JP5211194B2 - MIMO fading simulator - Google Patents

Description

  The present invention relates to a MIMO fading simulator.

  In wireless communication, MIMO (multi-input multi-output), which performs signal transmission using a plurality of transmission / reception antennas, has attracted attention. MIMO performs spatial multiplexing of signals by transmitting uncorrelated signals between a plurality of transmitting antennas, and can achieve a high transmission rate by separating them at a receiver using a plurality of receiving antennas. Also, highly reliable transmission can be realized by transmitting a signal having a correlation between transmitting antennas by space-time coding and improving diversity gain.

  In order to evaluate various transmission schemes in MIMO, a channel model that models the radio propagation environment is required. In standardization of wireless LAN based on IEEE802.11n, examination of MIMO is underway, and a channel model for evaluating a transmission method has already been proposed in Non-Patent Document 1.

  As a method for evaluating the MIMO transmission method, first, a computer simulation is performed based on a channel model. Next, we will make a prototype of a transmitter and a receiver and actually evaluate them in an indoor transmission experiment system using RF or IF signals. At that time, a fading simulator simulates an actual propagation path. That is, the fading simulator realizes a channel model with RF or IF analog input / output. A fading simulator for MIMO can also be created based on the channel model described in Non-Patent Document 1, but there are some problems.

  First, in the MIMO fading simulator, there are multiple analog input circuits (down converters) and analog output circuits (up converters) along with MIMO. Therefore, the channel model as set by the variation in circuit characteristics and distortion by the circuit is used. It cannot be realized. Furthermore, since the distortion is increased by the MIMO fading simulator, there is a problem that the output SNR is lowered. Non-Patent Document 1 does not specify these solutions.

  In Non-Patent Document 1, only omni antennas are assumed. However, since various antennas can be actually used, the MIMO fading simulator based on Non-Patent Document 1 cannot handle these antennas. is there.

  A high-precision MIMO fading simulator that solves these problems has not been proposed so far.

V. Erceg et al., “TGn Channel Models” IEEE 802.11-03 / 940r4, May 2004.

The conventional MIMO fading simulator has the following problems.
1. It is impossible to realize a channel model as set due to variations in circuit characteristics in a plurality of analog input / output circuits and distortion caused by circuits.
2. Since the antenna directivity and the coupling coefficient between elements cannot be taken into consideration, the MIMO channel when using an antenna other than the omni antenna cannot be realized with high accuracy.
In consideration of the above points, there is no MIMO fading simulator that can compensate for variations in circuit characteristics and circuit distortion in an analog input / output circuit, and that can generate channel coefficients in consideration of antenna directivity and inter-element coupling coefficients.

  The present invention has been made in view of such problems, and can calibrate variations in circuit characteristics and distortion caused by circuits in a plurality of RF or IF analog input / output circuits, that is, down converters and up converters. An object of the present invention is to provide a highly accurate MIMO fading simulator capable of deriving a correlation coefficient of a MIMO channel using antenna directivity and an inter-element coupling coefficient and generating a channel coefficient using the MIMO channel correlation coefficient.

According to the present invention, a MIMO fading simulator having M input terminals and N output terminals and simulating transmission path characteristics between M transmission antennas and N reception antennas is provided as follows:
M downconverters that convert input signals from the M input terminals into complex baseband input signals, and M is a positive integer.
M input calibration circuits for compensating the distortion generated by the down converter for the complex baseband input signal and outputting as a post-calibration input signal;
The channel formation formed by the calibrated input signals from the M input calibration circuits and the MN channels between the M inputs and the N outputs, which are respectively called J-tap finite impulse response filters, hereinafter referred to as FIR filters. Circuit, N and J are positive integers,
A channel coefficient generator that gives the MN FIR filters J channel coefficients that simulate the characteristics of each channel;
N output calibration circuits which are provided with N outputs of the channel forming circuit and which respectively perform distortion compensation and output as post-calibration output signals which are complex baseband signals;
N up-converters that up-convert the output signals after calibration of the N output calibration circuits and output the signals to the N output terminals, respectively.
And the channel coefficient generator includes the J channels based on antenna directivity and inter-element coupling coefficients of the M transmitting antennas and the N receiving antennas. Coefficients are generated and set for the MN FIR filters, and the N output calibration circuits compensate for the distortion generated by the N up-converters for the N outputs of the channel forming circuit, respectively. To generate the output signal after calibration,
MNJ complex Gaussian random number generators that generate complex Gaussian random numbers;
MNJ Doppler filters that convolve weighting coefficients so as to be a Doppler spectrum set for the complex Gaussian random number,
An antenna / spatial correlation coefficient generator that generates an antenna / spatial correlation coefficient using the antenna directivity and the inter-element coupling coefficient;
J antenna / spatial correlation coefficient multipliers that generate a correlation channel coefficient by multiplying the Doppler channel coefficient that is the output of the Doppler filter by the antenna / spatial correlation coefficient that is the output of the antenna / spatial correlation coefficient;
Furthermore, look at including a door,
The channel coefficient is obtained from the generated correlation channel coefficient.
It is configured as follows.

