JP4247532B2 - MIMO-OFDM reception system and receiver with high-precision timing recovery - Google Patents

Description

  The present invention relates to a radio transmission system based on orthogonal frequency division multiplexing (OFDM) such as mobile communication, wireless LAN, terrestrial digital TV broadcasting, etc. and OFDM such as MC-CDMA, and signals using a plurality of (single) transmission / reception antennas. The present invention relates to a reception method and a receiver in a MIMO-OFDM method that is extended to a spatially multiplexed MIMO method.

  In recent years, in mobile communications and wireless LANs, OFDM has attracted attention as a wireless transmission method that can realize high-reliability and high-speed transmission. The OFDM scheme is multicarrier transmission with a guard interval (GI) inserted, and can realize good transmission characteristics even under frequency selective fading conditions. With the aim of further increasing the speed of the OFDM system, the application of a MIMO (Multiple Input Multiple Output) system that spatially multiplexes signals using a plurality of transmission / reception antennas is being studied, and research is being vigorously conducted as MIMO-OFDM. .

FIG. 9 shows a conventional MIMO-OFDM system configuration. As shown in FIG. 9, in the conventional MIMO-OFDM system, the _{N T} transmit antennas and _{N R} receive antennas, the OFDM signal is spatially multiplexed, OFDM signals to different from each transmit antenna In the receiver, signal separation / detection of N _T OFDM signals is performed from N _R received signals.

The configuration of a conventional MIMO-OFDM transmitter is shown in FIG. As shown in FIG. 10, in a conventional MIMO-OFDM transmitter, first, an information bit sequence is divided into N _T sequences, each of which is error-correction-encoded and interleaved, and N OFDM by an OFDM modulator. A subcarrier OFDM signal is generated. Next, the OFDM signal is generated by converting the modulated signal in each subcarrier into a time signal by IFFT, and adding a part of the latter half of the time signal as GI to the first half of the time signal. A time signal generated by IFFT and a GI are collectively called an OFDM symbol, and a packet is composed of a plurality of OFDM symbols. These OFDM signals are waveform-shaped for band limitation, and then frequency-converted to an RF frequency band and transmitted from each transmitting antenna.

The configuration of a conventional MIMO-OFDM receiver is shown in FIG. As shown in FIG. 11, in a conventional MIMO-OFDM receiver, first, the received signal received by each antenna passes through a matched filter similar to the waveform shaping filter of the transmitter, and timing recovery is performed. Next, after removing the GI from the timing, is converted into sub-carrier signals of the N by FFT, since the sub-carrier signals of the N from the nature of the OFDM signal are orthogonal to each other, N _R number of each sub-carrier _NT signals from the received signal are detected by a maximum likelihood detector, and a detection signal is output. Note that the subcarrier signal is a signal obtained by converting a received signal or a synthesized signal obtained by linearly synthesizing the received signal into a signal of each subcarrier in the frequency domain by FFT. The detection signal of each subcarrier is serial-parallel converted for each transmission signal sequence, decoded after deinterleaving, and all sequences are integrated again. Technologies related to recent MIMO-OFDM are disclosed in Non-Patent Literature 1 to Non-Patent Literature 7.

  Conventional MIMO-OFDM receivers can be broadly classified as follows: (I) A linear synthesis type (for example, Non-Patent Document 2, Non-Patent Document 3, Non-Patent Document 5, and (See Non-Patent Document 6), (II) Maximum likelihood estimation type for calculating the likelihood by subtracting the replica signal from each received signal and performing signal determination (for example, Non-Patent Document 1, Non-Patent Document 2, Non-Patent Document (See Patent Literature 4 and Non-Patent Literature 6).

  The maximum likelihood estimation type MIMO-OFDM receiver has an excellent error rate characteristic although the calculation amount is increased as compared with the linear combination type, but the co-channel interference (CCI) as shown in FIG. The existing MIMO-OFDM system is not assumed, and it is considered that the transmission characteristics are significantly deteriorated by CCI from a neighboring cell or the like. However, as far as the inventor knows, no study has been made on a MIMO-OFDM reception system for suppressing CCI. In the first place, there is already a technique for suppressing CCI as disclosed in Non-Patent Document 7, for example, by stopping MIMO transmission, transmitting a single OFDM signal, and operating a plurality of receiving antennas as an adaptive array. Although proposed, this technique is not a technique related to the MIMO-OFDM reception system.

Patent Document 1 discloses a technique for suppressing interference due to a multipath delay wave exceeding the CCI and guard interval in the OFDM system and a technique for suppressing the latter interference in the MIMO-OFDM system. There is no suggestion or disclosure about the technique for suppressing both interferences in the system.
Japanese Patent Application No. 2003-003813 A. van Zelst, R. van Nee, and GAAwater, “Space Division Multiplexing (SDM) for OFDM Systems Multiplexing (SDM) for OFDM systems), IEEE Vehic.Tech. Conf., May 2000, p.1070-1074 Y. Li, J. H. Winters, N. R. Somberberger, "MIMO-OFDM for Wireless Communications: Signal Detection with Enhanced Channel Estee MISSION (MIMO-OFDM for wireless communications: signal detection with enhanced channel estimation) ", IEEE Trans. On Commun., Vol. 50, No. 9, September 2002, p.1471-1477 H. Sampath, S. Talwar, J. Terrado, V. Atzeg, A. Paul Large (H. Sampath, S. Talwar, J. Tellado, V. Erceg and A. Paulraj), “A Force・ Generation MIMO-OFDM broadband wireless system: design, performance, and field trial results ”, IEEE Comm. Magazine, No. Volume 40 Number 9, September 2002, p.143-149 S. Hori, M. Mizoguchi, T. Sakata, M. Nozokura (S. Hori, M. Mizoguchi, T. Sakata, and M. Norikura), “A New Branch Metric Generation Method for Soft Decision Viterbi Decoding in Coded OFDM-SDM Systems Employing MLD Over Frequency Selective MIMO Channels, IEICE Trans. Fundamentals. (IEICETrans. Fundamentals.), Volume E85-A, July 2002, p.1675-1684 S. Kurosaki, Y. Asai, T. Sugiama, M. Umehira, “SDM-COFDM Scheme Employing Feedforward Interchannel Interface Lens canceller for MIMO based broadband wireless LANs (SDM-COFDM scheme, feed-forward inter-channel interference cancellers for MIMO based broadband wireless LANs), IEICE Trans. Commun., Vol. E86-B, 2003 January, p.283-290 Jun Suyama, Hiroshi Suzuki, Kazuhiko Fukawa, "MIMO-OFDM turbo equalization for multipath delay exceeding guard interval", IEICE Technical Report, RCS2002-312, March 2003 Toru Nishikawa, Yoshitaka Hara, Keisuke Hara, "A Study on Adaptive Array for OFDM in Mobile Communications", IEICE Technical Report, RCS2000-232, March 2001 L. R. Bahar, J. Cocca, F. Jelenek, J. Rabib (LR Bahl, J. Cocke, F. Jelinek and J. Raviv), “Optimal Decoding Off Linear Codes “Optimal decoding of linear codes for minimizing symbol error rate”, IEEE Trans. On Inform. Theory, Vol. 20, March 1974, p.284- 287 P. Robertson, E. Villebrun and P. Hoecher, “A Comparison Off Optimal and Suboptimal MAP Decoding Algorithm Operating in the Log Domain (Acomparison of optimal and sub-optimal MAP decoding algorithms operating in the log domain) ”, ICC95 (in Proc. of ICC95), June 1995, p.1009-1013 G. Battail, “Ponderation des symbols decodes par l'algorithmede Viterbi”, Ann Telecomm. Vol. 42, January 1987, p. 31-38 J. Hangenauer and P. Hoeher, “AViterbi algorithm with soft-decision outputs and its application”, Improv. Off IEEE GLOBECOM89 (inProc. Of IEEE GLOBECOM89), November 1989, p1680-1686 S. Haykin, "Adaptive filter Theory", 3rd edition, Prentice Hall, 1993 K. Fukawa, “A suboptimal receiver structure for joint spatial-temporal filtering and multiuser detection in mobile radio communications.” Sub-Optimal Receiver Structure for Joint Spatial Temporal Filtering and Multi-User Detection in Mobile Radio Communications ”, IEEE Trans. On Vehicular Tech., Vol. 49, No. 4, July 2000, p.1196-1206 IEEEStd 802.11a, `` High-speed Physical Layer in the 5 GHz Band '', 1999 P. W. Walniansky, G. J. Fostiny, G. D. Golden, R. A. Valenzuela (PW Wolniansky, GJ Foschini, GD Goldenand RA Valenzuela), BLAST: an architecture for realizing very high data rate over the rich-scattering wireless channel (V-BLAST), 1998 URSI in. Symposium on signal system And Electronics (1998 URSI Int. Sympo. On Signals, Systems, and Electronics), 1998, p.295-300 Atsushi Suyama, Yoshitaka Hara, Hiroshi Suzuki, Yoshihide Kamio, Kazuhiko Fukawa, "OFDM reception method for multipath environment with delay profile exceeding guard interval", IEICE Technical Report, RCS2001-175, November 2001

