, the error signal Es as the processing signal parameter is initialized to the reception signal R _m , in step S2 the processing frequency parameter s is initialized to 1, and in step S3, the transmission signals of N transmitters are set to 1 respectively.
The received energy per symbol is measured, and the transmitter training symbol sequences are ordered in descending order. Here, it is assumed that the order is n = 1, 2, ..., N. In step S4, it is checked whether s is 1, and if it is 1, in step S5, the parameter n in the descending order of received energy is initialized to 1, and in step S6 it is checked whether s is 1 and 1 If there is, the error signal Es is obtained in step S7.
For the training symbol sequence TSS _n using t
Detection of 0, te, estimation of H _n , generation of a replica Re _n using these and the training symbol sequence TSS _n , subtracting that Re _n from Es, updating Es with the result, and storing Re _n To do. It is checked in step S8 whether n is equal to N. If not, n is incremented by 1 in step S9.
Then, the process returns to step S6. If n is equal to N in step S8, it is checked in step S10 whether s is equal to S. If they are not equal, s is incremented by 1 in step S11 and the process proceeds to step S12. When n = N in step S8, the error signal Es is an error signal obtained by subtracting all replicas Re _{1 to} Re _N from the received signal. If s is not 1 in step S4, the process proceeds to step S12. In step S12, the reception signal replicas Re _{1 to} Re _N and the error signal Es generated in the s−1th processing are fetched, and the process proceeds to step S5. Step S6
In s is not 1, in step S13, by adding the Re _n to generate an interference suppression received signal r _ICn in Es fetched in step S12, this interference suppression received signal t0, te, the training symbol sequence TSS _n To estimate H and generate a received signal replica Re _n using these and the training symbol sequence TSS _n .
Stored, and by subtracting the Re _n from the interference suppression received signal to generate an error signal Es, proceeds to step S8 to update the Es thereto.

If s is equal to S in step S10, then t0, te, and H at that time are output to the MIMO adaptive equalization unit 35 in step S14. Incidentally, in the s = Sth process, step S1
3, the reception signal replica Re _n is not generated, and Es is not generated when n = N. In step S15
The MIMO adaptive equalization unit 35 adaptively equalizes the received signal, and the signal sequence decoding unit 62 decodes the symbol sequence separated in step S16. When performing the processing corresponding to the eighth embodiment, the timing candidates tsn-m2, t are determined in step S13.
sn-m3 and the like are stored, error detection processing is performed in step S17, and when an error is detected, step S18
Then, n is set in the processing order of the transmitter signal and step S
Move to 13. When the order of the transmitter signals to be processed is reset for each number of times of processing, the process returns to step S3 after step S11, as indicated by the broken line in FIG. For the received energy measurement in step S3 after s = 2, the replicas Re ₁ to
Perform on Re _N.

Eleventh Embodiment An eleventh embodiment of the method corresponding to the sixth embodiment will be described with reference to FIG. In step S1, t0, t from the received signal
e is detected, H is estimated, and Re _{1 to} Re _N and Es are generated. In step S2, the received energy per symbol is measured for each of the N transmitter signals, and the transmitter training symbol sequences are ordered in descending order.
Here, the order is n = 1, 2, ..., N. In step S3, s is initialized to 2, and in step S4, s-1
The reception signal replicas Re _{1 to} Re _N and the error signal Es for the second time are fetched, n is initialized to 1 in step S 5, and step S 5
By adding the Re _n to generate an interference suppression received signal r _ICn captured in Es at 6, this interference suppression received signal t0,
te is detected, H _n is estimated, and these are used to generate and store a received signal replica Re _n, and this Re _n is subtracted from the interference suppression received signal r _ICn to generate an error signal Es, Es To this. In step S7, it is checked whether or not n is equal to N. If they are not equal, n is incremented by 1 in step S8 and the process returns to step S6. If n is equal to N in step S7, it is checked in step S9 whether s is equal to S. If they are not equal, s is incremented by 1 in step S10 and the process returns to step S4. If s is equal to S in step S9, then t0, te, and H at that time are output to the MIMO adaptive equalization unit 35 in step S11. Incidentally, in the s = Sth processing, step S6
In, the reception signal replica Re _n is not generated, and Es is not generated when n = N. MIMO adaptive equalization processing is performed on the received signal in step S12, and step S13
Then, the equalized symbol sequence is decoded into a signal sequence. In this case as well, although not particularly shown, it is also possible to perform an error detection process and obtain te and H by using a timing candidate if there is an error.

