JP7748675B2 - Base station device, weight generation method, and wireless communication system - Google Patents

Description

The present invention relates to a base station device, a weight generation method, and a wireless communication system.

To meet the ever-increasing demand for wireless communications, progress is being made in implementing multi-user MIMO (Multiple-Input Multiple-Output) technology, which multiplexes and transmits signals for multiple terminals over the same space, time, and frequency (see, for example, Non-Patent Document 1). With multi-user MIMO, it is assumed that the number of antenna elements at a terminal is smaller than the number of antenna elements at a base station. In this case, it is difficult for the terminal to separate signals for multiple terminals, so it is common for the base station to perform interference suppression processing (hereinafter referred to as "null formation") to prevent interference between multiple terminals before transmitting the signals.

To achieve this null formation, it is necessary to accurately grasp the state of the propagation path between the antenna element of the base station that transmits the signal and the antenna element of the terminal that receives the signal. This requires propagation path estimation (hereinafter also referred to as "channel estimation"), which involves transmitting a known signal between the transmitter and receiver in advance and estimating the state of the propagation path based on information about how much that signal has changed. This propagation path estimation requires transmitting a known signal, which results in communication overhead, so it is generally performed discretely when communication is needed or at regular intervals specified by communication standards.

As the frequency bands used in wireless communications expand, research is also underway into Massive MIMO technology, which uses multiple antenna elements for transmitting and receiving antennas, in order to compensate for the increased distance attenuation and increased noise power that accompanies wideband transmission (see, for example, Non-Patent Document 2). Massive MIMO employs multiple antenna elements, enabling more flexible signal control than ever before, and offering a high degree of antenna freedom.

However, wireless communication propagation paths change constantly as terminals move and the surrounding environment changes. Therefore, if the propagation path obtained by propagation path estimation changes from the propagation path actually used for communication, the effectiveness of null formation is weakened, resulting in inter-user interference and degrading transmission characteristics. Therefore, focusing on the high antenna flexibility of Massive MIMO, null space expansion technology is being studied, which suppresses inter-user interference by forming multiple nulls per terminal antenna, rather than the conventional spatial multiplexing transmission in which one null is formed per terminal antenna (see, for example, Non-Patent Documents 3 and 4, and Patent Document 1). While conventionally, after the minimum necessary null formation is performed, the remaining antenna freedom is used to obtain gain for the desired signal, null space expansion technology performs additional null formation, thereby enabling multi-user MIMO transmission that is robust to changes in the propagation path.

[About multi-user MIMO technology]
(Overview of Multi-User MIMO)
Coherent transmission and phased array antenna technology are basically technologies for improving line gain, and in order to increase the line capacity when a wide service area is covered by a single base station device, another wireless communication technology is required. On the other hand, since frequency resources are limited, here we will explain, for example, the multi-user MIMO technology discussed in Non-Patent Document 1 as a technology for utilizing limited resources with high frequency utilization efficiency.

Figure 2 is a schematic diagram showing an example configuration of a multi-user MIMO system. As shown in the figure, the multi-user MIMO system comprises a base station device 801 and terminal devices 802-1, 802-2, and 802-3 (terminal devices #1 to #3). In reality, a single base station device 801 accommodates a large number of terminal devices 802, but only a few of these (terminal devices 802-1 to 802-3 in the figure) are selected for communication. Each terminal device 802 generally has fewer transmit and receive antennas than the base station device 801. The following describes communication (downlink) from base station device 801 to terminal device 802.

Base station device 801 forms multiple directional beams using a large number of antenna elements. For example, consider the case where three MIMO channels are assigned to each of terminal devices 802-1 to 802-3, and a total of nine signal sequences are transmitted. In this case, the signal transmitted to terminal device 802-1 is adjusted so that the directional gain is extremely low (so that nulls are formed) in the directions of terminal devices 802-2 and 802-3, thereby suppressing interference with terminal devices 802-2 and 802-3. Similarly, the signal transmitted to terminal device 802-2 is adjusted so that the directional gain is extremely low in the directions of terminal devices 802-1 and 802-3. Similar processing is performed on terminal device 802-3. The reason for performing directivity control in this manner is that, for example, terminal device 802-1 has no way of knowing the information about the signals received by terminal devices 802-2 and 802-3, and therefore cooperative reception processing between the terminal devices 802 is not possible. In other words, it is extremely difficult to separate all nine signal sequences in the reception processing of terminal device 802-1, which has only three antennas. Therefore, interference separation is performed in advance on the transmitting side to prevent each of terminal devices 802-1 to 802-3 from receiving signals from other terminal devices 802. This concludes the overview of existing multi-user MIMO systems.

