JP2007300586A

JP2007300586A - Mimo detection system

Info

Publication number: JP2007300586A
Application number: JP2006152638A
Authority: JP
Inventors: Kazuhiko Fukawa; 和彦府川; Hiroshi Suzuki; 博鈴木; Satoshi Suyama; 聡須山
Original assignee: Tokyo Institute of Technology NUC
Current assignee: Tokyo Institute of Technology NUC
Priority date: 2006-05-01
Filing date: 2006-05-01
Publication date: 2007-11-15
Anticipated expiration: 2026-05-01
Also published as: JP4863459B2

Abstract

<P>PROBLEM TO BE SOLVED: To provide a detection system in which computational complexity can be remarkably reduced while maintaining bit error rate characteristics nearly equal with maximum likelihood detection in MIMO signal detection. <P>SOLUTION: A metric generating circuit 31 corresponding to a metric generating means generates a metric corresponding to a transmission signal candidate on the basis of a reception signal inputted through terminals 29-1 to 29-4, channel information of a transmission line inputted from a terminal 30, and the transmission signal candidate outputted by a signal candidate generating circuit 44. A minimum metric detector 32 corresponding to a minimum metric detecting means inputs the metric and transmission signal candidate to search for a minimum metric, determines a bit of the transmission signal candidate corresponding to the minimum value and outputs the bit to a terminal 33 as a determination value. The transmission signal candidate generating circuit 44 corresponding to a transmission signal candidate generating means adds generation noise generated by a noise generating circuit 38 to the reception signal, performs linear synthesis using a weight coefficient determined from the channel information by a linear conversion circuit 35, then performs hard decision that is a sort of quantization, and generates the transmission signal candidate. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

本発明は，携帯電話システム等の無線通信に関するものであり，特に複数の送受信アンテナを用いて空間多重伝送を行うＭＩＭＯ（ＭｕｌｔｉｐｌｅＩｎｐｕｔＭｕｌｔｉｐｌｅＯｕｔｐｕｔ）方式に関するものである． The present invention relates to wireless communication such as a mobile phone system, and more particularly to a MIMO (Multiple Input Multiple Output) system that performs spatial multiplexing transmission using a plurality of transmission / reception antennas.

携帯電話システム等の無線通信において，周波数帯域を広げずに伝送速度を高める技術として，複数の送受信アンテナを用いて空間多重伝送を行うＭＩＭＯ伝送が知られている．
図１に，送信アンテナ数Ｋ（Ｋは２以上の整数）のＭＩＭＯ伝送用送信機の構成を示す．まず，入力端子１から送信ビット系列がシリアル・パラレル変換器２−１へ入力され，Ｋ個のビット系列に分けられる．ビット系列は，各々対応するベースバンド変調回路３−１から３−Ｋへ端子７−１を通って入力され，送信信号に相当する複素シンボルが変調信号として生成される．この複素シンボルは同相成分と直交成分の２成分を持つが，一つの信号と見なす．ベースバンド変調回路３−１の出力である変調信号は，端子７−２と端子７−３を通り変調器４−１へ入力される．発振器６が出力するキャリア信号が端子７−４から入力し，変調器４−１は，このキャリア信号を用いて変調信号をＲＦ周波数帯へ周波数変換し，端子８を通して対応する送信アンテナ５−１で送信する．他の変調信号についても同様の操作が行われる．In wireless communication such as cellular phone systems, MIMO transmission that performs spatial multiplexing transmission using multiple transmitting and receiving antennas is known as a technique for increasing the transmission speed without expanding the frequency band.
Fig. 1 shows the configuration of a transmitter for MIMO transmission with the number of transmitting antennas K (K is an integer of 2 or more). First, the transmission bit sequence is input from the input terminal 1 to the serial / parallel converter 2-1, and is divided into K bit sequences. Each bit sequence is input from a corresponding baseband modulation circuit 3-1 to 3-K through a terminal 7-1, and a complex symbol corresponding to a transmission signal is generated as a modulation signal. Although this complex symbol has two components, an in-phase component and a quadrature component, it is regarded as one signal. The modulation signal that is the output of the baseband modulation circuit 3-1 is input to the modulator 4-1 through the terminals 7-2 and 7-3. The carrier signal output from the oscillator 6 is input from the terminal 7-4, and the modulator 4-1 converts the frequency of the modulated signal into the RF frequency band using this carrier signal, and transmits the corresponding transmitting antenna 5-1 through the terminal 8. Send with. The same operation is performed for other modulated signals.

図１の変調器４の構成を図２に示す．端子７−２と端子７−３からそれぞれ，変調信号である複素シンボルの同相成分と直交成分が入力され，端子７−４からキャリア信号が入力される．乗算器９−１は変調信号の同相成分にキャリア信号を乗算し，乗算器９−２は変調信号の直交成分に，移相器１０の出力である位相が９０度回転したキャリア信号を乗算する．これらの乗算結果は加算器１１で足し合わされ，増幅器１２で増幅された後，端子８から出力される． The configuration of the modulator 4 in FIG. 1 is shown in FIG. The in-phase component and the quadrature component of the complex symbol, which is a modulation signal, are input from the terminal 7-2 and the terminal 7-3, respectively, and the carrier signal is input from the terminal 7-4. The multiplier 9-1 multiplies the in-phase component of the modulation signal by the carrier signal, and the multiplier 9-2 multiplies the quadrature component of the modulation signal by the carrier signal whose phase that is the output of the phase shifter 10 is rotated by 90 degrees. . These multiplication results are added by the adder 11, amplified by the amplifier 12, and then output from the terminal 8.

また，図１のベースバンド変調回路３について，ＭＩＭＯシングルキャリア伝送用の構成を図３に示す．端子７−１からビット系列が複素シンボル生成回路１３に入力され，ビットに応じて複素シンボルが生成される．複素シンボルの同相成分と直交成分がそれぞれ，端子７−２及び７−３から出力される． The configuration for MIMO single carrier transmission of the baseband modulation circuit 3 in FIG. 1 is shown in FIG. A bit sequence is input from the terminal 7-1 to the complex symbol generation circuit 13, and a complex symbol is generated according to the bit. The in-phase component and quadrature component of the complex symbol are output from terminals 7-2 and 7-3, respectively.