The conventional MIMO-OFDM reception system does not assume an environment in which CCI exists as shown in FIG. 12, and there is a problem that transmission characteristics are significantly deteriorated if CCI exists. This is because in the conventional MIMO-OFDM system, signals corresponding to the number of transmission antennas are spatially multiplexed, and these signals are separated and detected in the receiver, so that the detector is less resistant to interference. In addition, in the CCI environment, the accuracy of timing reproduction performed in the receiver deteriorates, so that there is a high possibility that an error occurs in timing. Therefore, when an error occurs in timing, since past and future OFDM symbols leak into the Fourier transform section, code interference interference (ISI) and carrier due to the leakage of the current OFDM symbol shorter than the Fourier transform section. Interference (ICI) occurs. That is, the problem of causing further deterioration of transmission characteristics due to ISI or ICI occurs. Further, in the CCI environment, when a multipath delay exceeding the GI of the desired signal arrives, ISI and ICI are generated, so that the transmission characteristic is similarly deteriorated. Furthermore, it is difficult to estimate the channel response of a desired signal necessary for signal separation / detection in a CCI environment. This is because a desired signal is buried in interference due to CCI, resulting in a low SNR environment, and the estimation accuracy of channel estimation is significantly degraded. In the first place, in CCI environment, MIMO transmission is stopped, a single OFDM signal is transmitted, and a plurality of receiving antennas are operated as an adaptive array, so that CCI can be suppressed and high reliability can be realized. Therefore, there arises a problem that the transmission rate is reduced to 1 / _NT of the transmission rate of the normal MIMO-OFDM system.

  The present invention has been made under the circumstances as described above, and an object of the present invention is to suppress co-channel interference and interference generated by multipath delay waves exceeding GI and timing recovery error in the MIMO-OFDM system. In addition, it is possible to compensate for the degradation of transmission quality caused by the degradation of estimation accuracy of channel estimation, and to achieve high-speed signal transmission at all times, and the MIMO-OFDM reception system with high-accuracy timing recovery and reception Is to provide a machine.

The present invention relates to a MIMO-OFDM reception system and receiver with high-precision timing recovery, and the above object of the present invention is to provide N _T (N _T is a positive integer) different error correction coded OFDM signals. N _R (N _R is a positive integer) timing recovery processing for recovering timing from received signals received by antennas, and L (L is a positive integer) number of white signals to output the synthesized signal A linear synthesis process, L Fourier transform processes for converting the synthesized signal into a subcarrier signal, a signal detection process for performing a likelihood calculation using the subcarrier signal and outputting a detection signal, and the detection signal _NT error correction decoding processing that performs error correction decoding by deinterleaving, parameter estimation processing that collectively estimates parameters used in the linear synthesis processing and the signal detection processing, This is achieved by comprising a repetitive control process in which a part or all of the process is repeated.

Furthermore, the object of the present invention is to detect a temporary timing candidate based on a cross-correlation between the received signal and a known preamble signal or an autocorrelation of the received signal as the timing recovery process, and the parameter for each temporary timing candidate. The estimation process is performed, and the provisional timing candidate that minimizes the error between the composite signal output in the linear composition process using the estimated weighting coefficient and its replica is used as a timing, or the provisional timing candidate In the most probable candidates, the number of taps is increased and the parameter estimation process is performed. The number of taps necessary for the parameter estimation process is determined and the provisional timing is corrected according to the estimated weighting factor of each tap. Or by performing the above two timing reproduction processes, or Using the fractional interval transversal filter that filters the received signal and the combiner that synthesizes the output as the linear synthesis processing, using the weight coefficient of the fractional interval transversal filter estimated in the parameter estimation processing. By performing linear synthesis of the received signal, or as the Fourier transform process, the guard interval component of the synthesized signal is removed by the timing obtained from the timing reproduction process, or Fourier transform is performed, or The ratio of the power of the desired signal that changes depending on the starting position of the Fourier transform and the power of the generated interference is calculated from the estimated value of the channel response obtained from the parameter estimation process, and the Fourier transform is performed at the position where the power ratio becomes maximum. Or where the calculated power ratio is maximized Then, by performing two Fourier transforms with different ranges and performing in-phase synthesis, by converting the synthesized signal into a subcarrier signal, or as the signal detection processing, for L subcarrier signals Then, a replica signal is generated using a transfer function derived by Fourier transforming the estimated value of the channel response, and maximum likelihood estimation is performed using an absolute value square value of a difference between the subcarrier signal and the replica signal as a metric. Further, at the time of iterative control, the log posterior likelihood ratio of the coded bit that is the output of the error correction decoding process is fed back and used as a priori information to perform maximum posterior probability estimation, or the transfer function performs linear processing for separating the N _T signals, sequentially decodes perform interference cancellation and turn signal separation reliability of L sub-carrier signals using a By performing a process of outputting a detection signal according to the above, or as the error correction decoding process, the detection signals corresponding to the number of subcarriers are parallel-serial converted, deinterleaved, and error correction of software input / soft output is performed. Decoding, outputting a log likelihood ratio between the information bits and the coded bits, and performing a process of determining a received bit sequence based on the log likelihood ratio of the information bits, or the parameter estimation As a process, the weighting factor of the fractionally spaced transversal filter and the channel response used for the signal detection process are collectively calculated, and the square of the absolute value of the difference between the synthesized signal and the replica of the synthesized signal generated using the preamble signal The value is estimated by the least square method so that the value is minimized, or the received signal and the preamble signal are An error is detected by a cyclic code for the received bit sequence as an estimation based on the eigenvalue and eigenvector of the vector autocorrelation matrix of the element, or as the iterative control process, and no error is detected. In this case, the reception process is terminated, and when an error is detected, the log likelihood ratio of the coded bits or the expected value of the subcarrier signal calculated therefrom is used. A signal including the Fourier transform process for the synthesized signal by performing part or all again and repeating the series of processes, or as the Fourier transform process and the signal detection process Whether to perform discrete Fourier transform / signal separation linear processing for separating _NT signals by linear processing for separation and outputting the detection signal Alternatively, during the repetitive control, a desired subcarrier signal component is extracted from the combined signal by replica subtraction, and the extracted signals are combined by linear processing including the Fourier transform processing, and the detection signal is output. Or performing the parameter estimation process, and using the estimated weighting factor, the synthesized signal output by the linear synthesis process and its replica error, and the reception Performing the parameter estimation process on the received signal without performing the linear synthesis process and comparing the error with the replica of the received signal generated with the estimated value of the channel response, This is achieved more effectively by not performing the linear synthesis process when the value is small.