Next, a method for forming a directional beam will be described below. Here, a case will be described in which base station device 801 has nine antenna elements and each of terminal devices 802-1 to 802-3 has three antenna elements. For example, in FIG. 2, channel information between the jth (j = 1, ..., 9) antenna element of base station device 801 and the first antenna element of terminal device 802-1 is denoted as h _1j . Using channel information between each antenna element (j = 1, ..., 9) of base station device 801 and the first antenna element of terminal device 802-1, row vector h ₁ is denoted as (h ₁₁ , h ₁₂ , h ₁₃ , ..., h ₁₈ , h ₁₉ ). Similarly, the channel information between the j-th antenna element of the base station device 801 and the second and third antenna elements of the terminal device 802-1 is expressed as _h2j and _h3j , and the corresponding row vectors _h2 and _h3 are expressed as ( _h21 , _h22 , _h23 , ..., _h28 , _h29 ) and ( _h31 , _h32 , _h33 , ..., _h38 , _h39 ). The antenna elements of the terminal device 802-2 and the terminal device 802-3 are assigned similar consecutive numbers, and the row vectors _h4 to _h9 are expressed as ( _h41 , _h42 , _h43 , ..., _h48 , _h49 ) to ( _h91 , _h92 , _h93 , ..., _h98 , _h99 ).

Channel information refers to values that represent the state of the wireless communication propagation path, i.e., the channel, which is the actual propagation coefficient between the transmitting antenna and the receiving antenna, estimated or obtained by the transmitter or receiver (strictly speaking, this includes the effects of amplifiers and filters within the transmitter and receiver).

In addition, the nine systems of signals transmitted by base station device 801 are denoted as _t1 to _t9 , and the column vector having these as components is denoted as Tx ^[all] = ( _t1 , _t2 , _t3 , ..., _t8 , _t9 ) ^T . Here, the letter T on the right superscript represents the transpose of a vector or matrix. Similarly, the received signals at the nine antenna elements of terminal devices 802-1 to 802-3 are denoted as _r1 to _r9 , and the column vector having these as components is denoted as Rx ^[all] = ( _r1 , _r2 , _r3 , ..., _r8 , _r9 ) ^T . Finally, the matrix having row vectors _h1 to _h9 as the first to ninth row components is denoted as the overall channel matrix H ^[all] . Additionally, noise is denoted as n.

In this case, the following relationship (1) holds for the entire multi-user MIMO system.

In order to control the transmission directivity, a 9-row, 9-column transmission weight matrix W is introduced, and equation (1) is rewritten as equation (2) below.

Furthermore, if the transmission weight matrix W is decomposed into column vectors w ₁ to w ₉ and written as W = (w ₁ , w ₂ , w ₃ , ..., w ₈ , w ₉ ), then "H ^[all] ·W" in equation (2) can be expressed as the following equation (3).

Here, for example, consider selecting values of w1 to w3 so that the multiplication of six row vectors _h4 to _h9 by _three column vectors _w1 to w3 (the sum of the products of each component; in the case of complex vectors, this is different from the inner product) is all zero. At the same time, select values of _w4 to _w9 so that the multiplication of row vectors _h1 to _h3 and _h7 to _h9 by column vectors _{w4 to w6} , and _the multiplication of row vectors _h1 to _h6 by column vectors _w7 to _w9 are all zero.

Then, the 9-row, 9-column matrix H ^[all] ·W shown in equation (3) can be expressed as the following equation (4) using a 3-row, 3-column submatrix.