さらに，ＭＩＭＯ−ＯＦＤＭ（ＯｒｔｈｏｇｏｎａｌＦｒｅｑｕｅｎｃｙＤｉｖｉｓｉｏｎＭｕｌｔｉｐｌｅｘｉｎｇ）伝送用のベースバンド変調回路３の構成を，図４に示す．まず，端子７−１からビット系列がシリアル・パラレル変換器２−２へ入力され，サブキャリア数Ｎ（Ｎは２以上の整数）のビット系列に分けられる．Ｎ個のビット系列はそれぞれ，対応する複素シンボル生成回路１３−１から１３−Ｎへ入力され，複素シンボルが生成される．ＩＦＦＴ（ＩｎｖｅｒｓｅＦａｓｔＦｏｕｒｉｅｒＴｒａｎｓｆｏｒｍ）回路はこれらの複素シンボルに対して，各サブキャリアに応じたキャリア信号を乗算して合成し，マルチキャリア信号を生成する．ガードインターバル付加器１５は，このマルチキャリア信号の最後の部分をガードインターバルとして先頭に付加してＯＦＤＭ変調信号を生成し，その同相成分と直交成分をそれぞれ，端子７−２及び７−３へ出力する． Furthermore, FIG. 4 shows the configuration of a baseband modulation circuit 3 for MIMO-OFDM (Orthogonal Frequency Division Multiplexing) transmission. First, a bit sequence is input from the terminal 7-1 to the serial / parallel converter 2-2 and divided into bit sequences of N subcarriers (N is an integer of 2 or more). Each of the N bit sequences is input to a corresponding complex symbol generation circuit 13-1 to 13-N, and a complex symbol is generated. An IFFT (Inverse Fast Fourier Transform) circuit multiplies these complex symbols by a carrier signal corresponding to each subcarrier to generate a multicarrier signal. The guard interval adder 15 generates the OFDM modulation signal by adding the last part of the multicarrier signal as a guard interval to the head, and outputs the in-phase component and the quadrature component to terminals 7-2 and 7-3, respectively. Do it.

次に，受信アンテナ数Ｌ（Ｌは２以上の整数）のＭＩＭＯシングルキャリア伝送用の受信機の構成を図５に示す．まず，受信アンテナ１６−１から１６−Ｌで受信した受信波はそれぞれ，対応する受信回路２０−１から２０−Ｌにおいてベースバンドへ周波数変換され，受信信号として出力される．受信回路２０−１は，増幅器１２，ハイブリッド１７，乗算器９−１と９−２，移相器１０，低域通過フィルタ１８−１と１８−２，及びＡ／Ｄ変換器１９−１及び１９−２から構成され，増幅器１２とハイブリッド１７は受信波を増幅後分岐し，移相器１０は発振器６が出力するキャリア信号の位相を９０度回転させ，乗算回路９−１及び９−２は増幅された受信波にキャリア信号とキャリア信号の位相を９０度回転したものをそれぞれ乗算して，２つの乗算結果を出力する．これらの乗算結果は低域通過フィルタ１８−１及び１８−２で高周波成分が除去された後，ベースバンド信号である受信信号の同相成分と直交成分が抽出される．Ａ／Ｄ変換器１９−１及び１９−２は受信信号をディジタル信号に変換して出力する．信号検出器２１は，受信信号と，伝送路のチャネル情報として伝送路推定回路４３が出力するインパルス応答の推定値を基に，送信信号である複素シンボルを検出し送信ビット系列の判定値を出力端子２２へ出力する．伝送路推定回路４３は，入力端子２３から入力する既知のトレーニング信号と，受信信号を用いて，伝送路のインパルス応答を推定し，この推定値を出力する． Next, Fig. 5 shows the configuration of a receiver for MIMO single carrier transmission with the number of receiving antennas L (L is an integer of 2 or more). First, the received waves received by the receiving antennas 16-1 to 16-L are frequency-converted to the baseband in the corresponding receiving circuits 20-1 to 20-L, and are output as received signals. The receiving circuit 20-1 includes an amplifier 12, a hybrid 17, multipliers 9-1 and 9-2, a phase shifter 10, low-pass filters 18-1 and 18-2, and an A / D converter 19-1. The amplifier 12 and the hybrid 17 branch after amplifying the received wave, and the phase shifter 10 rotates the phase of the carrier signal output from the oscillator 6 by 90 degrees, thereby multiplying circuits 9-1 and 9-2. Multiplies the amplified received signal by the carrier signal and the carrier signal phase rotated by 90 degrees, and outputs two multiplication results. From these multiplication results, high-frequency components are removed by the low-pass filters 18-1 and 18-2, and then the in-phase component and the quadrature component of the received signal which is the baseband signal are extracted. The A / D converters 19-1 and 19-2 convert the received signal into a digital signal and output it. The signal detector 21 detects a complex symbol which is a transmission signal based on the received signal and an estimated value of an impulse response output from the transmission path estimation circuit 43 as channel information of the transmission path, and outputs a transmission bit sequence determination value. Output to terminal 22. The transmission path estimation circuit 43 estimates the impulse response of the transmission path using a known training signal input from the input terminal 23 and the received signal, and outputs this estimated value.