Twelfth Embodiment Next, referring to FIG. 31, a first method corresponding to the seventh embodiment will be described.
Two examples will be described. In step S1, the error signal parameter Es is initialized with the received signal R _m , and in step S2
In step S3, the processing number parameter s is initialized to 1.
Then, the received energy per symbol is measured for the training symbol sequences of N transmitters, and the transmitted training signal sequences are divided into groups G1 to GX in descending order. It is checked in step S4 if s is 1, and if it is 1, the group parameter x is initialized to 1 in step S5, and it is checked in step S6 if s is 1, and if it is 1, the group GX is set in step S7. Error signal (received signal) Es is subjected to parallel processing using the training symbol sequence belonging to
Detects, estimates the H, further generates the reception signal replica Re _n Using these, stored and subtracted each replica Re _n generated from the error signal Es, updates Es in the results.

In step S8, it is checked whether x is equal to X,
If they are not equal, x is incremented by 1 in step S9 and step S
Returning to step 6, if x and X are equal in step S8, it is checked in step S10 whether s is equal to S. If they are not equal, s is incremented by 1 in step S11 and the process proceeds to step S12. If s is not 1 in step S4, the process proceeds to step S12. Re _{1 to} Re _N generated in the s−1th time in step S12
And Es are fetched and the process proceeds to step S5. Step S6
If s is not 1 in step S13, in step S13, each Re _n corresponding to the training symbol sequence belonging to the group GX is added to the error signal Es to generate a plurality of interference suppression reception signals, and the plurality of interference suppression reception signals are generated. The signals are processed in parallel by using the training symbol sequences belonging to the group GX to detect t0 and te, respectively, H is estimated, and further, each received signal replica Re _n is generated, updated and stored, and Corresponding generated replicas Re _n are respectively subtracted from the plurality of interference suppression reception signals to generate E _S , Es is updated to Es, and the process proceeds to step S8. If s is equal to S in step S10, step S
At 14, the t0, te, and H at that time are supplied to the MIMO adaptive equalization unit 35. In step S13, at the Sth time, Re _n
Is not generated, and Es is not generated when x = X. In step 15, the received signal is adaptively equalized, and in step 16, the equalized symbol sequence is decoded. Although not shown in the figure, if there is an error by performing error detection processing if necessary, using timing candidates,
It is also possible to recalculate t0, te, and H.
Furthermore, although not particularly shown in FIGS. 29 to 31, in these methods, the reliability of the detection timing may be determined and only the replicas that do not correspond to the timing with low reliability may be output. Although the equalization range is detected in the above description, the equalization range may have a fixed length and may not be detected.

In the above description, the synchronization timing generator 12 _n and the start timing / range candidate generator 21 _n respectively obtain the correlation between the training symbol sequence TSS _n and the received signal, but the correlation generated by the synchronization timing generator 12 _n May be stored and used by the start timing / range candidate generation unit 21 _n . When synchronization detection and channel estimation are performed in the serial type and composite type processing stages 42 as shown in FIGS. 22 and 25, for example, in the serial type, channel estimation is performed for each user in the first processing stage 42 ₁ . It is considered that the accuracy of channel estimation deteriorates. For example, when the received signal of the user 1 (having the highest received power) is detected by the channel estimation means for each user with the configuration shown in FIG. 17, the user 2 is included in the received signal to be used. This is because the received signal of the user N is included. By the way, in the second processing stage 42 _{2 and} thereafter, it is considered that the received signals used for channel estimation cancel the received signals of other users, so that it does not lead to a large deterioration in channel estimation accuracy.
Therefore, as a means for solving this problem, the first processing stage 4
2, ₁ as indicated by the broken line in FIG. 22 sync detector 82 ₁ -
The channel estimation unit 28 (shown in FIG. 9) includes the synchronization start timings of all users detected at 82 _N , the training symbol sequences TSS _{1 to} TSS _N, and the received signal (a signal that has not been canceled by the replica). It is sufficient to input all the channel states H to the replica generation units 45 _{1 to} 45 _{n and} to supply them to the replica generation units 45 _{1 to} 45 _n .

[0040]

As described above, according to the present invention, MI
It is possible to improve the estimation accuracy of the channel state in the MO receiver, the detection accuracy of the equalization start timing, and the accuracy of determining the equalization range. Especially in packet communication, channel state estimation, equalization start timing detection,
It is considered necessary to determine the equalization range, and in such a case, the present invention makes it possible to shorten the training symbol sequences for channel estimation and synchronization, and to reduce the amount of information to be transmitted. Can be increased.

[Brief description of drawings]

FIG. 1 is a diagram illustrating a configuration example of a transmitter corresponding to a turbo receiver.

FIG. 2 is a diagram illustrating a functional configuration example of a MIMO turbo receiver.

FIG. 3 is a diagram illustrating a functional configuration example of a MIMO adaptive equalization unit.

FIG. 4 is a diagram showing a functional configuration of a single-user adaptive equalizer.