  According to the MIMO fading simulator of the present invention, by using an input calibration circuit or an output calibration circuit, it is possible to compensate for variations in circuit characteristics among M down-converters or N up-converters, and antenna directivity. By deriving the correlation coefficient of the MIMO channel based on, and generating the channel coefficient using it, a fading transmission path close to the real environment can be simulated.

The block diagram which shows the function structure of the MIMO fading simulator by this invention. Block diagram illustrating a functional configuration of an input calibration circuit 4 _m in FIG. The block diagram which shows the function structure of the output calibration circuit 10n in FIG. The block diagram which shows the function structure of the channel coefficient generator 7 in FIG. The block diagram which shows the other structural example of the MIMO fading simulator by this invention.

Hereinafter, will be described with reference to the drawings forms condition for carrying out the present invention.

First, a description will be given shape state for carrying out the invention according to the MIMO fading simulator. The configuration of the MIMO fading simulator is shown in FIG. The MIMO fading simulator of the present invention simulates an M × N channel transmission path formed between M antennas (not shown) on the transmission side and N antennas (not shown) on the reception side. To do. M and N are integers of 1 or more. Accordingly, M input terminals 1 ₁ to 1 _M correspond to M transmitting antennas on the transmitting side, and output terminals 15 ₁ to 15 _N correspond to N receiving antennas on the receiving side.
As shown in FIG. 1, the MIMO fading simulator according to the present invention includes M down-converters 2 ₁ to 2 _M connected to _M input terminals 1 ₁ to _{1 M} , and M analog-to-digital conversions. (Hereinafter referred to as ADC) 3 _{1 to} 3 _M , M input calibration circuits 4 ₁ to 4 _M , channel forming circuit 5, channel coefficient generator 7, and N adders 8 ₁ to 8 _N , N noise generators 9 _{1 to} 9 _N , N output calibration circuits 10 ₁ to 10 _N , N digital-analog converters (hereinafter referred to as DACs) 11 _{1 to} 11 _N , , N up-converters 12 _{1 to} 12 _N and N variable attenuators 13 _{1 to} 13 _N connected to _N output terminals 15 ₁ to 15 _N.
The channel forming circuit 5 is a circuit for forming MN channels between M inputs and N outputs, and N channels provided corresponding to each input terminal 1 _m (m = 1, 2,..., M). And N FIR filters 5 _{m1 to} 5 _mN for defining the transmission characteristics of the M, and an M input adder 6 _n provided corresponding to each output terminal 15 _n (n = 1, 2,..., N). Has been.

MIMO fading simulator performs frequency conversion and quadrature demodulation using downconverter 2 ₁ to 2 _M for RF or IF analog input signal input from input terminals 1 ₁ to 1 _M and inputs complex baseband into a signal, further converts the input signal is a digital signal by sampling using ADC 3 ₁ to 3 _M.

Each input calibration circuit 4 ₁ to 4 _M is the same structure, is composed of an input distortion compensation circuit 4A and the distortion learning circuit 4B as shown input calibration circuit 4 _m 2, the input signal as will be described later Distortion compensation is performed and an input signal after calibration is output. The post-calibration input signal from each input calibration circuit 4 _m is distributed to N FIR filters 5 _{m1 to} 5 _mN simulating N channels in the channel forming circuit 5, and set by the channel coefficient generator 7. The transmission line characteristics are affected. The output of the nth FIR filter 5 _mn among the N FIR filters 5 _{m1 to} 5 _mN corresponding to each input terminal 1 _m is given to the M input adder 6 _n corresponding to the output terminal 15 _n , and therefore Each M input adder 6 _n adds the outputs of M FIR filters 5 _{1n to} 5 _Mn .