In equation (4), H ^[1] , H ^[2] , and H ^[3] are 3-row, 3-column matrices, and "0" is a 3-row, 3-column matrix whose components are all zero. By selecting a transformation matrix that satisfies these conditions as the transmission weight matrix W, equation (4) can be decomposed into three relational equations expressed by the following equations (5-1) to (5-3).

Here, Tx ^[1] = ( _t1 , _t2 , _t3 ) ^T , Tx ^[2] = ( _t4 , _t5 , _t6 ) ^T , Tx ^[3] = ( _t7 , _t8 , _t9 ) ^T , Rx ^[1] = ( _r1 , _r2 , _r3 ) ^T , Rx ^[2] = ( _r4 , _r5 , _r6 ) ^T , Rx ^[3] = ( _r7 , _r8 , _r9 ) ^T . In this way, it can be considered that one base station device performs one-to-one MIMO communication, that is, three systems of single-user MIMO communication are performing communication simultaneously in parallel.

Next, an example of a method for determining transmission weight vectors w ₁ to w ₉ will be described below. The procedure is to determine transmission weight vectors w ₁ to w ₃ for terminal device 802-1, then determine transmission weight vectors w ₄ to w ₆ for terminal device 802-2, and transmission weight vectors w ₇ to w ₉ for terminal device 802-3.

First, as a first step, six basis vectors _e4 to _e9 are obtained in a six-dimensional subspace spanned by six row vectors _h4 to _h9 for terminal devices 802-2 and 802-3. There are various methods for obtaining these vectors, including the Gram-Schmidt orthogonalization method, but the Gram-Schmidt orthogonalization method will be used as an example here.
First, one row vector _h4 is focused on, and a vector in this direction with an absolute value of 1 is defined as a basis vector _e4 . The basis vector _e4 is expressed by the following equation (6).

In equation (6), (h ₄ h ₄ ^H ) is a scalar quantity that means the square of the absolute value of the same vector, and division by the square root of this value means normalizing the row vector h _4. Also, "h ₄ ^H " is the Hermitian conjugate vector of the row vector h ₄ , and is a vector obtained by transposing the rows and columns and taking the complex conjugate of each component.

Next, we focus on row vector _h5 , and calculate row vector _h5 ' by canceling the component in the direction of basis vector _e4 from this row vector, and then normalize it. Row vector _h5 ' and basis vector _e5 are expressed by the following equations (7-1) and (7-2).

(h ₅ e ₄ ^H ) in equation (7-1) means the projection of row vector h ₅ in the direction of basis vector e _4. Similar processing is performed as in the following equations (8-1) and (8-2).

Here, the range of the sum of Σ in equation (8-1) is the sum for integer i in the range 4≦i≦(j-1) (j is an integer between 5 and 9). In other words, this means canceling out the component in the direction of the already determined specified vector. In this way, six basis vectors _e4 to _e9 can be determined.

Next, in the second step, transmission weight vectors _w1 to _w3 for terminal device 802-1 are obtained. First, the components of the six-dimensional subspace spanned by basis vectors _e4 to _e9 are cancelled from row vectors _h1 to _h3 . Specifically, this is expressed by the following equation (9).

Here, j in equation (9) is an integer from 1 to 3, and the range of the sum of Σ is the sum for integers i in the range of 4≦i≦9. The three-dimensional space spanned by the three row vectors h ₁ ' to h ₃ ' thus obtained is orthogonal to all of the row vectors h ₄ to h ₉ described above. By selecting three vectors (which do not necessarily have to be orthogonal vectors) within this three-dimensional space and setting the complex conjugate vectors of these vectors as transmission weight vectors w ₁ to w ₃ , it is possible to suppress interference with other terminal devices 802-2 and 802-3.

The three vectors can be selected in any manner, but for example, if three orthogonal vectors that form a unitary matrix obtained by performing singular value decomposition are used, eigenmode transmission can be limited to a subspace that does not cause interference to other terminal devices 802, enabling efficient transmission.

Finally, in the third step, if the same processing is performed for the terminal device 802-2 and the terminal device 802-3, the overall transmission weight vectors w ₁ to w ₉ can be finally determined.