In short, in the present invention, N _T different OFDM signals that have been subjected to error correction coding are received by N _R antennas, and timing is recovered by timing recovery processing that also operates in the CCI environment. CCI is suppressed by linear synthesis processing with L transversal filters to be whitened, and a synthesized signal is output. Next, GI removal and Fourier transform are performed on the combined signal to convert it into a subcarrier signal. Further, likelihood calculation is performed using the transfer function estimated in each subcarrier, _NT signal detection processing is performed, and the output is subjected to error correction decoding after deinterleaving.

  In the present invention, the parameters used in the linear synthesis process and the signal detection process are estimated collectively. Specifically, the absolute square value of the difference between the synthesized signal obtained by synthesizing the received signal by the linear synthesis process and the replica of the synthesized signal generated by the inner product of the preamble signal known by transmission and reception and the estimated value of the channel response is minimized. The parameter is estimated by the least square method so as to satisfy the above, or the parameter is estimated by the eigenvalue and eigenvector of the autocorrelation matrix of the vector having the received signal and the preamble signal as elements.

  Furthermore, in the present invention, when an error is detected in the decoded result, a process for repeating a part or all of the above process is performed. At that time, each process is performed by feeding back the expected value of the subcarrier signal that can be calculated from the log likelihood ratio of the encoded bits, which is the output of the decoder. In addition, a desired subcarrier signal of an arbitrary transmitting antenna is extracted by replica subtraction from the combined signal in which CCI is suppressed, and the signal component is combined by linear processing including Fourier transform processing, and the logarithm of the encoded bit A turbo equalization process that outputs a likelihood ratio is performed. Since this processing directly performs replica subtraction on the synthesized signal in the time domain, it is possible to remove ISI and ICI that occur when multipath delay exceeding GI arrives.

In the present invention, the above-described means enables the following highly reliable MIMO-OFDM transmission.
-CCI can be suppressed by linear synthesis processing-High-accuracy timing reproduction can be performed in the CCI environment-Parameter estimation can be performed well in the CCI environment-Reliable high-speed transmission can be maintained in the CCI environment without reducing the transmission speed Can achieve good transmission characteristics even when received power is low due to repetitive processing in cooperation with error correction processingInterference that occurs when multipath delay exceeding GI arrives in CCI environment Can be removed

  According to the MIMO-OFDM reception system and receiver with high-accuracy timing recovery according to the present invention, it is possible to suppress co-channel interference and interference due to multipath delay waves exceeding the guard interval, and to correct timing synchronization errors. There is an excellent effect that a reliable MIMO-OFDM reception system can be realized.

  Specifically, since a system such as a wireless LAN is unlicensed, a situation is assumed in which a plurality of access points are spatially arranged in a limited frequency band. In these systems using the MIMO-OFDM scheme, it is considered that transmission characteristics deteriorate due to interference from other access points and user terminals, and communication may not be possible depending on the situation. In particular, in a wireless LAN, there are many cases in which a user terminal is stationary and used, so that the possibility of improving the communication status decreases. Therefore, it is considered that a reception method that can suppress interference becomes important. Therefore, it goes without saying that the present invention having an excellent effect of eliminating co-channel interference is very useful for a system such as a wireless LAN.

  Further, when the service range of the wireless LAN is widened, transmission characteristics are similarly deteriorated due to interference generated by multipath delay waves exceeding the guard interval. The MIMO-OFDM receiver with high-precision timing recovery according to the present invention is a receiver that can operate satisfactorily even when both co-channel interference and multipath delay waves exceeding the guard interval exist. A highly reliable MIMO-OFDM reception system can be realized.

That is, according to the present invention, as follows:
-CCI can be suppressed by linear synthesis processing-High-accuracy timing reproduction can be performed in the CCI environment-Parameter estimation can be performed well in the CCI environment-Reliable high-speed transmission can be maintained in the CCI environment without reducing the transmission speed Can achieve good transmission characteristics even when received power is low due to repetitive processing in cooperation with error correction processingInterference that occurs when multipath delay exceeding GI arrives in CCI environment It is possible to obtain an excellent effect that can be removed.

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

FIG. 1 shows an example of an embodiment for carrying out the present invention. As shown in FIG. 1, the features of the present invention are a timing regenerator for recovering timing from received signals received by _NR antennas, and L linearities for whitening the received signal and outputting a synthesized signal. A combiner, an L number of Fourier transformers that convert the combined signal into a subcarrier signal, a signal detector that performs likelihood calculation using the subcarrier signal and outputs a detected signal, and an error in the detected signal _NT error correction decoders that perform correction decoding, a parameter estimator that collectively estimates parameters used in the linear synthesizer and the signal detector, and a repetitive controller that repeatedly performs part or all of the above processing It is in the point comprised from.

Embodiments of a MIMO-OFDM reception system and a receiver according to the present invention will be specifically described. Here, consider a MIMO-OFDM transmission system that receives N _T different OFDM signals subjected to error correction coding using N _R antennas.

  The configuration of a MIMO-OFDM receiver capable of suppressing co-channel interference in the present invention is shown in the block diagram of FIG. The MIMO-OFDM receiver according to the present invention first detects a temporary timing candidate based on a cross-correlation between a received signal and a known preamble signal and an autocorrelation of the received signal by a timing regenerator. Next, parameter estimation is performed based on the temporary timing candidates. In parameter estimation, the absolute value square value of the difference between the replica of the synthesized signal generated by the inner product of the preamble signal and the estimated value of the channel response and the synthesized signal obtained by linearly synthesizing the received signal with the fractional interval transversal filter is minimized. Thus, the least square method is used. Alternatively, using the preamble signal and the received signal as vectors, parameter estimation is performed using the eigenvalues and eigenvectors of the autocorrelation matrix of the vectors. In the parameter estimation, the weighting coefficient and channel response of the transversal filter are collectively estimated. In the timing regenerator, parameter estimation is performed for each timing candidate, and reception processing is performed at a timing at which the error between the combined signal and the replica is minimized, or a parameter is obtained by a transversal filter with an increased number of filter taps. The estimation is performed, and the minimum number of taps necessary for parameter estimation is determined and the timing error is corrected based on the estimated weighting factor of each tap.

  Next, the received signal is linearly synthesized by a weighting factor obtained by parameter estimation in a fractional interval transversal filter. Then, FFT is performed after removing the guard interval component of the OFDM signal from the combined signal, and the log likelihood ratio of the bit of the signal transmitted from each antenna by the maximum likelihood detector is output in each subcarrier. Also, as signal detection processing, only the desired subcarrier composite signal transmitted from any transmitting antenna is extracted from the composite signal including the guard interval, so that a composite signal replica other than the desired subcarrier is generated and combined. After removing from the signal, turbo equalization processing can be used in which only the signal of the desired subcarrier is subjected to Fourier transform and synthesis by linear processing and converted to a log likelihood ratio. However, since this process requires a log likelihood ratio of all bits, it is used only during repetitive control, and the synthesized signal is processed only by linear processing for the first time.