The output of the M input adder 6 _n corresponding to the nth receiving antenna on the receiving side is added with the noise given from the noise generator 9 _n by the adder 8 _n , and the addition result is given to the output calibration circuit 10 _n. . As shown in FIG. 3, each output calibration circuit 10 _n includes an output distortion compensation circuit 10A, a training signal generation circuit 10B, and a signal selection switch 10C. As will be described later, an input signal of the output calibration circuit 10 _n is used. On the other hand, a post-calibration output signal that compensates for distortion generated by the up-converter 12 _n is generated. The DAC 11 _n converts the calibrated output signal into a complex baseband output signal that is an analog signal. The up-converter 12 _n converts the complex baseband output signal into RF or IF by orthogonal modulation and frequency conversion, and the variable attenuator 13 _n has a path loss set by a path loss multiplier of the channel coefficient generator 7 described later. At the same time, the output of the up-converter 12 _n is attenuated so that the output signal level has an optimum dynamic range, and an analog output signal is output from the output terminal 15 _n . These processes are performed on each of N post-calibration output signals.

歪学習回路４Ｂでは、以下の３つのステップによりダウンコンバータ２_mのDCオフセットを表す2次元のベクトルa^d _mとIQインバランスおよびチャネル間偏差を表す２行２列の行列D_mを求める。
(a) 第１のステップとして基準信号発生器（図示せず）を入力端子１_mに接続する。
(b) 第２のステップとして基準信号発生器の出力をオフにした状態、即ちs^t _m=0の状態
でADC３_mの出力信号

を観測し、式(1) におけるDCオフセットを表すベクトルa^d _mを推定する。

(c) 第３のステップとして基準信号発生器より予め決められた校正用のトレーニング信号

を入力してADC３_mの出力s^bb _mを測定し、式(1) によりIQインバランスおよびチャネル間偏差を表す行列D_mを推定する。
入力歪補償回路４Ａでは以下の式(3) によりADC３_mの出力信号s^bb _mに対してDCオフセットとIQインバランスおよびチャネル間偏差D_mの補償を行い、校正後入力信号

を出力する。
ここでは基準信号発生器を入力端子１_mに接続したが、図３に示した出力校正回路１０_nのトレーニング信号生成回路１０Ｂにおいて出力歪補償を行ったトレーニング信号を生成し、信号選択スイッチ１０Ｃをトレーニング信号生成回路１０Ｂ側に切り替え、出力端子１５_nを入力端子１_mに結線することによっても同様の校正を行うことができる。 Hereinafter, the main part will be described in more detail.
FIG. 2 shows the configuration of the input calibration circuit 4 _m (m = 1, 2,..., M). The input calibration circuit 4 _m includes a distortion compensation circuit 4A and a distortion learning circuit 4B. The input signal of the m-th down-converter 2 _m is ^st _m , the two-dimensional vector representing the DC offset of the down-converter 2 _m is a ^d _m, and a 2-by-2 matrix representing IQ imbalance and interchannel deviation. the When D _m, the output of the Dow converter 2 _m, output s ^bb _m of thus ADC 3 _m is expressed as follows.

In the distortion learning circuit 4B, a two-dimensional vector a ^d _m representing the DC offset of the down converter 2 _m and a 2-by-2 matrix D _m representing IQ imbalance and inter-channel deviation are obtained by the following three steps.
(a) As a first step, a reference signal generator (not shown) is connected to the input terminal 1 _m .
(b) While the machine is turned off the output of the reference signal generator as a second step, namely s ^t _m = 0 in the state ADC 3 _m of the output signal

And a vector a ^d _m representing the DC offset in the equation (1) is estimated.

(c) As a third step, a training signal for calibration predetermined by the reference signal generator

Input to the measured output s ^bb _m of ADC 3 _m, to estimate the matrix D _m representing the IQ imbalance and interchannel deviation by the formula (1).
Performs compensation of the DC offset and the IQ imbalance and interchannel deviation D _m relative to the output signal s ^bb _m of ADC 3 _m by the input distortion compensating circuit 4A in the following equation (3), after calibration input signal

Is output.
Here, the reference signal generator is connected to the input terminal 1 _m , but a training signal subjected to output distortion compensation is generated in the training signal generation circuit 10B of the output calibration circuit 10 _n shown in FIG. The same calibration can be performed by switching to the training signal generation circuit 10B side and connecting the output terminal 15 _n to the input terminal 1 _m .