また，図６にＭＩＭＯ−ＯＦＤＭ伝送用受信機の構成を示す．まず，受信アンテナ１６−１から１６−Ｌで受信した受信波はそれぞれ，対応するＯＦＤＭ受信回路２６−１から２６−Ｌにおいてベースバンドへ周波数変換された後，各サブキャリア成分に分離され，受信信号として出力される．ＯＦＤＭ受信回路２６−１は受信回路２０−１，ガードインターバル除去回路２４，及びＦＦＴ（ＦａｓｔＦｏｕｒｉｅｒＴｒａｎｓｆｏｒｍ）演算回路２５から構成され，受信回路２０−１は，発振器６が出力するキャリア信号を用いて受信波をベースバンドへ周波数変換する．ガードインターバル除去回路２４は，周波数変換された信号からガードインターバルに相当する部分を除去し，ＦＦＴ演算回路２５は，ＩＦＦＴの逆操作であるＦＦＴを行い，各サブキャリア成分へ分離し，受信信号として出力する．したがって，サブキャリア数Ｎ個の同相成分と直交成分が生成される．信号検出はサブキャリア毎に行い，第１サブキャリアに相当する受信信号は信号検出器２１−１へ入力され，第Ｎサブキャリアに相当する受信信号は信号検出器２１−Ｎへ入力される．信号検出器２１−１から２１−Ｎは，受信信号と，ＤＦＴ（ＤｉｓｃｒｅｔｅＦｏｕｒｉｅｒＴｒａｎｓｆｏｒｍ）回路２７が出力するチャネルの周波数応答推定値をチャネル情報として用い，送信信号である複素シンボルを検出し送信ビット系列の判定値を出力する．パラレル・シリアル変換回路２８は，信号検出器２１−１から２１−Ｎが出力する送信ビット系列の判定値をパラレル・シリアル変換して出力端子２２へ出力する．ＤＦＴ回路２７は，伝送路推定回路４３が出力する伝送路のインパルス応答の推定値にＤＦＴを行い，各サブキャリア周波数における周波数応答の推定値を出力する． Figure 6 shows the configuration of the receiver for MIMO-OFDM transmission. First, the received waves received by the receiving antennas 16-1 to 16-L are frequency-converted to baseband in the corresponding OFDM receiving circuits 26-1 to 26-L, respectively, and then separated into subcarrier components. It is output as a signal. The OFDM receiving circuit 26-1 includes a receiving circuit 20-1, a guard interval removing circuit 24, and an FFT (Fast Fourier Transform) arithmetic circuit 25. The receiving circuit 20-1 uses a carrier signal output from the oscillator 6. Frequency conversion of received wave to baseband. The guard interval removal circuit 24 removes a portion corresponding to the guard interval from the frequency-converted signal, and the FFT operation circuit 25 performs FFT, which is the inverse operation of IFFT, and separates it into each subcarrier component as a received signal. Output. Therefore, in-phase and quadrature components with N subcarriers are generated. Signal detection is performed for each subcarrier, a received signal corresponding to the first subcarrier is input to the signal detector 21-1, and a received signal corresponding to the Nth subcarrier is input to the signal detector 21-N. The signal detectors 21-1 to 21 -N use the received signal and the channel frequency response estimation value output from the DFT (Discrete Fourier Transform) circuit 27 as channel information, detect complex symbols that are transmission signals, and transmit bits The judgment value of the series is output. The parallel / serial conversion circuit 28 performs parallel / serial conversion on the determination value of the transmission bit series output from the signal detectors 21-1 to 21 -N and outputs the result to the output terminal 22. The DFT circuit 27 performs DFT on the estimated value of the impulse response of the transmission line output from the transmission line estimation circuit 43, and outputs the estimated value of the frequency response at each subcarrier frequency.

図５及び図６の信号検出器２１に適用できる各種検出方式の内，最適方式はビット誤り率を最小にできる最尤検出（非特許文献１参照）である．最尤検出に基づく信号検出器２１の構成を図７に示す．まず，各受信アンテナからの受信信号が端子を介してメトリック生成回路３１へ入力する．具体的には，端子２９−１と２９−２からは受信アンテナ１６−１の受信信号の同相成分及び直交成分が，端子２９−３と２９−４からは受信アンテナ１６−Ｌの受信信号の同相成分及び直交成分が，それぞれ入力する．伝送路のチャネル情報として端子３０から，ＭＩＭＯシングルキャリア伝送の場合には伝送路のインパルス応答の推定値が，ＭＩＭＯ−ＯＦＤＭ伝送の場合には伝送路の周波数応答の推定値が入力する．メトリック生成手段に相当するメトリック生成回路３１は，受信信号，伝送路のチャネル情報，及び信号候補生成回路３４が出力する送信信号候補を基に，送信信号候補に対応するメトリックを生成する．このメトリックについて以下数式を用いて，ＭＩＭＯシングルキャリア伝送を例に説明する．なお，ベースバンドの信号は全て，同相成分を実部，直交成分を虚部とする複素表示で表すものとする． Among various detection methods applicable to the signal detector 21 of FIGS. 5 and 6, the optimum method is maximum likelihood detection (see Non-Patent Document 1) that can minimize the bit error rate. Figure 7 shows the configuration of the signal detector 21 based on maximum likelihood detection. First, the received signal from each receiving antenna is input to the metric generation circuit 31 via the terminal. Specifically, the in-phase and quadrature components of the reception signal of the reception antenna 16-1 are received from the terminals 29-1 and 29-2, and the reception signal of the reception antenna 16-L is received from the terminals 29-3 and 29-4. In-phase and quadrature components are input respectively. As channel information of the transmission path, an estimated value of the impulse response of the transmission path is input from the terminal 30 in the case of MIMO single carrier transmission, and an estimated value of the frequency response of the transmission path is input in the case of MIMO-OFDM transmission. The metric generation circuit 31 corresponding to the metric generation means generates a metric corresponding to the transmission signal candidate based on the received signal, the channel information of the transmission path, and the transmission signal candidate output from the signal candidate generation circuit 34. This metric will be explained below using MIMO single carrier transmission as an example, using the following formula. All baseband signals are expressed in a complex representation with the in-phase component as the real part and the quadrature component as the imaginary part.

ｋは整数）送信アンテナと第ｐ受信アンテナ間のインパルス応答をｈ_ｐｋ，第ｋ送信アンテナからの送信信号，即ち複素シンボルをｓ_ｋとすると，ｙ_ｐは

と表すことができる．なお，ｎ_ｐは第ｐ受信アンテナの受信信号に加わる雑音であり，ｐが

数選択性と仮定した．

となる．

トを求め，判定値として端子３３へ出力する．なお，この最小メトリック検出器３２は最

と第ｐ受信アンテナ間の周波数応答の推定値に置き換えればよい．
このように最尤検出は，数式２のメトリックを送信信号候補の数だけ計算しなければならず，複素シンボルの多値数をＭとするとＭ^Ｋ回計算する必要がある．したがって，ＭやＫが大きい値のとき演算量が膨大になってしまうという問題がある．

k is an integer) If the impulse response between the transmitting antenna and the p-th receiving antenna is h _pk , and the transmission signal from the k-th transmitting antenna, that is, the complex symbol is s _k , y _p is

It can be expressed as. Note that n _p is noise added to the received signal of the p-th receiving antenna, and p is

Assumed number selectivity.

It becomes.

Is output to the terminal 33 as a judgment value. The minimum metric detector 32 is the maximum

And the estimated value of the frequency response between the pth receiving antennas.
Thus maximum likelihood detection has to compute a metric of Equation 2 for the number of transmission signal candidates, certain multi-level number of complex symbols is necessary to calculate M ^K times when the M. Therefore, there is a problem that the amount of calculation becomes enormous when M and K are large values.