  The log likelihood ratio of the output bit is MAP decoded after deinterleaving, and the log likelihood ratio of the information bit and the log likelihood ratio of the encoded bit are output. The received bit is determined and the packet in the physical layer is played back. At that time, however, the packet error detection is performed on the received bits by the cyclic code in the repetitive controller, and if there is no error, the idle state is kept until the next packet is received. If there is an error, the log a likelihood ratio of the encoded bits is estimated by using the log likelihood ratio of the encoded bits as feedback information to the maximum likelihood detector. Output the ratio. It is also possible to improve the accuracy by generating an expected value of the subcarrier signal from the logarithmic likelihood ratio of the encoded bits, and performing timing reproduction and parameter estimation again using the expected value. All or part of the series of reception processing is repeated. In addition, the repeat controller controls the number of times the reception process is repeated and stops the repetition.

Note that in the MIMO-OFDM reception system and receiver of the present invention, N _T and N _R are positive integers and include 1, so that when N _T = 1, it is considered a reception system and receiver related to the OFDM system. Needless to say, the technical scope of the present invention includes the OFDM system.

  The above basic operation will be described in more detail using mathematical expressions. First, a MIMO-OFDM signal model will be briefly described, and then the present invention will be described.

FIG. 10 shows the configuration of the assumed MIMO-OFDM transmitter. First, the information bit sequence is divided into _NT sequences and then input to an OFDM modulator to generate an OFDM signal. These OFDM signals are subjected to waveform shaping due to band limitation, and then frequency-converted to an RF frequency band and transmitted from _NT transmission antennas. The OFDM signal at the k-th (1 ≦ k ≦ N _T ) transmission antenna of the first user is defined as follows. The symbol length T _S of the OFDM symbol of N subcarriers is N, where T _{F is} the period of FFT and IFFT, and T _G T _S = T _F + T _G Further, the sampling interval _Δt is _Δt = _TF / N. Therefore, the OFDM signal s _k (m) at the time mΔ _t (iTs ≦ mΔ _t <(i + 1) Ts) of the k-th transmitting antenna is

となる。ここで、ｂ_ｋｎ(ｉ)は第ｋ送信アンテナ、第ｉシンボル、第ｎ(０≦ｎ≦Ｎ−１)サブキャリアにおける変調信号である。また、Δ_ＧはΔ_Ｇ＝Ｔ_Ｇ／Δ_ｔで定義する整数であり、ＯＦＤＭシンボルのシンボル番号ｉはｉ_ｉ≦ｉ≦ｉ_ｆとする。

It becomes. Here, b _kn (i) is a modulated signal in the k-th transmission antenna, the i-th symbol, and the n-th (0 ≦ n ≦ N−1) subcarrier. Also, delta _G is an integer defined in _{Δ G = T G / Δ t} , symbol number i of the OFDM symbol is a _{i i} ≦ i ≦ _{i f.}

はｒ_ｌ(ｔ)を第ｌ要素とするＮ_Ｒ次元受信信号ベクトルとする。また、Ｄ波のマルチパスが到来し、伝搬路の時間変動は無視できるものとする。第ｋ送信アンテナのｄ(１≦ｄ≦Ｄ)パスの複素包絡線をｈ_ｋｄ、その遅延時間と到来角をτ_ｋｄ及びφ_ｋｄとする。このとき、

は Let r _l (t) be the received signal in the baseband of the l th (1 ≦ l ≦ N _R ) receiving antenna;

Is an N _{R-dimensional} received signal vector for the first l element r _l (t). In addition, it is assumed that a D-wave multipath arrives and the time variation of the propagation path can be ignored. Let h _{kd be} the complex envelope of the d (1 ≦ d ≦ D) path of the k-th transmitting antenna, and let τ _kd and φ _kd be its delay time and arrival angle. At this time,

Is

となる。ただし、Ｍ_ｉ＝ｉ_ｉ(Ｎ＋Δ_Ｇ)、Ｍ_ｆ＝ｉ_ｆ(Ｎ＋Δ_Ｇ)−１である。ここで、ｐ_ｓ(ｔ)は送受信機における波形整形用フィルタの合成インパルス応答であり、コサインロールオフフィルタを仮定する。また、

は、到来角φの到来波に対するＮ_Ｒ次元アンテナ応答ベクトルである。

は、他の(Ｋ−１)ユーザ分の信号を合成したＮ_Ｒ次元干渉波ベクトルであり、ＣＣＩ成分を表す。

は各受信アンテナの熱雑音信号を要素に持つＮ_Ｒ次元雑音信号ベクトルである。

It becomes. However, M _i = i _i (N + Δ _G ) and M _f = i _f (N + Δ _G ) −1. Here, p _s (t) is a combined impulse response of the waveform shaping filter in the transceiver, and a cosine roll-off filter is assumed. Also,

Is an _NR- dimensional antenna response vector for an incoming wave with an arrival angle φ.

Is an _NR- dimensional interference wave vector obtained by synthesizing signals for other (K-1) users, and represents a CCI component.

Is N _{R-dimensional} noise signal vector having a thermal noise signal of each receiving antenna element.

  The following description will be made using the received signal model of Equation 2 above.

The linear composition process will be described. In the linear synthesizing process, a whitening filter that generates a synthesized signal by filtering the received signals at each receiving antenna is realized by a fractional interval transversal filter. Here, as shown in FIG. 1, a linear synthesizer, a Fourier transformer, a replica generation unit of signal detection processing, and a metric generation unit are collectively referred to as a branch metric generator, and this receiver generates L branch metrics. It is composed of a vessel. At this time, if the sampling period is Δ = Δ _t / η (η: positive integer), the combined signal y _r (l, m) in the l-th branch metric generator is

となる。ここで、^Ｈは複素共役転置を表し、Ｍは整数で、分数間隔トランスバーサルフィルタのタップ数は(２Ｍ＋１)となる。また、

は

に乗算されるＮ_Ｒ次元のフィルタの重み係数ベクトルであり、

と

は(２Ｍ＋１)Ｎ_Ｒ次元ベクトルとなる。

はパラメータ推定器によって推定される。

It becomes. Here, ^H represents a complex conjugate transpose, M is an integer, and the number of taps of the fractionally spaced transversal filter is (2M + 1). Also,

Is

_NR dimensional filter weight coefficient vector multiplied by

When

Becomes a (2M + 1) N _R- dimensional vector.

Is estimated by a parameter estimator.

The signal detection process will be described. The replica signal y _s (l, m) of y _r (l, m) in the l-th branch metric generator is _expressed as follows by considering the delayed wave component up to the delay time D _{0 Δt} as the desired signal of the first user: It represents as follows.

これをベクトル表示すると

If you display this as a vector

となる。ここで、^＊は複素共役を表す。また、

と

はＮ_Ｔ(Ｄ_０＋１)次元ベクトルであり、チャネル応答及びＯＦＤＭ信号候補に対応する。したがって、フィルタの合成出力からレプリカ信号を減算したα(ｌ,ｍ)は

It becomes. Here, ^* represents a complex conjugate. Also,

When

Is an N _T (D ₀ +1) dimensional vector and corresponds to channel response and OFDM signal candidates. Therefore, α (l, m) obtained by subtracting the replica signal from the combined output of the filter is

と表すことができる。ここで、

と

は(２Ｍ＋１)Ｎ_Ｒ＋Ｎ_Ｔ(Ｄ_０＋１)次元ベクトルである。

It can be expressed as. here,

When

Is a (2M + 1) N _R + N _T (D ₀ +1) dimensional vector.