となる。さらに、M入力加算器６_nは、s_mn(t_k)をmに関してM個加算してn番目の出力端子１５_nから出力されるチャネル受信信号y_n(t_k)を生成する。チャネル受信信号y_n(t_k)は

となる。 The channel coefficient generator 7 generates and outputs channel coefficients. Details of the channel coefficient generator 7 will be described later. The J tap FIR filter 5 _mn of the channel forming circuit 5 convolves the channel coefficient with the signal input from the m (m = 1, 2,..., M) -th input terminal 1 _m , and n (n = 1, 2,..., N) to the output terminal 15 _n . Further, the channel coefficient h _{mn, j} (t _k ) (j = 1, 2,..., J) for the FIR filter 5 _{mn is} assumed. Here, t _k is a sampling time and is an integer multiple of the sampling interval Δt. The FIR filter 5 _mn receives the post-calibration input signal s _m (t _k ), which is the output of the m-th input calibration circuit 4, and uses the channel coefficient h _{mn, j} (t _k ) as a weighting coefficient to _generate a channel signal. s _mn (t _k ) is generated. At this time, the channel signal s _mn (t _k ) is

It becomes. Further, the M input adder 6 _n adds M s _mn (t _k ) with respect to m to generate a channel reception signal y _n (t _k ) output from the _nth output terminal 15 _n . The channel received signal y _n (t _k ) is

It becomes.

The noise generator 9 _n generates a complex Gaussian random number w _n (t _k ) whose mean is 0 as noise and whose variance is a set value. Further, the adder 8 _n adds the channel reception signal y _n (t _k ) and the noise w _n (t _k ) to output a signal.
r _n (t _k ) = y _n (t _k ) + w _n (t _k ) (6)
Is output.

As shown in FIG. 3, the output calibration circuit 10 _n includes an output distortion compensation circuit 10A, a training signal generation circuit 10B, and a signal selection switch 10C. The output calibration circuit 10 _n performs distortion compensation on the output signal, and outputs a post-calibration output signal. Is output. Training generated by the training signal generation circuit 10B with a two-dimensional vector representing the DC offset of the n-th up-converter 12 _n as a ^u _n and a 2-by-2 matrix representing IQ imbalance and inter-channel deviation as M _n. the signal and r ^t _n, when the output signal of the up-converter 12 _n and s _n, the output signal s _n is
s _n = M _n r ^t _n + M _n a ^u _n (7)
It is expressed. Therefore, when performing the distortion learning, using the distortion learning circuit 4B of the input calibration circuit 4 _m, obtaining the M _n and a ^u _n upconverter 12 _n by the following four steps.

(c) 第３のステップとしてトレーニング信号生成回路１０Ｂにより予め決められた校正用のトレーニング信号

を出力し、その状態で歪学習回路４Ｂにより校正後入力信号s_mを観測して式(7) によりIQインバランスおよびチャネル間偏差を表す行列M_nを推定する。
(d) 第４のステップとして、第２のステップで推定したM_na^u _nと第３のステップで推定したM_nからDCオフセットを表すベクトルa^u _nを求める。得られたM_nとa^u _nは出力校正回路１０_nの出力歪補償回路１０Ａに与えられる。
出力歪補償回路１０Ａでは以下の式(9) により加算器８_nの出力信号ベクトル

に対してDCオフセットとIQインバランスおよびチャネル間偏差の補償を行い、校正後出力信号

を出力する。

ここでは出力端子１５_nを入力端子１_mに結線し、歪学習回路４ＢでDCオフセットとIQインバランスおよびチャネル間偏差を学習したが、出力端子１５_nに信号解析器を接続して信号の解析を行うことによっても同様の校正を行うことができる。 (a) As a first step, the signal selection switch 10C is switched to the training signal generation circuit 10B side, and the output terminal 15 _n is connected to the input terminal 1 _m .
(b) While the machine is turned off the output of the training signal generating circuit 10B as the second step, i.e., in the state of r ^t _n = 0, observing after calibration input signal s _m by the distortion learning circuit 4B. Since distortion of the down converter 2 _m has been compensated by the input distortion compensation circuit 4A is a s _m = s _n. Accordingly, the product of the vector a ^u _n representing the DC offset and the matrix M _n representing the IQ imbalance and the inter-channel deviation can be estimated from the equation (7).