以上説明したように，最尤検出はビット誤り率を最小にできるものの，演算量が膨大になるという問題がある．この演算量を削減できる準最適検波方式として，受信信号に線形操作を行う線形受信が知られている．線形受信の一種であるＭＭＳＥ（ＭｉｎｉｍｕｍＭｅａｎＳｑｕａｒｅＥｒｒｏｒ）（非特許文献２参照）に基づく信号検出器２１の構成を図８に示す．まず，各受信アンテナからの受信信号が端子を介して線形変換回路３５へ入力する．具体的には，端子２９−１と２９−２からは受信アンテナ１６−１の受信信号の同相成分及び直交成分が，端子２９−３と２９−４からは受信アンテナ１６−Ｌの受信信号の同相成分及び直交成分が，それぞれ入力する．伝送路のチャネル情報として端子３０から，ＭＩＭＯシングルキャリア伝送の場合には伝送路のインパルス応答の推定値が，ＭＩＭＯ−ＯＦＤＭ伝送の場合には伝送路の周波数応答の推定値が入力する．線形変換回路３５は，チャネル情報から算出する重み付け係数を用いて受信信号に線形操作である線形合成を行い，Ｋ個の合成信号を生成する．硬判定器３６−１から３６−Ｋはそれぞれ，合成信号に量子化の一種である硬判定を行い，合成信号に最も近い複素シンボルを求める．その複素シンボルから送信ビットを推定し，判定値としてパラレル・シリアル変換器２８へ入力する．パラレル・シリアル変換器２８は判定値をパラレル・シリアル変換して端子３３へ出力する． As explained above, maximum likelihood detection can minimize the bit error rate, but has the problem that the amount of computation is enormous. As a sub-optimal detection method that can reduce the amount of computation, linear reception is known which performs linear operation on the received signal. The configuration of the signal detector 21 based on MMSE (Minimum Mean Square Error) (see Non-Patent Document 2), which is a type of linear reception, is shown in FIG. First, the received signal from each receiving antenna is input to the linear conversion circuit 35 via a terminal. Specifically, the in-phase and quadrature components of the reception signal of the reception antenna 16-1 are received from the terminals 29-1 and 29-2, and the reception signal of the reception antenna 16-L is received from the terminals 29-3 and 29-4. In-phase and quadrature components are input respectively. As channel information of the transmission path, an estimated value of the impulse response of the transmission path is input from the terminal 30 in the case of MIMO single carrier transmission, and an estimated value of the frequency response of the transmission path is input in the case of MIMO-OFDM transmission. The linear conversion circuit 35 performs linear synthesis, which is a linear operation, on the received signal using a weighting coefficient calculated from the channel information, and generates K synthesized signals. Each of the hard discriminators 36-1 to 36-K performs a hard decision which is a kind of quantization on the synthesized signal, and obtains a complex symbol closest to the synthesized signal. The transmission bit is estimated from the complex symbol and input to the parallel / serial converter 28 as a judgment value. The parallel / serial converter 28 converts the judgment value from parallel to serial and outputs the result to the terminal 33.

上記の線形変換回路３５の動作について数式を用いて説明する．まず，Ｌ次元受信信号ベクトルｙ，Ｋ次元送信信号ベクトルｓ，Ｌ次元雑音ベクトルｎを以下のように定義する．

を（ｐ，ｋ）成分とするＬ×Ｋ行列とする．合成信号を成分とするＫ次元ベクトルをｘとすると

と表すことができる．ここで，Ｗは重み付け係数を成分とするＫ×Ｌ行列，ＩはＫ×Ｋ単位行列である．ＷはＨから求めることができ，ｘを硬判定したものがｓの推定値となる．
この線形受信は最尤検出に較べ演算量を大幅に削減できるものの，送信信号候補の数を実質１にしていることと等価で，ビット誤り率が著しく劣化するという問題がある．The operation of the linear conversion circuit 35 will be described using mathematical expressions. First, an L-dimensional received signal vector y, a K-dimensional transmitted signal vector s, and an L-dimensional noise vector n are defined as follows.

Is an L × K matrix with (p, k) components. If a K-dimensional vector whose component is a composite signal is x,

It can be expressed as. Here, W is a K × L matrix whose components are weighting coefficients, and I is a K × K unit matrix. W can be obtained from H, and a hard decision of x is an estimate of s.
Although this linear reception can greatly reduce the amount of computation compared to maximum likelihood detection, it is equivalent to making the number of transmission signal candidates substantially 1, and has the problem that the bit error rate significantly deteriorates.

Ｘ．ＺｈｕａｎｄＲ．Ｄ．Ｍｕｒｃｈ，“ＰｅｒｆｏｒｍａｎｃｅａｎａｌｙｓｉｓｏｆｍａｘｉｍｕｍｌｉｋｅｌｉｈｏｏｄｄｅｔｅｃｔｉｏｎｉｎａＭＩＭＯａｎｔｅｎｎａｓｙｓｔｅｍ，” ＩＥＥＥＴｒａｎｓａｃｔｉｏｎｏｎＣｏｍｍｕｎｉｃａｔｉｏｎｓ，ｖｏｌ．５０，ｎｏ．２，ｐｐ．１８７−１９１，Ｆｅｂｒｕａｒｙ２００２．X. Zhu and R.K. D. Murch, “Performance analysis of maximum Likelihood detection in a MIMO antenna system,” IEEE Transactions on Communications, vol. 50, no. 2, pp. 187-191, February 2002. ＳｉｍｏｎＨａｙｋｉｎ，ＡｄａｐｔｉｖｅＦｉｌｔｅｒＴｈｅｏｒｙＴｈｉｒｄＥｄｉｔｉｏｎ，Ｐｒｅｎｔｉｃｅ−Ｈａｌｌ出版，１９９６年．Simon Haykin, Adaptive Filter Theory Third Edition, published by Prentice-Hall, 1996.

このように，ＭＩＭＯ検波方式の内，最尤検出は最適検波方式であり最小ビット誤り率を達成できるが，演算量が膨大になってしまう．この演算量を削減するため，従来から線形受信等の準最適検波方式が提案されているが，演算量を削減するとビット誤り率が大幅に劣化するという問題があった． Thus, of the MIMO detection methods, maximum likelihood detection is an optimal detection method and can achieve the minimum bit error rate, but the amount of computation becomes enormous. In order to reduce this amount of computation, quasi-optimal detection methods such as linear reception have been proposed in the past, but there was a problem that the bit error rate deteriorated significantly when the amount of computation was reduced.