を生成する。 In this method, D _{0 Δt} is set to be equal to or less than the GI length, and the log likelihood function is calculated as the sum of L branch metrics except α (l, m) corresponding to GI as shown in the following _equation (16).

Is generated.

上記数１６のＹ_ｌ(ｉ,ｎ)は、
ｙ_ｒ(ｌ,ｍ)
の離散フーリエ変換、すなわち、サブキャリア信号であり、また、上記数１６のＨ_ｌ(ｋ_１,ｎ)は第ｋ_１送信アンテナからの伝送路の伝達関数に相当し、下記数１７、数１８でそれぞれ表される。

Y _l (i, n) in the above equation 16 is
y _r (l, m)
Discrete Fourier transform, that is, a subcarrier signal, and H _l (k ₁ , n) in the above equation 16 corresponds to the transfer function of the transmission path from the k _{1 th} transmission antenna. Respectively.

上記数１６の近似は、各シンボル区間でＦＦＴを行い、周波数領域で最尤推定を行うことと等価になる。なお、

に関係する項は

The approximation of Equation 16 is equivalent to performing FFT in each symbol section and performing maximum likelihood estimation in the frequency domain. In addition,

The terms related to

だけであり、この項から{ｂ_ｋｎ(ｉ)|１≦ｋ≦Ｎ_Ｔ}が信号判定できる。即ち、信号検出処理では、上記数１９を最小にする

を選び、これを{ｂ_ｋｎ(ｉ)|１≦ｋ≦Ｎ_Ｔ}の判定値とする最尤推定を行う。全てのサブキャリアについて上記処理を行い、Ｌ個のサブキャリア信号からＮ_Ｔ個の送信信号が抽出される。

From this term, {b _kn (i) | 1 ≦ k ≦ N _T } can be determined as a signal. That is, in the signal detection process, the number 19 is minimized.

Is selected, and maximum likelihood estimation is performed using this as a determination value of {b _kn (i) | 1 ≦ k ≦ N _T }. The above processing is performed for all subcarriers, and _NT transmission signals are extracted from L subcarrier signals.

Actually, the following processing is performed to calculate the log likelihood ratio. Assume that a set of coded bits defining {b _kn (i) | 1 ≦ k ≦ K} is Θ, and a set obtained by removing the coded bits θ _p from Θ is Θ _p . At this time, the log likelihood ratio λ _p of θ _p can be calculated as follows.

この軟判定情報λ_ｐを検波信号として誤り訂正復号処理へ出力する。

The soft decision information λ _p is output as a detection signal to the error correction decoding process.

を帰還して事前情報として用いて最大事後確率推定を以下のように行う。 During iterative control, the log-likelihood ratio of the coded bits that are the output of the error correction decoder

Is used as a priori information to estimate the maximum posterior probability as follows.

同様に、λ_ｐを検波信号として誤り訂正復号処理へ出力する。また、符号化されたビットの対数尤度比

を用いて、変調信号の構成ビットとその出現確率から変調信号の期待値を生成し、それを上記数１９の信号候補

として用いることができ、計算量を削減することができる。

Similarly, output to the error correction decoding processing lambda _p as the detection signal. Also, the log likelihood ratio of the encoded bits

Is used to generate the expected value of the modulation signal from the constituent bits of the modulation signal and the appearance probability thereof, and this is used as the signal candidate of Equation 19 above.

As a result, the amount of calculation can be reduced.

Alternatively, linear processing by MMSE or ZF (Zero Forcing) that separates _NT signals from L subcarrier signals Y _l (i, n) using the transfer function H _l (k ₁ , n), or reliability Depending on whether to perform sequential decoding (V-BLAST) to perform signal separation and interference cancellation in the order of the degree, processing to output a detection signal is performed. The linear processing by MMSE and ZF is described in detail in Non-Patent Document 2 and Non-Patent Document 5, and the sequential decoding is described in Non-Patent Document 15 in detail.

The error correction decoding process will be described. In the error correction decoding process, the log likelihood ratio λ _p of the encoded bits calculated by the signal detector is deinterleaved and used as a priori information to perform MAP decoding. At that time, the error correction decoder calculates a log likelihood ratio between the encoded bit and the information bit by the MAP decoding algorithm. The log-likelihood ratio of information bits is used to determine the received bit sequence, the log-likelihood ratio of coded bits is interleaved, and prior information for generating the expected value of the modulated signal in each subcarrier and estimating the maximum posterior probability As used in repeated control.

  The operation of MAP decoding is described in detail in Non-Patent Document 8, Non-Patent Document 9, Non-Patent Document 10, and Non-Patent Document 11.

として表されるので、この推定アルゴリズムについて述べる。 The parameter estimation process will be described. The parameter estimator of FIG. 1 estimates the channel response necessary to generate the weighting factor and transfer function of the whitening filter. These parameters are given by

This estimation algorithm will be described.

  The filtering synthesis of the received signal is equivalent to the whitening process in space and time.

が得られる。ここで、〈〉はアンサンブル平均を表す。この式から、

は

の固有ベクトルに比例したベクトルとなることがわかる。したがって、

を以下の手順に従い推定する。

Is obtained. Here, <> represents an ensemble average. From this formula:

Is

It can be seen that the vector is proportional to the eigenvector. Therefore,

Is estimated according to the following procedure.

の自己相関行列の

導出：

が送受信間で既知なプリアンブル信号区間において、

の自己相関行列

を下記数２５に従い計算する。 1.

Of the autocorrelation matrix

Derivation:

Is a known preamble signal interval between transmission and reception,

Autocorrelation matrix

Is calculated according to Equation 25 below.

ただし、ｍの和はプリアンブル信号区間に限定する。

However, the sum of m is limited to the preamble signal interval.

の固有展開：

の固有展開を行い、固有値{λ_ｑ}と対応する正規化固有ベクトル

を求める。 2.

Specific expansion of:

Eigenexpansion of, and the normalized eigenvector corresponding to the eigenvalue {λ _q }

Ask for.

であるので、λ_ｑが〈|α(ｌ,ｍ)|^２〉と一致する。これは残留干渉波と雑音信号との平均電力和に相当する。また、信号電力は

の第（２Ｍ＋１）Ｎ_Ｒ＋１要素から第（２Ｍ＋１）Ｎ_Ｒ＋Ｎ_Ｔ(Ｄ_０＋１) 要素の２乗和に比例するので、

とした時のＳＩＮＲは、下記数２６のγ_ｑに比例する。 3. Calculation of SINR:

Therefore, λ _q matches <| α (l, m) | ² >. This corresponds to the average power sum of the residual interference wave and the noise signal. The signal power is

Is proportional to the sum of squares of the (2M + 1) N _R +1 element to the (2M + 1) N _R + N _T (D ₀ +1) element of

The SINR is proportional to γ _q in the following equation ₍ 26).

４．

の算出：
上記数２６の大きい順から固有ベクトル

をＬ個選び、この固有ベクトルに

を乗算したものを

とする。この操作はＳＩＮＲによる重み付けに相当する。

4).

Calculation of:
The eigenvectors in descending order of Equation 26 above

Choose L and use this eigenvector

Multiplied by

And This operation corresponds to weighting by SINR.

  Further, when L = 1, parameter estimation by the least square method is possible. The parameter estimation by the least square method will be described in a sixth embodiment.