5A is a diagram for explaining a conventional symbol synchronization timing determination method, and FIG. 5B is a diagram for explaining a conventional equalization range determination method.

FIG. 6 shows the structure of a proposed MIMO receiver.

FIG. 7 is a diagram illustrating a functional configuration example of a synchronization timing generation unit.

FIG. 8 is a diagram illustrating a functional configuration example of a start timing / range candidate generation unit.

FIG. 9 is a diagram illustrating a functional configuration example of a channel estimation unit.

FIG. 10 is a diagram showing a functional configuration of a MIMO receiver that generates a proposed equalization start timing signal common to the antennas and an equalization range candidate signal.

11 is a diagram showing a configuration example of a synchronization timing generation unit in FIG.

12 is a diagram showing a configuration example of a start timing / range candidate generation unit in FIG.

FIG. 13 is a diagram showing a functional configuration of a receiver according to the first embodiment of the present invention.

14 is a diagram showing a functional configuration example of a processing stage 42 _s in FIG.

15 is a diagram showing a functional configuration example of a replica generator 45 in FIG.

FIG. 16 is a diagram showing a functional configuration example of a processing stage 42 _s according to a second embodiment of the present invention.

17 is a diagram showing a functional configuration example of a channel estimation unit 28 in FIG.

FIG. 18 is a diagram showing a functional configuration example of a receiver according to a third embodiment of the present invention.

FIG. 19 is a diagram showing a functional configuration example of a receiver of a fourth embodiment of the present invention.

20 is a diagram showing a functional configuration example of a processing stage 42 _s in FIG. 19.

FIG. 21A is a diagram showing an example of the functional configuration of a processing order detection unit 48 in the fifth embodiment of the present invention, and B is a diagram showing another example.

FIG. 22 is a diagram showing a functional configuration example of a processing stage 42 _s in the fifth embodiment.

FIG. 23 is a diagram showing a functional configuration example of a receiver of a sixth embodiment of the present invention.

FIG. 24 is a diagram showing an example of the functional configuration of a processing order detection unit 48 in the sixth embodiment.

FIG. 25 is a diagram showing a functional configuration example of a processing stage 42 _s according to a seventh embodiment of the present invention.

FIG. 26 is a diagram illustrating a configuration example in which a processing unit is repeatedly used to form each processing stage.

FIG. 27 is a diagram showing a main part of a functional configuration of an eighth embodiment of the present invention.

FIG. 28 is a flow chart showing an example of the processing procedure of the method of the present invention corresponding to the first to fourth embodiments.

FIG. 29 is a flowchart showing an example of the processing procedure of the method of the present invention corresponding to the fifth embodiment.

FIG. 30 is a flowchart showing an example of the processing procedure of the method of the present invention corresponding to the sixth embodiment.

FIG. 31 is a flowchart showing an example of the processing procedure of the method of the present invention corresponding to the seventh embodiment.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Shigeru Tomisato             2-11-1, Nagatacho, Chiyoda-ku, Tokyo Stock             Ceremony company NTT Docomo (72) Inventor Hiroto Suda             2-11-1, Nagatacho, Chiyoda-ku, Tokyo Stock             Ceremony company NTT Docomo F term (reference) 5K022 DD00 DD32 DD34 FF00                 5K046 AA05 BB01 EE02 EE03 EE06                       EE37 EF03

Claims (10)