本発明は，このような課題に鑑みてなされたものであり，最尤検出と同程度のビット誤り率特性を維持しつつ，演算量を大幅に削減できる検波方式を提供することを目的とする． The present invention has been made in view of such problems, and an object of the present invention is to provide a detection method that can significantly reduce the amount of computation while maintaining bit error rate characteristics comparable to that of maximum likelihood detection. .

本発明のＭＩＭＯ検波方式によれば，上記目的は前記特許請求の範囲に記載した手段により達成される．即ち，本発明のＭＩＭＯ検波方式は，（ｉ）受信信号，伝送路のチャネル情報，及び複数の送信信号候補から，送信信号候補に対応するメトリックを生成するメトリック生成手段，（ｉｉ）メトリックの最小値を探索し，その最小値に対応する送信信号候補のビットを判定値として出力する最小メトリック検出手段，（ｉｉｉ）受信信号とチャネル情報を基に，線形操作と量子化により複数の送信信号候補を生成する送信信号候補生成手段とから構成される．従来技術と異なる点は，線形操作と量子化により複数の送信信号候補を求め，そのメトリックを生成することにある．
また，本発明のＭＩＭＯ検波方式は，複数の受信アンテナで受信した受信波を周波数変換して得られる信号を受信信号とし，伝送路のインパルス応答の推定値をチャネル情報にできる．
また，本発明のＭＩＭＯ検波方式は，複数の受信アンテナで受信した受信波を周波数変換した後，各サブキャリア成分に分離して得られる信号を受信信号とし，伝送路の周波数応答の推定値をチャネル情報にできる．
さらに，本発明のＭＩＭＯ検波方式の（ｉｉｉ）送信信号生成手段は，受信信号に生成した雑音を加算し，チャネル情報から求める重み付け係数を用いて線形操作を行った後，量子化により複数の送信信号候補を生成する．
加えて，本発明のＭＩＭＯ検波方式の（ｉｉｉ）送信信号生成手段は，受信信号に対してチャネル情報から求める重み付け係数を用いて線形操作を行い，初期値を求め，初期値とチャネル情報と受信信号とを基に更新値をさらに求め，初期値に更新値を加えた後，量子化により複数の送信信号候補を生成する．According to the MIMO detection system of the present invention, the above object is achieved by the means described in the claims. That is, the MIMO detection method of the present invention includes (i) metric generation means for generating a metric corresponding to a transmission signal candidate from the received signal, channel information of the transmission path, and a plurality of transmission signal candidates. A minimum metric detecting means for searching for a value and outputting a bit of a transmission signal candidate corresponding to the minimum value as a determination value; (iii) a plurality of transmission signal candidates by linear operation and quantization based on the reception signal and channel information; Is composed of transmission signal candidate generation means for generating. The difference from the prior art is that multiple transmission signal candidates are obtained by linear operation and quantization, and the metrics are generated.
In addition, the MIMO detection method of the present invention can use a signal obtained by frequency-converting received waves received by a plurality of receiving antennas as a received signal, and use an estimated value of the impulse response of the transmission path as channel information.
Also, the MIMO detection method of the present invention performs frequency conversion on received waves received by a plurality of receiving antennas, and then uses a signal obtained by separating each of the subcarrier components as a received signal, and calculates an estimated value of the frequency response of the transmission path. Can be channel information.
Further, the transmission signal generation means (iii) of the MIMO detection system of the present invention adds the generated noise to the received signal, performs a linear operation using a weighting coefficient obtained from the channel information, and then performs a plurality of transmissions by quantization. Generate signal candidates.
In addition, the (iii) transmission signal generation means of the MIMO detection system of the present invention performs a linear operation using a weighting coefficient obtained from the channel information on the received signal, obtains an initial value, and obtains an initial value, channel information, and reception The update value is further obtained based on the signal, and after adding the update value to the initial value, multiple transmission signal candidates are generated by quantization.

本発明は，以下に記載されるような効果を奏する．
請求項１記載の発明のＭＩＭＯ検波方式によれば，最尤検出と同程度のビット誤り率特性を維持しつつ，演算量を大幅に削減できる．
請求項２記載の発明のＭＩＭＯ検波方式によれば，ＭＩＭＯシングルキャリア伝送に適用できる．
請求項３記載の発明のＭＩＭＯ検波方式によれば，ＭＩＭＯ−ＯＦＤＭ伝送に適用できる．
請求項４記載の発明のＭＩＭＯ検波方式によれば，送信信号候補の数を抑え演算量を削減できる．
請求項５記載の発明のＭＩＭＯ検波方式によれば，送信信号候補の数をさらに抑え演算量を大幅に削減できる．The present invention has the following effects.
According to the MIMO detection system of the first aspect of the present invention, the amount of calculation can be greatly reduced while maintaining the bit error rate performance comparable to that of maximum likelihood detection.
The MIMO detection system according to the second aspect of the present invention can be applied to MIMO single carrier transmission.
The MIMO detection system according to the third aspect of the invention can be applied to MIMO-OFDM transmission.
According to the MIMO detection system of the fourth aspect of the invention, the number of transmission signal candidates can be reduced and the amount of calculation can be reduced.
According to the MIMO detection system of the fifth aspect of the invention, the number of transmission signal candidates can be further reduced and the amount of calculation can be greatly reduced.

以下，本発明を実施するための最良の形態について説明する．
まず，第１の発明のＭＩＭＯ検波方式を用いた信号検出器２１の構成（請求項４）を図９に示す．図７に示す従来の最尤検出に基づく信号検出器２１との違いは，信号候補生成回路３４を送信信号候補生成手段に相当する送信信号候補生成回路４４に置き換えた点にある．なお，送信信号候補生成回路４４は，加算回路３７−１から３７−Ｌ，雑音生成回路３８，線形変換回路３５，量子化器３９−１から３９−Ｋ，及びパラレル・シリアル変換器４０から構成され，以下では送信信号候補生成回路４４の動作について詳述する．The best mode for carrying out the present invention will be described below.
First, FIG. 9 shows the configuration of the signal detector 21 using the MIMO detection system of the first invention (claim 4). The difference from the signal detector 21 based on the conventional maximum likelihood detection shown in FIG. 7 is that the signal candidate generation circuit 34 is replaced with a transmission signal candidate generation circuit 44 corresponding to transmission signal candidate generation means. The transmission signal candidate generation circuit 44 includes adder circuits 37-1 to 37-L, a noise generation circuit 38, a linear conversion circuit 35, quantizers 39-1 to 39-K, and a parallel / serial converter 40. The operation of the transmission signal candidate generation circuit 44 will be described in detail below.