  A simplified configuration of the receiver of the present invention in which a branch metric generator including L linear synthesis processes and a Fourier transform process is shared, that is, L = 1 will be described. First, the metric generator of FIG. 1 is reduced to one. The configuration is shown in FIG. At this time, instead of the above equation (19)

を用いて、{ｂ_ｋｎ(ｉ)|１≦ｋ≦Ｎ_Ｔ}の信号判定を行う。このとき、パラメータ推定は固有展開ではなく、上記数１３のα(１,ｍ)の絶対値２乗を最小にするよう逐次的最小２乗法であるＲＬＳやＬＭＳアルゴリズムを適用する（ＲＬＳやＬＭＳアルゴリズムについては、非特許文献１２に詳しく書かれている）。すなわち、下記数２８で定める評価関数Ｊ(ｍ)を最小にするよう

を推定する。

_Is used to perform signal determination of {b _kn (i) | 1 ≦ k ≦ N _T }. At this time, parameter estimation is not an eigenexpansion, but an RLS or LMS algorithm, which is a sequential least square method, is applied so as to minimize the absolute value square of α (1, m) in Equation 13 (RLS or LMS algorithm) Is described in detail in Non-Patent Document 12.) That is, the evaluation function J (m) defined by the following equation 28 is minimized.

Is estimated.

ただし、λ(０＜λ≦１）は忘却係数である。また、推定パラメータが全て零になることを避けるため、一つのパラメータを定数に固定する（詳細は非特許文献１３を参照）。ここでは、ｃ_１(１,０)を１とする。

However, λ (0 <λ ≦ 1) is a forgetting factor. Further, in order to avoid that the estimation parameters are all zero, one parameter is fixed to a constant (refer to Non-Patent Document 13 for details). Here, c ₁ (1,0) is set to 1.

と既知なプリアンブル信号ｓ_ｐ(ｍ)との相互相関処理や受信信号の自己相関処理により仮タイミング候補を検出する。時刻ｍでの受信信号とプリアンブル信号とのスライディング相関値Ｐ_ｍは Timing reproduction processing will be described. In the timing recovery process, the received signal vector

And detecting the temporary timing candidates by the autocorrelation process of the known preamble signal s _p (m) and the cross-correlation processing and reception signal. The sliding correlation value P _m between the received signal and the preamble signal at time m is

となる。ここで、ｉ_ｐはプリアンブル信号のシンボル数である。

It becomes. Here, i _p is the number of symbols of the preamble signal.

とする。各仮タイミング候補

に対してパラメータ推定処理を行い、上記数２８の評価関数Ｊ(ｍ)が最小となるタイミングを最適なタイミングとする。 The temporary timing candidate is set to α (0 <α <1) times the peak value P _max of P _m as a threshold, and a time exceeding the threshold is a temporary timing candidate.

And Each temporary timing candidate

The parameter estimation process is performed on the above, and the timing at which the evaluation function J (m) of Equation 28 is minimized is set as the optimum timing.

_{Alternatively} , the provisional timing m _T is determined from the first time that exceeds the threshold value by taking α times the peak value P _max of P _m as a threshold value. Actually, since the position of the temporary timing is shifted forward by the value of α, the position detected by m ₀ depending on α is shifted backward to be the temporary timing. That is, the temporary timing m _T is

となる。なお、αは上記数３０のｍｉｎ_ｍ[Ｐ_ｍ≧αＰ_ｍａｘ]の分散を最小にする値を選び、ｍ_０はその平均値から決定する。次に、決定した仮タイミングにおいてパラメータ推定用のタップ数を（２Ｍ＋１)Ｎ_Ｒ＋Ｎ_Ｔ(Ｄ_０＋１)
以上に設定し、パラメータ推定処理を行い、各タップの重み係数の大きさにより仮タイミングが正しい位置からシフトした場合に起こるタップの位置のシフト量を検出して、仮タイミングを補正する。具体的には、

または、

や

の重み係数の内で大きさの小さい方から５０％はあまり受信特性には影響を及ぼさないタップとして、その平均値を

とする。タップの大きさが

のβ_１(β_１＞１)倍のタップを抽出し、その最も先頭となるタップが先行波の位置であると決定して、仮タイミングを補正する。また、各タップの重み係数の大きさより必要なタップ数の決定も行う。閾値を

のβ_２(β_１＞β_２＞１)倍とし、最大遅延時間の遅延波の位置も検出して、パラメータ推定の精度を上げるため、最小な有効タップ数を推定する。遅延波の電力は先行波に比べ平均的には小さいため、パラメータβ_２を別に設ける。

It becomes. Here, α is a value that minimizes the variance of min _m [P _m ≧ αP _max ] in Equation 30 above, and m ₀ is determined from the average value. Next, the number of taps for parameter estimation is set to (2M + 1) N _R + N _T (D ₀ +1) at the determined temporary timing.
The parameter estimation process is set as described above, and the shift amount of the tap position that occurs when the temporary timing is shifted from the correct position is detected based on the magnitude of the weight coefficient of each tap to correct the temporary timing. In particular,

Or

And

50% of the weighting factors from the smallest of the weighting factors are taps that do not affect the reception characteristics so much, and the average value is set as a tap.

And The size of the tap

Is extracted, β ₁ (β ₁ > 1) times as many taps are extracted, the leading tap is determined to be the position of the preceding wave, and the temporary timing is corrected. In addition, the number of necessary taps is determined based on the weight coefficient of each tap. Threshold

Β ₂ (β ₁ > β ₂ > 1) times, and the position of the delayed wave having the maximum delay time is also detected, and the minimum number of effective taps is estimated in order to improve the accuracy of parameter estimation. Since the power of the delayed wave is small on average as compared to the preceding wave, it provided the parameters beta ₂ separately.

  Furthermore, processing using the above two types of methods is also conceivable. Parameter estimation processing is performed with redundancy for each temporary timing candidate, and the estimated tap weight is calculated in the temporary timing candidate that minimizes the evaluation function. The temporary timing is corrected and the required number of taps is determined based on the coefficient size.

The Fourier transform process will be described. The Fourier transform process performs Fourier transform by removing the GI component of the synthesized signal at the timing obtained from the timing reproduction process. FIG. 3 shows the state of the composite signal and the FFT interval. Usually, the GI of the first path component in the composite signal in the figure.
The processing is performed in the Fourier transform section 2 in which FFT is performed from the end position.

  When the Fourier transform start position is changed, future and past OFDM signals leak, so that ISI and ICI occur, and the signal power of the desired signal also decreases. At this time, the signal power and the interference power of ISI and ICI can be calculated using the estimated value of the channel response. Details of the calculation method are described in detail in Non-Patent Document 16.

A ratio (P _r ) between the signal power and the interference power is calculated, and a Fourier transform process is performed at a position where P _r is maximized. Actually, since ISI and ICI depend on the modulation signal of each OFDM symbol, ensemble averaging is performed on the modulation signal to estimate the interference power. When the delay amount in the combined signal is within GI as shown in FIG. 3, the symbol head is set to 0, and _Pr is constant from D ₀ −1 to Δ _G which is the head of the D _0th path component. Therefore, two Fourier transforms are performed in the Fourier transform interval 1 and the Fourier transform interval 2, and the phase amount corresponding to the shift of the start position of the Fourier transform interval 1 to the Fourier transform interval 2 is corrected. In-phase synthesis of the signal after Fourier transform. By doing in this way, it becomes possible to perform a Fourier-transform process in the area longer than a normal Fourier-transform area equivalently.

Further, when the delay amount in the combined signal as shown in FIG. 4 is not less than GI similarly by changing the starting position of the Fourier transform to calculate the P _r, performing a Fourier transform at a position where P _r is the maximum. At this time, the transfer function necessary for the signal detection process is also adjusted in accordance with the shift amount of the Fourier transform position. For example, for a subcarrier signal whose signal power is reduced due to the occurrence of ICI, a transfer function that considers the reduction is generated, and a matched filter for the transmission path that takes into account the distortion component of the signal can be realized. To.