[Claims] 1. An equalization start timing signal, an equalization range signal, and a channel state estimation value for a signal obtained by receiving an integer N transmission signal series of 2 or more integers in the same frequency band by an integer M antennas of 2 or more. A MIMO adaptive equalization unit for performing adaptive equalization processing to separate it into N transmitter-compatible symbol sequences, and a signal sequence decoding unit for decoding the N output symbol sequences from the MIMO adaptive equalization unit, respectively. And a training sequence generation unit that respectively generates a training symbol sequence of the same symbol sequence as the training symbol sequence in each transmission signal sequence, and inputs the above-mentioned training symbol sequence and received signal,
First to generate equalization start timing signal and equalization range signal
A synchronization detection unit, a channel estimation unit that estimates the channel state by inputting the training symbol sequence, the equalization start timing signal, the equalization range signal, and the received signal, the training symbol sequence, the equalization start timing signal, and the like. The replica generation unit that generates a replica of the received signal of the training symbol with the digitized range signal and the estimated channel state as input, and the received signal and the received replica are input to receive the received signal replica of the training symbol sequence from the received signal. An error signal generation unit that subtracts an error signal, the above error signal, and at least one received signal replica for a training symbol sequence of a target transmission signal sequence are input, and these are added to obtain the target transmission signal. A signal for generating an interference-suppressed received signal that suppresses signals other than the received signal of the sequence. A MIMO receiver comprising: a calculation unit; and a second synchronization detection unit that receives the interference suppression received signal and the training symbol sequence as input and generates an equalization start timing signal and an equalization range signal. 2. The receiver according to claim 1, wherein the first synchronization detection unit, the replica generation unit, the addition unit, and the second synchronization detection unit are provided for each transmission signal sequence, and the addition unit includes: There is only one training symbol sequence received signal replica of the target transmission signal sequence that is input. 3. The receiver according to claim 2, wherein the reliability determination unit determines the reliability of the detection timing detected by the synchronization detection unit, and the detection timing determined by the reliability determination unit to have low reliability. And means for not outputting the replica corresponding to the one determined to be. 4. The receiver according to claim 1, further comprising: a processing order detector that measures received energy per symbol for each of N transmission signal sequences and determines a larger order of the received energy. Subtraction for subtracting the detection unit, the channel estimation unit, the replica generation unit, and the replica from the replica generation unit from the received signal input to the first synchronization detection unit and supplying the subtraction signal to the second synchronization detection unit. The transmission signal sequence processing unit with the unit is cascaded in the determined order, the output signal of the subtraction unit of the last transmission signal sequence processing unit of the cascade connection is obtained as an error signal from the error signal generation unit, The training symbol sequence received signal replica input to the adder corresponds to the one having the largest received energy per symbol. 5. The receiver according to claim 1, wherein the received energy per symbol is measured for N transmission signal sequences, and the N transmission signal sequences are divided into a plurality of groups in descending order. A parallel processing unit that includes a processing order detection unit and is configured to perform processing by the first synchronization detection unit, the channel estimation unit, and the replica generation unit in parallel for a plurality of transmission signal sequences belonging to the group is configured, These parallel processing units are connected in series in the order of the larger groups by a subtraction unit that subtracts the replica from the replica generation unit from the input signal of the first synchronization detection unit and inputs the subtracted signal to the next parallel processing unit. . 6. A signal obtained by receiving an integer N number of transmission signal sequences of the same frequency band of 2 or more by an integer M number of antennas of 2 or more by using equalization start timing, an equalization range, and a channel state. Of the equalization start timing and the equalization range using the received signal and the training symbol sequence for each transmission signal sequence. First synchronization detection process for detection, received signal, training symbol sequence for each transmission signal sequence, channel estimation process for estimating channel state using equalization start timing and equalization range, and training symbol for each transmission signal sequence A replica generation process for generating a replica of the received signal of the training symbol using the sequence, the equalization start timing, and the estimated channel state, Error signal generation process of generating an error signal by subtracting the training symbol reception signal replica from the reception signal, and adding a reception signal replica of at least one target transmission signal sequence to the error signal except the target transmission signal sequence Interference suppression reception signal generation process for generating an interference suppression reception signal that suppresses the reception signal of, the second synchronization for detecting the equalization start timing and the equalization range using the interference suppression reception signal and the corresponding training symbol sequence. A MIMO receiving method comprising: a detection process. 7. The receiving method according to claim 6, wherein the first synchronization detection process and the replica generation process are executed in parallel for each transmission signal sequence, and N generated training symbol reception signal replicas are generated as the error signal. In the process, the error signal is generated by subtracting it from the received signal, and the interference suppression received signal generating process of adding the replica for each one of the target transmission signal sequences to the error signal is executed in parallel. The second synchronization detection process is performed on the generated N interference suppression reception signals in parallel. 8. The receiving method according to claim 6, wherein the process of determining the reliability of the detected equalization start timing, and the replica for the transmission signal sequence corresponding to the equalization start timing determined to have low reliability Stop using. 9. The receiving method according to claim 6, wherein the received energy per symbol is measured for each of the N transmitted signal sequences, the order of the received energy is determined in order, and the received signal sequence is determined for each transmitted signal sequence. The synchronization detection process, the channel estimation process, the replica generation process, and the error signal generation process are executed in the order determined above, and the error signal generated by the processing for each transmission signal sequence is used as the reception signal used for the processing of the transmission signal sequence. Are added to form a reception signal for the next processing for each transmission signal sequence, and the interference suppression reception signal generation process and the second synchronization detection process are executed for the error signal generated by the last processing for each transmission signal sequence. . 10. The receiving method according to claim 6, wherein the received energy per symbol is measured for each of N transmitted signal sequences, and a plurality of N transmitted signal sequences are grouped in descending order of received energy. For each group, the first synchronization detection process, the channel estimation process, and the replica generation process are performed in parallel for each transmission signal sequence belonging to the group, in the order of the group with the largest received energy. Then, at that time, the replica generated in the processing for each group is subtracted from the received signal for the group processing to generate an error signal, and this error signal is set as the received signal for the processing for the next group, and the final For the error signal obtained by the processing for each group, the interference suppression signal generation process and the second synchronization detection error To run.