The above is an example of how to calculate the transmission weight matrix W.

3 is a flowchart showing an example of a procedure for calculating a transmission weight matrix W in a multi-user MIMO system. First, in calculating the transmission weight matrix W, channel matrices H for all terminal devices 802 to be multiplexed are obtained (step S801). A serial number is assigned to each destination terminal device 802, and k is a variable indicating the serial number. First, k is initialized (step S802). Next, k is counted up (step S803), and a partial channel matrix (referred to as H ^main here for convenience) for the terminal device 802 (#k) corresponding to the current value of k is extracted (step S804), and partial channel matrices (referred to as H ^sub here for convenience) for the other destination terminal devices 802 are extracted (step S805).

Furthermore, the orthogonal basis vectors of the subspace spanned by each row vector of the partial channel matrix H ^sub are calculated, and these are designated as basis vectors {e _j } (step S806). Next, as a process equivalent to equation (9), the components relating to the basis vector {e _j } calculated in step S806 are cancelled from the partial channel matrix H ^main for the terminal device 802 (#k) of interest, and this is designated as the matrix ∼H ^main (step S807). Here, in step S807, H with a tilde (~) above it is written as ∼H. Similarly, in the following mathematical expressions, when a character with a symbol such as a hat (^) above it is written, the symbol is written before the character.

Furthermore, any orthogonal basis vector of the subspace spanned by the row vectors of the matrices ~ ^Hmain is calculated and set as the basis vector {e _k } (step S808). Here, the any basis vector may be, for example, a vector that constitutes a right singular matrix when the matrices ~ ^Hmain are subjected to singular value decomposition. After that, a transmission weight vector {w _{k } for the signal of the terminal device 802(#k) is determined as the Hermitian conjugate vector (column vector obtained by transposing the complex conjugate vector) of each vector of the basis vector {e k }} (step S809).

Here, it is determined whether the transmission weight vectors for all destination terminal devices 802 have been determined (step S810), and if there are any remaining terminal devices 802, the processes from step S803 to step S809 are repeated. If the transmission weight vectors for all terminal devices 802 have been determined, a transmission weight matrix W is determined as a matrix with the transmission weight vectors {w _k } as column vectors (step S811), and the process ends.

In addition, since channel information generally differs for each frequency component, for example, for a wideband signal, such as a signal using the OFDM (Orthogonal Frequency Division Multiplexing) modulation method, a similar transmission weight is calculated for each frequency component, i.e., for each subcarrier. Also, since the case where each of the terminal devices 802-1 to 802-3 has three antenna elements has been described here, step S808 includes the process of calculating the orthogonal basis vectors of the subspace spanned by each row vector of the matrix ~H ^main . However, if the terminal device has only one antenna, step S808 simply corresponds to normalizing the row vector corresponding to the matrix ~H ^main .

The above is a general method for calculating transmit and receive weights for multi-user MIMO, which assumes that the terminal device is equipped with multiple antennas and block diagonalizes the entire channel matrix as shown in equation (4). However, there are several other variations of the same method for calculating transmit and receive weights. These variations do not necessarily require the terminal device to have a single antenna, but for simplicity, the following explanation will assume that N terminal devices with a single antenna are spatially multiplexed simultaneously. Other methods for calculating transmit and receive weights are explained below.

First, with regard to the transmitting and receiving weights of the base station device 801, a ZF (Zero Forcing) pseudo-inverse matrix expressed by the following equations (10-1) and (10-2) may be calculated for the entire channel matrix H ^[all] shown in equation (1) etc., and used as the transmitting weight and the receiving weight.

Here, assuming that the number of terminal devices to be spatially multiplexed is N and the number of antenna elements of the base station device 801 is K (N<K), taking the downlink as an example, the size of the overall channel matrix H ^[all] is N×K (N rows and K columns). If the rank of H[all ^] is N, the size of the matrix H ^[all] ·H ^[all]H is N×N, and an inverse matrix exists, and a pseudo-inverse matrix can be obtained using equation (10-1). Generally, if the value of K is sufficiently redundant with respect to N, the rank of this N×N matrix will stably be N, and an inverse matrix will stably exist. Similarly, with regard to the uplink receiving weights corresponding to reception by the base station device 801, the size of the overall channel matrix H ^[all] is K×N (K rows and N columns), and the size of the matrix H ^[all]H ·H ^[all] is also N×N. Generally, an inverse matrix exists, and a ZF-type pseudo-inverse matrix expressed by the following equation (10-2) may be calculated and used as the receiving weight.