  The repeated control process will be described. First, error detection using a cyclic code is performed on the received bit sequence determined by the error correction decoder. If no error is detected, the reception process is terminated, and if an error is detected, the log likelihood ratio of the encoded bits is fed back and the process proceeds to the iterative control process. In the iterative control process, the log likelihood ratio of the encoded bits is input to the maximum likelihood detector and used as prior information for maximum a posteriori probability estimation. Further, using the log likelihood ratio, an expected value of the modulation signal is generated from the constituent bits of the modulation signal and the appearance probability thereof. Then, in addition to the preamble signal, a replica of the OFDM symbol in the data section is generated from the expected value of the modulation signal, and timing reproduction processing and parameter estimation processing are also performed in the data section to improve accuracy. In other words, a part of or all of each process in the MIMO-OFDM reception system of the present invention is repeated to improve the processing accuracy and reduce errors, thereby realizing a highly reliable reception system. The repetition control is performed by error detection using a cyclic code, and no error is detected or processing is performed up to the maximum number of repetitions.

  A method of performing discrete Fourier transform / signal separation linear processing or turbo equalization processing as Fourier transform processing and signal detection processing will be described. In the Fourier transform process, the GI component of the synthesized signal is removed and the Fourier transform is performed, so that the signal power of the GI component is wasted. Further, when a multipath delay wave exceeding GI is present in the synthesized signal, ISI and ICI are generated, so that signal separation cannot be performed satisfactorily. Therefore, a discrete Fourier transform / signal separation linear process or a turbo equalization process is introduced to the combined signal in which the CCI is suppressed, and the signal power of the GI component is combined and the ISI and ICI are removed. Non-Patent Document 6 describes in detail the discrete Fourier transform / signal separation linear processing and turbo equalization processing.

In the discrete Fourier transform / signal separation linear processing, linear equalization of the composite signal, _NT signal separation, and discrete Fourier transform are collectively performed using the estimated value of the channel response. Since these processes are all linear processes, they can be realized by multiplying a vector having a composite signal as a component by a composite weight matrix.

  In the turbo equalization process, a signal component of a desired subcarrier in a specific transmission signal is extracted from the synthesized signal by replica subtraction using the expected value of the modulation signal during repetitive control. The replica is generated from the current OFDM symbol obtained by inverse Fourier transform of the expected value of the modulated signal other than the desired subcarrier, and the past and future OFDM symbols generated by the expected value of the modulated signal, and the estimated channel response. . The extracted signal component is converted into a subcarrier signal by a linear equalization process including a Fourier transform process, and is output as a detection signal. The above processing is performed for all subcarriers of the transmission signal.

  A reception method for controlling whether or not to perform linear synthesis processing will be described. First, parameter estimation processing is performed, the received signal is linearly synthesized using the estimated weighting coefficient, and an error between the output synthesized signal and its replica, that is, the evaluation function J (m) of the above equation 28 is calculated. Next, linear synthesis of the received signal is not performed, a replica is generated for the received signal, parameter estimation processing is performed so that the difference is minimized, and an error between the received signal and the generated replica is calculated. Both errors are compared, and if the error is small when linear synthesis is performed, linear synthesis processing is performed on the received signal, and subsequent processing is performed on the synthesized signal. In the opposite case, the subsequent processing is performed on the received signal without performing linear synthesis processing. By doing in this way, when CCI does not exist, the deterioration of the precision accompanying the increase in the tap in the parameter estimation process which estimates the weighting coefficient of the tap of a linear composition process can be suppressed.

  In order to quantitatively evaluate the effectiveness of the MIMO-OFDM reception system of the present invention, a computer simulation was performed based on a 5 GHz band wireless LAN (IEEE 802.11a) (see Non-Patent Document 14).

  The simulation specifications are shown in Table 1 below. An independent packet is transmitted from each transmission antenna. The packet is composed of a channel estimation preamble signal and data, and each packet has 6 symbols and 30 symbols. A Gold sequence is used for pilots constituting the preamble signal, and different sequences are arranged every two symbols.

マルチパス伝搬路は１７波指数減衰モデルとし、各パスはレイリーフェージングで変動し、最大遅延時間は０.８μｓ、直接波と最大遅延波の電力比をξ(ｄＢ)とする。受信アンテナはアンテナ間隔λ_ｒ／２（λ_ｒ：波長）のリニアアレーとし、異なるユーザ信号の平均到来角度差を６０度、各パスは標準偏差がψのガウス分布とした。同一チャネル干渉を考慮しない従来の最尤受信方式（従来方式）、本発明方式（Ｌ＝２）、及び本発明の簡略化構成（Ｌ＝１）の３種類のシミュレーションを行った。誤り訂正符号は付加しておらず、繰り返し制御も行っていない。タイミング再生は理想的に行われているとし、フーリエ変換処理としては従来通りＧＩを除去してＦＦＴを行った。

The multipath propagation path is a 17-wave exponential attenuation model, each path fluctuates due to Rayleigh fading, the maximum delay time is 0.8 μs, and the power ratio between the direct wave and the maximum delay wave is ξ (dB). The receiving antenna was a linear array with an antenna interval of λ _r / 2 (λ _r : wavelength), the average arrival angle difference of different user signals was 60 degrees, and each path had a Gaussian distribution with a standard deviation of ψ. Three types of simulations were performed: a conventional maximum likelihood reception method (conventional method) that does not consider co-channel interference, the present method (L = 2), and the simplified configuration (L = 1) of the present invention. No error correction code is added and no repetitive control is performed. Timing reproduction is ideally performed, and as a Fourier transform process, GI was removed as before and FFT was performed.

1. Interference Characteristics FIG. 5 shows the average bit error rate (BER) characteristics for co-channel interference, and the horizontal axis in the figure is the signal-to-noise power ratio E _b / N ₀ per bit. When the number of interfering users is 0, the conventional method that can sufficiently obtain path diversity shows the best characteristics. The reason is considered to be the SINR degradation due to insufficient number of taps of the transversal filter in the system of the present invention, and the reduction of the diversity effect by further reducing the metric generator to one in the simplified system. In addition, about the former, the deterioration can be suppressed by using the method of Example 11. FIG. On the other hand, when the number of interfering users is 1, the characteristics of the conventional system are greatly degraded. However, although the system of the present invention and its simplified system show some degradation, sufficient BER characteristics are obtained.

2. CIR Characteristics FIG. 6 shows an average BER characteristic with respect to CIR, and CIR in the figure indicates a signal-to-interference power ratio. The number of receiving antennas is 3 and 4. When the amount of interference is large, the method of the present invention and its simplified method show excellent BER characteristics. However, as the amount of interference decreases, the characteristics of the conventional method are improved and the best BER characteristics are shown. The reason why the characteristic difference between the method of the present invention and the simplified method becomes smaller when the CIR becomes lower is that the degree of freedom of the receiving antenna is used for interference suppression and the diversity effect cannot be obtained. In addition, with the increase in the number of receiving antennas, the method of the present invention improves the characteristics because the antenna diversity effect can be sufficiently obtained.

3. Angular Spread Characteristics FIG. 7 shows the average BER characteristics with respect to the angular spread of the path. When there is no interference, the characteristics of the conventional system and the system of the present invention are improved as the angle spread increases. However, the simplified system does not improve the characteristics even if the angle spread increases. That is, the simplified method does not provide path diversity due to angular spread. In the co-channel interference environment, the characteristics of the method of the present invention and the simplified method deteriorate as the angle spread increases. This is because the number of separable interference waves increases as the angle spread increases. .