(c) As a third step, a training signal for calibration predetermined by the training signal generation circuit 10B

Outputs, estimates the matrix M _n representing the IQ imbalance and interchannel deviation by the formula (7) by observing the calibration after the input signal s _m by the distortion learning circuit 4B in that state.
(d) a fourth step, determine the vector a ^u _n representing the DC offset from the M _n a ^u _n estimated by the second step M _n estimated by the third step. M _n and a ^u _n obtained is applied to the output distortion compensation circuit 10A of the output calibration circuit 10 _n.
In the output distortion compensation circuit 10A, the output signal vector of the adder 8 _n is obtained by the following equation (9).

DC offset, IQ imbalance and inter-channel deviation compensation, and output signal after calibration

Is output.

Here, the output terminal 15 _n is connected to the input terminal 1 _m , and the DC offset, IQ imbalance, and inter-channel deviation are learned by the distortion learning circuit 4B, but a signal analyzer is connected to the output terminal 15 _n to analyze the signal. The same calibration can be performed by performing the above.

From the above, according to the shape condition for carrying out the present invention, by using the input calibration circuit and output the calibration circuit, M-number of the down-converter and the N IQ imbalance and DC offset between up-converter or the like Can compensate for distortions and circuit characteristics variations, and realize a highly accurate MIMO channel as set.

Next, a configuration example of the channel coefficient generator 7 is shown in FIG. As shown in FIG. 4, the channel coefficient generator 7 according to the present invention includes first to J-th path component generators 14 _{1 to} 14 _J and an antenna spatial correlation coefficient generator 14K. J is an integer of 1 or more. The first path component generator 14 ₁ includes MN complex Gaussian random number generators 14A _{1 to} 14A _MN , MN Doppler filters 14B _{1 to} 14B _MN , an antenna / spatial correlation coefficient multiplier 14C, and MN pieces. Path power multipliers 14D _{1 to} 14D _MN , MN line-of-sight component adders 14E _{1 to} 14E _MN , MN fluorescent light component adders 14F _{1 to} 14F _MN , and MN path loss multipliers 14G ₁ to 14G _MN and MN interpolators 14H _{1 to} 14H _MN . The other path component generators 14 _{2 to} 14 _J have the same configuration.

Each path component generator 14 _j generates a complex Gaussian random number using the complex Gaussian random number generators 14A _{1 to} 14A _MN , and further filters the Doppler spectrum set using the Doppler filters 14B _{1 to} 14B _MN. Process. This process is realized by convolving a weighting coefficient with a complex Gaussian random number using an FIR filter or an infinite impulse response (IIR) filter. At the time t _{h when the} channel is generated, the MN complex Gaussian random number generators 14A _{1 to} 14A _MN and the Doppler filters 14B _{1 to} 14B _MN perform these processes, respectively. Each of the complex Gaussian random number generators 14A _{1 to} 14A _MN generates random numbers independent of each other.

となる。さらに、送信と受信アンテナの素子間結合を表す係数行列をそれぞれM行M列のQ_t、N行N列のQ_rとすると、アンテナ・空間相関係数行列Rは、

となり、アンテナ指向性、電力角度スペクトラム、素子間結合、アレイ応答ベクトルを設定することで、アンテナ・空間相関係数を計算できる。Rは、パス毎に電力角度スペクトラムを設定することで、パス毎に異なったものを生成することも可能である。 The antenna / spatial correlation coefficient generator 14K uses a predetermined antenna directivity and power angle spectrum for M transmitting antennas and N receiving antennas, and uses an antenna / spatial correlation coefficient matrix of MN rows and MN columns. Generate R. It is assumed that the transmission M-dimensional array response vector including the antenna directivity is a _t (φ _t ), and the reception N-dimensional array response vector is a _r (φ _r ). However, φ _t and φ _r are the radiation direction from the transmission antenna and the arrival direction to the reception antenna. At this time, the antenna / spatial correlation coefficient matrix R _t for transmission M rows and M columns and the N / N antenna / spatial correlation coefficient matrix R _r for reception in consideration of the antenna directivity are the power angle spectrum in the radial direction. Using P _t (φ _t ) and the power angle spectrum P _r (φ _r ) in the direction of arrival,

It becomes. Furthermore, assuming that the coefficient matrix representing the coupling between the elements of the transmitting and receiving antennas is Q _{t of} M rows and M columns and Q _{r of} N rows and N columns, the antenna-spatial correlation coefficient matrix R is

Thus, the antenna / spatial correlation coefficient can be calculated by setting the antenna directivity, power angle spectrum, inter-element coupling, and array response vector. R can generate different values for each path by setting a power angle spectrum for each path.