送信信号候補数をＣ（Ｃは２以上の整数）とする．雑音生成回路３８はＣ回人工的に複素ガウス信号を生成するが，１回目は値が零となる信号をＬ個生成し，２回目以降は平均値零，分散ξで互いに無相関となる複素ガウスをＬ個発生させる．加算回路３７−１から３７−Ｌは，受信信号に雑音生成回路３８が生成する生成雑音を加算する．具体的には，端子２９−１と２９−２から入力する受信アンテナ１６−１の受信信号の同相成分及び直交成分については，それぞれ１番目の生成雑音の同相成分及び直交成分を加算し，端子２９−３と２９−４から入力する受信アンテナ１６−Ｌの受信信号の同相成分及び直交成分については，それぞれＬ番目の生成雑音の同相成分及び直交成分を加算する．加算回路３

信号を生成する．この合成信号を成分とするＫ次元ベクテルをｘとすると

るＫ×Ｌ行列であり，１回目は生成雑音が加わっていないことと等価なので数式７のＷと等しくなるが，２回目以降は分散ξの生成雑音が新たに加わっているのでＷと異なる．なお，ＭＭＳＥではなくＺＦ（ＺｅｒｏＦｏｒｃｉｎｇ）による線形合成を行うこともでき

とすることもできる．量子化器３９−１から３９−Ｋはそれぞれ，上記の合成信号に量子化の一種である硬判定を行い，合成信号に最も近い複素シンボルを求め，パラレル・シリアル変換器４０へ入力する．パラレル・シリアル変換器４０はこの複素シンボルをパラレル・シリアル変換し，送信信号候補として出力する．
このように，従来の最尤検出では全ての送信信号候補に対してメトリックを計算する必要があったが，本発明では送信信号候補数をＣに削減でき，演算量を大幅に削減できる．また，従来のＭＭＳＥでは送信信号候補の数を実質１としているのに対し，本発明はＭＭＳＥによる送信信号候補を含むＣ個の送信信号候補に対してメトリックを計算し，最小メトリックに相当する送信信号候補を選択するので，ＭＭＳＥよりもビット誤り率が改善できる．
なお，繰り返し処理を行う場合，受信信号の代わりに，最小メトリックに対応する送信信号候補もしくはそのｘを使うこともでき，さらなるビット誤り率改善が期待できる．Let C be the number of transmission signal candidates (C is an integer of 2 or more). The noise generation circuit 38 artificially generates a complex Gaussian signal C times. The first generation generates L signals whose values are zero, and the second and subsequent times are complex values that are uncorrelated with each other with an average value of zero and a variance ξ. Generate L Gaussians. The adding circuits 37-1 to 37-L add the generated noise generated by the noise generating circuit 38 to the received signal. Specifically, for the in-phase component and the quadrature component of the received signal of the receiving antenna 16-1 input from the terminals 29-1 and 29-2, the in-phase component and the quadrature component of the first generated noise are added, respectively. For the in-phase component and the quadrature component of the received signal of the receiving antenna 16-L input from 29-3 and 29-4, the in-phase component and the quadrature component of the Lth generated noise are added, respectively. Adder circuit 3

Generate a signal. If the K-dimensional vector whose component is this synthesized signal is x,

This is equivalent to the fact that the generated noise is not added at the first time, and is equal to W in Equation 7. However, the second time and later are different from W because the generated noise of variance ξ is newly added. It is also possible to perform linear synthesis using ZF (Zero Forcing) instead of MMSE.

Can also be used. Each of the quantizers 39-1 to 39-K performs a hard decision which is a kind of quantization on the above synthesized signal, obtains a complex symbol closest to the synthesized signal, and inputs it to the parallel-serial converter 40. The parallel / serial converter 40 performs parallel / serial conversion on the complex symbol and outputs it as a transmission signal candidate.
As described above, in the conventional maximum likelihood detection, it is necessary to calculate metrics for all transmission signal candidates. However, in the present invention, the number of transmission signal candidates can be reduced to C, and the amount of calculation can be greatly reduced. Also, in the conventional MMSE, the number of transmission signal candidates is substantially 1, whereas in the present invention, metrics are calculated for C transmission signal candidates including transmission signal candidates by MMSE, and transmission corresponding to the minimum metric is performed. Since the signal candidate is selected, the bit error rate can be improved compared to MMSE.
When iterative processing is performed, a transmission signal candidate corresponding to the minimum metric or its x can be used instead of the received signal, and further improvement of the bit error rate can be expected.

次に，第２の発明のＭＩＭＯ検波方式を用いた信号検出器２１の構成（請求項５）を図１０に示す．図７に示す従来の最尤検出に基づく信号検出器２１との違いは，信号候補生成回路３４を送信信号候補生成手段に相当する送信信号候補生成回路４５に置き換えた点にある．なお，送信信号候補生成回路４５は，線形変換回路３５，量子化器４１−１から４１−Ｋ，更新値演算回路４２，加算回路３７−１から３７−Ｋ，量子化器３９−１から３９−Ｋ，及びパラレル・シリアル変換器４０から構成され，以下では送信信号候補生成回路４５の動作について詳述する． Next, FIG. 10 shows the configuration of the signal detector 21 using the MIMO detection system of the second invention (Claim 5). The difference from the signal detector 21 based on the conventional maximum likelihood detection shown in FIG. 7 is that the signal candidate generation circuit 34 is replaced with a transmission signal candidate generation circuit 45 corresponding to transmission signal candidate generation means. The transmission signal candidate generation circuit 45 includes a linear conversion circuit 35, quantizers 41-1 to 41-K, an update value calculation circuit 42, adder circuits 37-1 to 37-K, and quantizers 39-1 to 39. The operation of the transmission signal candidate generation circuit 45 will be described in detail below.