4). Timing Offset Characteristic FIG. 8 shows an average BER characteristic with respect to the timing offset when an error occurs in the timing reproduction. In the method of the present invention and its simplified method, the fractional interval transversal filter can suppress the deterioration of the BER characteristic even in the situation where the timing offset occurs. On the other hand, in the conventional method, it can be seen that the characteristics are greatly deteriorated due to the sampling offset.

  Note that the MIMO-OFDM reception system and receiver with high-accuracy timing recovery of the present invention are not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the present invention. Of course.

It is a block diagram which shows the structure of the MIMO-OFDM receiver which can suppress the co-channel interference in this invention. It is a block diagram which shows the structure (in the case of L = 1) of the simplification system of the MIMO-OFDM receiver which can suppress the co-channel interference in this invention. In this invention, it is a schematic diagram which shows the optimal Fourier-transform process in case the delay amount in a synthetic | combination signal is less than GI. In this invention, it is a schematic diagram which shows the optimal Fourier-transform process in case the delay amount in a synthetic | combination signal is more than GI. It is a figure which shows the BER characteristic with respect to co-channel interference. It is a figure which shows the BER characteristic with respect to the amount of interference. It is a figure which shows the BER characteristic with respect to an angular spread. It is a figure which shows the BER characteristic with respect to timing offset. It is a block diagram of the conventional MIMO-OFDM system. It is a block diagram of the conventional MIMO-OFDM transmitter. It is a block diagram of the conventional MIMO-OFDM receiver. 1 is a schematic diagram showing a MIMO-OFDM system in which co-channel interference exists. FIG.

Claims (11)

A timing recovery process for recovering timing from received signals received by N _R (N _R is a positive integer) antennas of N _T (N _T is a positive integer) different error correction encoded OFDM signals; ,
L (L is a positive integer) linear combination processing for whitening the received signal and outputting a combined signal;
L Fourier transform processes for converting the combined signal into a subcarrier signal;
Signal detection processing for performing likelihood calculation using the subcarrier signal and outputting a detection signal;
_NT error correction decoding processes for deinterleaving the detected signal and performing error correction decoding;
A parameter estimation process for collectively estimating parameters used in the linear synthesis process and the signal detection process;
Repetitive control processing in which a part or all of the above processes are repeated;
A MIMO-OFDM reception system with high-accuracy timing recovery characterized by comprising:   As the timing recovery process, a temporary timing candidate is detected based on a cross-correlation between the received signal and a known preamble signal or an autocorrelation of the received signal, the parameter estimation process is performed for each temporary timing candidate, and an estimated weight The temporary timing candidate that minimizes the error between the combined signal output by the linear combining process using a coefficient and its replica is set as the timing, or the number of taps in the most probable candidate among the temporary timing candidates The parameter estimation process is performed by increasing the number of taps necessary for the parameter estimation process and the provisional timing is corrected according to the estimated weighting factor of each tap, or the timing is reproduced, or The MIMO- with high-precision timing recovery according to claim 1, wherein two timing recovery processes are performed. OFDM reception method.   As the linear synthesis process, using the fractional interval transversal filter that filters the received signal and the combiner that synthesizes the output, the received signal with the weight coefficient of the fractional interval transversal filter estimated in the parameter estimation process The MIMO-OFDM reception system with high-accuracy timing recovery according to claim 1, wherein linear synthesis is performed.   As the Fourier transform process, the guard interval component of the synthesized signal is removed by the timing obtained from the timing recovery process and Fourier transform is performed, or the power and generation of a desired signal that changes depending on the start position of the Fourier transform The ratio of the interference power to be calculated is calculated from the estimated value of the channel response obtained from the parameter estimation process, and the Fourier transform is performed at the position where the power ratio becomes the maximum, or the calculated power ratio becomes the maximum The MIMO-OFDM reception system with high-accuracy timing recovery according to claim 1, wherein the combined signal is converted into a subcarrier signal by performing in-phase combining by performing two Fourier transforms having different ranges at positions. As the signal detection processing, a replica signal is generated for the L subcarrier signals using a transfer function derived by Fourier transform of the estimated value of the channel response, and the subcarrier signal and the replica signal Maximum likelihood estimation is performed using the square of the absolute value of the difference as a metric. Further, at the time of iterative control, the logarithmic likelihood ratio of the encoded bit that is the output of the error correction decoding process is fed back and used as prior information. Maximum a posteriori probability estimation or linear processing for separating _NT signals from L subcarrier signals using the transfer function, and sequential decoding for signal separation and interference cancellation in the order of reliability 5. The MIMO-OFDM reception system with high-accuracy timing recovery according to claim 4, wherein processing for outputting a detection signal is performed depending on whether or not to perform.   As the error correction decoding process, the detection signals for the number of subcarriers are subjected to parallel-serial conversion, deinterleaving, soft input / soft output error correction decoding, and log likelihood ratio of information bits and encoded bits. The MIMO-OFDM reception system with high-accuracy timing recovery according to claim 1, wherein a process of determining a received bit sequence based on a log likelihood ratio of the information bits is performed.   As the parameter estimation processing, the weighting coefficient of the fractionally spaced transversal filter and the channel response used for the signal detection processing are collectively calculated, and the absolute difference between the combined signal and the replica of the combined signal generated using the preamble signal 4. The high accuracy according to claim 3, wherein estimation is performed by a least square method so that a square value is minimized, or estimation is performed using an eigenvalue and an eigenvector of an autocorrelation matrix of a vector having the received signal and the preamble signal as elements. A MIMO-OFDM reception scheme with timing recovery.   As the iterative control process, error detection is performed on the received bit sequence using a cyclic code. If no error is detected, the reception process is terminated, and if an error is detected, the received bit sequence is encoded. The high logarithm of claim 6, wherein a part or all of each of the processes is performed again using a log likelihood ratio of bits or an expected value of a subcarrier signal calculated therefrom, and a series of processes is further repeated. MIMO-OFDM reception system with precision timing recovery. As the Fourier transform process and the signal detection process, a discrete Fourier transform that separates _NT signals from the synthesized signal by linear processing that performs signal separation including the Fourier transform process and outputs the detection signal Perform signal separation linear processing or extract the desired subcarrier signal component from the combined signal by replica subtraction during the repetitive control, and combine the extracted signal by linear processing including the Fourier transform processing The MIMO-OFDM reception system with high-accuracy timing recovery according to claim 5, wherein turbo equalization processing for outputting the detection signal is performed.   Performing the parameter estimation process, using the estimated weighting factor, the synthesized signal output by the linear synthesis process and its replica error, the received signal and the received signal without performing the linear synthesis process 5. The parameter estimation process is performed, and an error with a replica of the received signal generated with the estimated value of the channel response is compared. When the latter error is small, the linear synthesis process is not performed. A MIMO-OFDM reception system with high-precision timing recovery as described in 1. A timing regenerator for recovering timing from received signals received by N _R (N _R is a positive integer) number of N _T (N _T is a positive integer) number of different error correction encoded OFDM signals; ,
L (L is a positive integer) linear synthesizers that whiten the received signal and output a synthesized signal;
L Fourier transformers for converting the combined signal into subcarrier signals;
A signal detector that performs likelihood calculation using the subcarrier signal and outputs a detection signal;
_NT error correction decoders for deinterleaving the error detection signal to perform error correction decoding;
A parameter estimator that collectively estimates parameters used in the linear synthesizer and the signal detector;
A repetitive controller that repeats part or all of the above processing;
A MIMO-OFDM receiver with high-precision timing recovery.

Images