となる。 The antenna / spatial correlation coefficient multiplier 14C in the j-th path component generator 14 _j multiplies the Doppler channel coefficient u _{mn, j} (t _h ) by the antenna / spatial correlation coefficient to obtain a correlation channel coefficient g _{mn, j} ( t _h ) is generated. Now, suppose that the MN-dimensional Doppler channel coefficient vector having Doppler channel coefficients as elements is u _j (t _h ), the MN-dimensional correlation channel coefficient vector g _j (t _h ) having correlation channel coefficients as elements is Using the output R of the spatial correlation coefficient generator 14K

It becomes.

The path power multiplier 14D _{mn, j} , line-of-sight component adder 14E _{mn, j} , fluorescent light adder 14F _{mn, j} , and path loss multiplier 14G _{mn, j} are for the correlation channel coefficient g _{mn, j} (t _h ) Based on the set power delay profile, the correlation channel coefficient is multiplied by the path amplitude, the line-of-sight component and the channel fluctuation component due to the fluorescent light are added, and the channel coefficient before interpolation is multiplied by the attenuation coefficient due to propagation and shadowing. h _{mn, j} (t _h ) is generated. At this time, the attenuation coefficient is optimally adjusted so that the quantization noise does not increase in consideration of realizing the attenuation amount to be set by the variable attenuator 15 _n .

The interpolator 14H _{mn, j} interpolates the pre-interpolation channel coefficient h _{mn, j} (t _h ) generated at the time t _h to obtain the channel coefficient h _{mn, j} (t _k ) at the sampling time t _k. Is generated.

From the above, according to the shape condition for carrying out the present invention, by generating the channel coefficients using antenna spatial correlation coefficients based on antenna directivity and power angle spectrum in channel coefficient generator, The fading transmission path of the MIMO channel taking into account the antenna directivity can be simulated.

FIG. 5 shows another embodiment according to the MIMO fading simulator. The configuration of FIG. 5 is the same as the configuration of FIG. 1 except that the distortion adding circuit 16 _m is inserted between each input calibration circuit 4 _m and the FIR filters 5 _{m1 to} 5 _{mN in} the configuration of FIG. A duplicate description is omitted. Since the distortion due to the down converter 2 _m and the up converter 12 _n is compensated by the input calibration circuit 4 _m and the output calibration circuit 10 _n , the distortion does not increase by passing through the MIMO fading simulator. However, the distortion adding circuits 16 _{1 to} 16 _M are used so that the influence of distortion or the like generated in the transmitter can be verified by the MIMO fading simulator according to the present invention. The post-calibration input signal that is the output of the input calibration circuit 4 _m is input to the distortion adding circuit 16 _m , added with distortion based on the set function, and output as a distortion added input signal. The distortion-added input signal is input to the FIR filters 5 _{m1 to} 5 _mN instead of the post-calibration input signal, and all subsequent processing is the same.

From the above, according to the shape condition for carrying out the present invention, by adding the distortion based on the function set for the input signal using a distortion adding circuit, MIMO fading influence by any strain Can be verified by simulator.

The present invention is not limited to the shape condition for carrying out the respective invention described above, it can naturally be adopted various other structures without departing from the gist of the present invention.

1 ₁ to 1 _M : input terminal, 2 ₁ to 2 _M : down converter, 3 _{1 to} 3 _M : ADC, 4 ₁ to 4 _M : input calibration circuit, 5: channel forming circuit: 5 _{11 to} 5 _MN : J tap FIR filter, 6 _{1 to} 6 _N : M input adder, 7: channel coefficient generator, 8 ₁ to 8 _N : adder, 9 _{1 to} 9 _N : noise generator, 10 ₁ to 1
0 _N : Output calibration circuit, 11 _{1 to} 11 _N : DAC, 12 _{1 to} 12 _N : Upconverter, 13 ₁ to 1
3 _N : Variable attenuator, 15 ₁ to 15 _N : Output terminal, 14 _{1 to} 14 _J : Path component generator, 14A _{1 to} 14A
_MN: complex Gaussian random number generator, 14B ₁ ~14B _MN: Doppler filters, 14C: antenna spatial correlation coefficient multiplier, 14K: antenna spatial correlation coefficient generator, 14D ₁ ~14D _MN: Path power multiplier, 14E _{1 to} 14E _MN : Line-of-sight component adder, 14F _{1 to} 14F _MN : Fluorescent light component adder, 14G _{1 to} 14G _MN
: Path loss multiplier, 14H _{1 to} 14H _MN : Interpolator, 16 _{1 to} 16 _M : Distortion adding circuit