In addition, in the case of MMSE (Minimum Mean Square Error) weights, which are known as similar transmitting and receiving weights, if the noise power is σ ² , the following equations (11-1) and (11-2) may be used instead of equations (10-1) and (10-2). Note that "I" in equations (11-1) and (11-2) is an N×N (N rows and N columns) unit matrix.

(Example of Multi-User MIMO Device Configuration)
4 is a schematic block diagram showing an example of the configuration of a base station device 80 in a multi-user MIMO system. As shown in the figure, the base station device 80 includes a transmitter 81, a receiver 85, an interface circuit 87, a MAC (Medium Access Control) layer processing circuit 88, and a communication control circuit 820. The MAC layer processing circuit 88 includes a scheduling processing circuit 881.

The base station device 80 inputs and outputs data to and from external devices or networks via the interface circuit 87. The interface circuit 87 detects, from the input data, data to be transferred over the wireless line and outputs the detected data to the MAC layer processing circuit 88. The MAC layer processing circuit 88 performs MAC layer processing in accordance with instructions from the communication control circuit 820, which manages and controls the operation of the entire base station device 80. Here, MAC layer processing includes converting data input and output by the interface circuit 87 and data transmitted and received over the wireless line, and adding MAC layer header information. During this processing, the scheduling processing circuit 881 performs various scheduling processes, including combining terminal devices that simultaneously perform spatial multiplexing in multi-user MIMO transmission. The scheduling processing circuit 881 outputs the scheduling results to the communication control circuit 820. In multi-user MIMO, signals are transmitted to multiple terminal devices at once, so multiple signal sequences are output from the MAC layer processing circuit 88 to the transmitter 81.

Figure 5 is a schematic block diagram showing an example of the configuration of a transmitter 81 in a base station device 80 in a multi-user MIMO system. As shown in the figure, the transmitter 81 includes transmission signal processing circuits 811-1 to 811-L (L is an integer equal to or greater than 2), addition and synthesis circuits 812-1 to 812-K (K is an integer equal to or greater than 2), IFFT (Inverse Fast Fourier Transform) & GI (Guard Interval) adding circuits 813-1 to 813-K, D/A (Digital-to-Analog) converters 814-1 to 814-K, a local oscillator 815, mixers 816-1 to 816-K, filters 817-1 to 817-K, high-power amplifiers (HPAs) 818-1 to 818-K, antenna elements 819-1 to 819-K, and a transmission weight processing unit 830. The transmission signal processing circuits 811-1 to 811-L and the transmission weight processing unit 830 are connected to the communication control circuit 820 shown in Figure 4.

The transmission weight processing unit 830 includes a channel information acquisition circuit 831, a channel information storage circuit 832, and a multi-user MIMO (MU-MIMO) transmission weight calculation circuit 833. Here, the subscript L in the names of the transmission signal processing circuits 811-1 to 811-L in the figure indicates the number of simultaneous spatial multiplexing operations. Furthermore, the subscript K in the names of the summing circuits 812-1 to 812-K and the antenna elements 819-1 to 819-K indicates the number of antenna elements included in the base station device 80.

In multi-user MIMO, signals are transmitted simultaneously to multiple terminal devices. Therefore, multiple signal sequences are input from the MAC layer processing circuit 88 to the transmitter 81, and these signal sequences are then input to the transmission signal processing circuits 811-1 to 811-L. When the data (data inputs #1 to #L) to be transmitted to each destination terminal device is input from the MAC layer processing circuit 88, the transmission signal processing circuits 811-1 to 811-L generate wireless packets to be transmitted over the wireless link and perform modulation processing. Here, if, for example, OFDM modulation is used, the signals of each signal sequence are modulated for each frequency component. Furthermore, the modulated baseband signals are multiplied by transmission weights for each frequency component. The signals multiplied by the transmission weights corresponding to each antenna element 819-1 to 819-K undergo the remaining signal processing as necessary, and are input to the summing and combining circuits 812-1 to 812-K as sampled data of the baseband transmission signal.