まず，各受信アンテナからの受信信号が端子を介して線形変換回路３５へ入力する．具体的には，端子２９−１と２９−２からは受信アンテナ１６−１の受信信号の同相成分及び直交成分が，端子２９−３と２９−４からは受信アンテナ１６−Ｌの受信信号の同相成分及び直交成分が，それぞれ入力する．線形変換回路３５は，端子３０から入力するチャネル情報から重み付け係数を算出し，この重み付け係数を用いて受信信号に線形操作である線形合成を行い，Ｋ個の合成信号を生成する．この合成信号が初期値ｘに相当する．量子化器４１−１から４１−Ｋはそれぞれ，合成信号に量子化の一種である硬判定を行い，合成信号に最も近い複素シンボルを求め更新値演算回路４２に入力する．更新値演算回路４２は，初期値を硬判定した複素シンボル，受信信号，並びにチャネル情報から更新値を求める．加算回路３７−１から３７−Ｋは初期値に更新値を加算し，量子化器３９−１から３９−Ｋはこの加算結果に対して量子化の一種である硬判定を行い，加算結果に最も近い複素シンボルを求めパラレル・シリアル変換器４０へ入力する．パラレル・シリアル変換器４０はこの複素シンボルをパラレル・シリアル変換し，送信信号候補として出力する． First, the received signal from each receiving antenna is input to the linear conversion circuit 35 via a terminal. Specifically, the in-phase and quadrature components of the reception signal of the reception antenna 16-1 are received from the terminals 29-1 and 29-2, and the reception signal of the reception antenna 16-L is received from the terminals 29-3 and 29-4. In-phase and quadrature components are input respectively. The linear conversion circuit 35 calculates a weighting coefficient from the channel information input from the terminal 30 and performs linear composition, which is a linear operation, on the received signal using the weighting coefficient to generate K composite signals. This synthesized signal corresponds to the initial value x. Each of the quantizers 41-1 to 41-K performs a hard decision which is a kind of quantization on the synthesized signal, obtains a complex symbol closest to the synthesized signal, and inputs the complex symbol to the update value calculation circuit. The update value calculation circuit 42 obtains an update value from the complex symbol whose initial value is hard-decided, the received signal, and the channel information. The adder circuits 37-1 to 37-K add the updated value to the initial value, and the quantizers 39-1 to 39-K make a hard decision as a kind of quantization for the addition result, Find the nearest complex symbol and input it to the parallel-serial converter 40. The parallel / serial converter 40 performs parallel / serial conversion on the complex symbol and outputs it as a transmission signal candidate.

上記の更新値演算回路４２の動作について以下，数式を用いて詳述する．まず，合成信号を硬判定して得られる複素シンボル，即ち量子化器４１−１から４１−Ｋの出力を成分

ある．また，Ｐは数式１３で定義するＫ×Ｋ行列である．
μ_ｒはｒ＝０では０とする．即ち，ｒ＝０の送信信号候補はｘを単に硬判定したものとなり，ＭＭＳＥと同じになる．ｒが１以上では

このように，μ_ｒはＫ（Ｍ−１）＋１個値が存在する．送信信号候補は，ｘにｕを加算した後，硬判定したものであるから，μ_ｒの数，即ちＫ（Ｍ−１）＋１個存在する．したがって，従来の最尤検出では全ての送信信号候補に対してメトリックを計算する必要があったが，本発明では送信信号候補数をＫ（Ｍ−１）＋１に削減でき，演算量を大幅に削減できる．また，従来のＭＭＳＥでは送信信号候補の数を実質１としているのに対し，本発明はＭＭＳＥによる送信信号候補を含むＫ（Ｍ−１）＋１個の送信信号候補に対してメトリックを計算し，最小メトリックに相当する送信信号候補を選択するので，ＭＭＳＥよりもビット誤り率が改善できる．

く線形操作を行ってもよい．また，初期値としてｘの代わりにｘの硬判定値を用いることも可能である．さらに，数式１２のｕの代わりに

としてもよい．ただし，ｅは零ベクトルとならない任意のＬ次元ベクトルである．The operation of the update value calculation circuit 42 will be described in detail below using mathematical expressions. First, a complex symbol obtained by hard-decision of the synthesized signal, that is, the output of the quantizers 41-1 to 41-K

is there. P is a K × K matrix defined by Equation 13.
μ _r is set to 0 when r = 0. That is, the transmission signal candidate of r = 0 is a hard decision of x, which is the same as MMSE. When r is 1 or more

Thus, mu _r is K (M-1) +1 or value exists. Transmitted signal candidates, after adding u to x, since it is obtained by hard decision, the number of mu _r, i.e. K (M-1) +1 or present. Therefore, in the conventional maximum likelihood detection, it is necessary to calculate a metric for all transmission signal candidates. However, in the present invention, the number of transmission signal candidates can be reduced to K (M−1) +1, which greatly increases the amount of calculation. Can be reduced. Further, in the conventional MMSE, the number of transmission signal candidates is substantially 1, whereas the present invention calculates a metric for K (M−1) +1 transmission signal candidates including transmission signal candidates by MMSE, Since the transmission signal candidate corresponding to the minimum metric is selected, the bit error rate can be improved compared to MMSE.

It is also possible to perform linear operations. It is also possible to use a hard decision value of x instead of x as an initial value. Furthermore, instead of u in Equation 12

It is good. However, e is an arbitrary L-dimensional vector that is not a zero vector.

最後に，本発明は上述の発明を実施するための最良の形態に限らず本発明の要旨を逸脱することなくその他種々の構成を採り得ることはもちろんである． Finally, the present invention is not limited to the best mode for carrying out the invention described above, and various other configurations can be adopted without departing from the gist of the present invention.