Claims (5)

In a MIMO fading simulator having M input terminals and N output terminals and simulating transmission path characteristics between M transmitting antennas and N receiving antennas,
M downconverters that convert input signals from the M input terminals into complex baseband input signals, and M is a positive integer.
M input calibration circuits for compensating the distortion generated by the down converter for the complex baseband input signal and outputting as a post-calibration input signal;
The channel formation formed by the calibrated input signals from the M input calibration circuits and the MN channels between the M inputs and the N outputs, which are respectively called J-tap finite impulse response filters, hereinafter referred to as FIR filters. Circuit, N and J are positive integers,
A channel coefficient generator that gives the MN FIR filters J channel coefficients that simulate the characteristics of each channel;
N output calibration circuits which are provided with N outputs of the channel forming circuit and which respectively perform distortion compensation and output as post-calibration output signals which are complex baseband signals;
N up-converters that up-convert the output signals after calibration of the N output calibration circuits and output the signals to the N output terminals, respectively.
And the channel coefficient generator includes the J channels based on antenna directivity and inter-element coupling coefficients of the M transmitting antennas and the N receiving antennas. Coefficients are generated and set for the MN FIR filters, and the N output calibration circuits compensate for the distortion generated by the N up-converters for the N outputs of the channel forming circuit, respectively. And generating the post-calibration output signal by performing processing as described above,
MNJ complex Gaussian random number generators that generate complex Gaussian random numbers;
MNJ Doppler filters that convolve weighting coefficients so as to be a Doppler spectrum set for the complex Gaussian random number,
An antenna / spatial correlation coefficient generator that generates an antenna / spatial correlation coefficient using the antenna directivity and the inter-element coupling coefficient;
J antenna / spatial correlation coefficient multipliers that generate a correlation channel coefficient by multiplying the Doppler channel coefficient that is the output of the Doppler filter by the antenna / spatial correlation coefficient that is the output of the antenna / spatial correlation coefficient;
Furthermore, look at including a door,
The channel coefficient is obtained from the generated correlation channel coefficient.
This is a MIMO fading simulator. In the MIMO fading simulator according to claim 1,
M analog-to-digital converters that convert the outputs of the M down-converters into digital signals and supply them to the M input calibration circuits,
N noise generators each generating noise,
N adders for adding the noise from the N noise generators to the N outputs of the channel forming circuit and inputting the addition results to the N output calibration circuits, respectively.
N digital-to-analog converters that convert the post-calibration output signals from the N output calibration circuits into analog signals and supply them to the N up-converters, respectively
N variable attenuators for adjusting and outputting the output levels of the N upconverters,
And a MIMO fading simulator.   2. The MIMO fading simulator according to claim 1, wherein each of the input calibration circuits compensates for the estimated DC offset and IQ imbalance by learning, a distortion learning circuit that estimates the corresponding DC offset and IQ imbalance of the down converter. A MIMO fading simulator comprising: an input distortion compensation circuit that processes the signal from the down converter to generate the post-calibration input signal.   4. The MIMO fading simulator according to claim 3, wherein each of the output calibration circuits processes a training signal generation circuit that generates a known training signal and a corresponding output signal of the channel formation circuit to generate the post-calibration output signal. The up-converter by inputting a training signal of the training signal generation circuit to any one of the M input terminals from a corresponding one of the N output terminals during distortion learning. DC offset and IQ imbalance are estimated by learning using the distortion learning circuit of the input calibration circuit, and the output distortion compensation circuit is configured to form the channel based on the estimated DC offset and IQ imbalance of the up-converter. MIM characterized by processing the corresponding output signal of the circuit to generate the post-calibration output signal O fading simulator.   5. The MIMO fading simulator according to claim 1, wherein a distortion adding circuit is provided between the input calibration circuit and the channel forming circuit, and the post-calibration input is an output of the input calibration circuit. A MIMO fading simulator, wherein a signal is input to the distortion adding circuit, distortion is added based on a set function, and the signal is output and input to the channel forming circuit.

Images