従来のＭＩＭＯ無線送信機のブロック構成図である．It is a block diagram of a conventional MIMO radio transmitter. 図１の変調器のブロック構成図である．It is a block block diagram of the modulator of FIG. ＭＩＭＯシングルキャリア伝送における，図１のベースバンド変調回路のブロック構成図である．It is a block block diagram of the baseband modulation circuit of FIG. 1 in MIMO single carrier transmission. ＭＩＭＯ−ＯＦＤＭ伝送における，図１のベースバンド変調回路のブロック構成図である．It is a block block diagram of the baseband modulation circuit of FIG. 1 in MIMO-OFDM transmission. ＭＩＭＯシングルキャリア伝送における，従来の無線受信機のブロック構成図である．It is a block diagram of a conventional radio receiver in MIMO single carrier transmission. ＭＩＭＯ−ＯＦＤＭ伝送における，従来の無線受信機のブロック構成図である．It is a block diagram of a conventional radio receiver in MIMO-OFDM transmission. 図５と図６の信号検出器で，従来の最尤検出を用いたブロック構成図である．It is a block block diagram using the conventional maximum likelihood detection in the signal detector of FIG. 5 and FIG. 図５と図６の信号検出器で，従来のＭＭＳＥを用いたブロック構成図である．FIG. 7 is a block diagram of the signal detector shown in FIGS. 5 and 6 using a conventional MMSE. 図５と図６の信号検出器で，第１の発明のブロック構成図である．5 is a block configuration diagram of the first invention in the signal detectors of FIGS. 5 and 6. FIG. 図５と図６の信号検出器で，第２の発明のブロック構成図である．5 is a block diagram of the second invention in the signal detectors of FIGS. 5 and 6. FIG.

符号の説明Explanation of symbols

１入力端子，２シリアル・パラレル変換器，３ベースバンド変調回路，４変調器，５送信アンテナ，６発振器，７端子，８端子，９乗算器，１０移相器，１１加算器，１２増幅器，１３複素シンボル生成回路，１４ＩＦＦＴ演算回路，１５ガードインターバル付加器，１６受信アンテナ，１７ハイブリッド，１８低域通過フィルタ，１９Ａ／Ｄ変換器，２０受信回路，２１信号検出器，２２出力端子，２３入力端子，２４ガードインターバル除去回路，２５ＦＦＴ演算回路，２６ＯＦＤＭ受信回路，２７ＤＦＴ回路，２８パラレル・シリアル変換器，２９端子，３０端子，３１メトリック生成回路，３２最小メトリック検出器，３３端子，３４信号候補生成回路，３５線形変換回路，３６硬判定器，３７加算回路，３８雑音生成回路，３９量子化器，４０パラレル・シリアル変換器，４２更新値演算回路，４３伝送路推定回路，４４送信信号候補生成回路，４５送信信号候補生成回路1 input terminal, 2 serial / parallel converter, 3 baseband modulation circuit, 4 modulator, 5 transmitting antenna, 6 oscillator, 7 terminal, 8 terminal, 9 multiplier, 10 phase shifter, 11 adder, 12 amplifier, 13 complex symbol generation circuit, 14 IFFT arithmetic circuit, 15 guard interval adder, 16 receiving antenna, 17 hybrid, 18 low-pass filter, 19 A / D converter, 20 receiving circuit, 21 signal detector, 22 output terminals, 23 inputs Terminal, 24 guard interval removal circuit, 25 FFT operation circuit, 26 OFDM reception circuit, 27 DFT circuit, 28 parallel-serial converter, 29 terminal, 30 terminal, 31 metric generation circuit, 32 minimum metric detector, 33 terminal, 34 signal candidate generation Circuit, 35 linear conversion circuit, 36 hard discriminator, 37 addition circuit, 38 noise generation circuit, 39 quantity Encoder, 40 a parallel-serial converter, 42 update value calculation circuit, 43 channel estimator 44 transmits the signal candidate generator circuit, 45 the transmission signal candidate generator circuit

Claims

受信信号と，伝送路のチャネル情報と，複数の送信信号候補とを基に，前記送信信号候補に対応するメトリックを生成するメトリック生成手段と，
前記メトリックの最小値を探索し，その最小値に対応する前記送信信号候補のビットを判定値として出力する最小メトリック検出手段と，
前記受信信号と前記チャネル情報を基に，線形操作と量子化により複数の前記送信信号候補を生成する送信信号候補生成手段とから構成されることを特徴とするＭＩＭＯ検波方式．Metric generation means for generating a metric corresponding to the transmission signal candidate based on the received signal, channel information of the transmission path, and a plurality of transmission signal candidates;
A minimum metric detecting means for searching for a minimum value of the metric and outputting a bit of the transmission signal candidate corresponding to the minimum value as a determination value;
A MIMO detection system comprising: transmission signal candidate generation means for generating a plurality of transmission signal candidates by linear operation and quantization based on the received signal and the channel information.

請求項１の前記受信信号は，複数の受信アンテナで受信した受信波を周波数変換して得られる信号とし，請求項１の前記チャネル情報は，伝送路のインパルス応答の推定値とすることを特徴とするＭＩＭＯ検波方式． The received signal of claim 1 is a signal obtained by frequency-converting received waves received by a plurality of receiving antennas, and the channel information of claim 1 is an estimated value of an impulse response of a transmission path. MIMO detection method.

請求項１の前記受信信号は，複数の受信アンテナで受信した受信波を周波数変換した後，各サブキャリア成分に分離して得られる信号とし，請求項１の前記チャネル情報は，伝送路の周波数応答の推定値とすることを特徴とするＭＩＭＯ検波方式． The received signal of claim 1 is a signal obtained by frequency-converting received waves received by a plurality of receiving antennas and then separating the signals into subcarrier components, and the channel information of claim 1 is a frequency of a transmission path. A MIMO detection system characterized by an estimated response value.

請求項１の前記送信信号生成手段は，前記受信信号に生成した雑音を加算し，前記チャネル情報から求める重み付け係数を用いて線形操作を行った後，量子化により複数の前記送信信号候補を生成することを特徴とするＭＩＭＯ検波方式． The transmission signal generation means according to claim 1 adds the generated noise to the reception signal, performs a linear operation using a weighting coefficient obtained from the channel information, and generates a plurality of transmission signal candidates by quantization. MIMO detection system characterized by

請求項１の前記送信信号生成手段は，前記受信信号に対して前記チャネル情報から求める重み付け係数を用いて線形操作を行い，初期値を求め，前記初期値と前記チャネル情報と前記受信信号とを基に更新値をさらに求め，前記初期値に前記更新値を加えた後，量子化により複数の前記送信信号候補を生成することを特徴とするＭＩＭＯ検波方式． The transmission signal generating means according to claim 1 performs a linear operation using a weighting coefficient obtained from the channel information on the received signal, obtains an initial value, and obtains the initial value, the channel information, and the received signal. A MIMO detection method characterized by further obtaining an update value based on the initial value, adding the update value to the initial value, and then generating a plurality of transmission signal candidates by quantization.