Signals input to summing circuits 812-1 to 812-K are combined for each frequency component. The combined signals are converted from frequency-domain signals to time-domain signals by IFFT & GI adding circuits 813-1 to 813-K, and further processing such as the insertion of guard intervals and waveform shaping between OFDM symbols (between blocks of block transmission in the case of SC-FDE (Single-Carrier Frequency Domain Equalization)) is performed. For each antenna element 819-1 to 819-K, D/A converters 814-1 to 814-K convert the digital sampling data into a baseband analog signal. Furthermore, each analog signal is multiplied by the local oscillation signal input from local oscillator 815 by mixers 816-1 to 816-K, and upconverted to a radio frequency signal. Here, since the upconverted signal contains frequency components outside the band of the channel to be transmitted, filters 817-1 to 817-K remove these frequency components to generate the electrical signal to be transmitted. The generated signal is amplified by high-power amplifiers 818-1 to 818-K and transmitted from antenna elements 819-1 to 819-K.

In Figure 5, after additive synthesis of the signals of each frequency component is performed in additive synthesis circuits 812-1 to 812-K, processing such as IFFT processing, guard interval insertion, and waveform shaping is performed. However, these processes may be performed in transmit signal processing circuits 811-1 to 811-L, and the IFFT & GI adding circuits 813-1 to 813-K may be omitted. In this case, the remaining signal processing as necessary after transmit weight multiplication in transmit signal processing circuits 811-1 to 811-L refers to processing such as IFFT processing, guard interval insertion, and waveform shaping.

In addition, the transmission weights multiplied in the transmission signal processing circuits 811-1 to 811-L are obtained from the multi-user MIMO transmission weight calculation circuit 833 provided in the transmission weight processing unit 830 during signal transmission processing. In the transmission weight processing unit 830, the channel information acquisition circuit 831 separately acquires the channel information acquired by the receiving unit 85 via the communication control circuit 820, and stores this information in the channel information storage circuit 832 while updating it sequentially. When transmitting a signal, in accordance with instructions from the communication control circuit 820, the multi-user MIMO transmission weight calculation circuit 833 reads channel information corresponding to the destination station from the channel information storage circuit 832 and calculates the transmission weights based on the read channel information. The multi-user MIMO transmission weight calculation circuit 833 outputs the calculated transmission weights to the transmission signal processing circuits 811-1 to 811-L.

The communication control circuit 820 also manages all communication-related control, including destination station management and overall timing control. The communication control circuit 820 outputs information indicating the destination station, etc. to the transmission weight processing unit 830, which performs signal processing related to the calculation of the transmission weights described above.

Figure 6 is a schematic block diagram showing an example of the configuration of a receiver 85 in a base station device 80 in a multi-user MIMO system. As shown in the figure, the receiver 85 includes antenna elements 851-1 to 851-K, low-noise amplifiers (LNAs) 852-1 to 852-K, a local oscillator 853, mixers 854-1 to 854-K, filters 855-1 to 855-K, A/D (analog-to-digital) converters 856-1 to 856-K, FFT (Fast Fourier Transform) circuits 857-1 to 857-K, receive signal processing circuits 858-1 to 858-L, and a receive weight processing unit 860. The receive signal processing circuits 858-1 to 858-L and the receive weight processing unit 860 are connected to the communication control circuit 820 shown in Figure 4. The receiving weight processing unit 860 includes a channel information estimation circuit 861 and a multi-user MIMO (MU-MIMO) receiving weight calculation circuit 862.

Signals received by antenna elements 851-1 to 851-K are amplified by low-noise amplifiers 852-1 to 852-K. The amplified signals are multiplied by the local oscillation signal output from local oscillator 853 in mixers 854-1 to 854-K, and the amplified signals are downconverted from radio frequency signals to baseband signals. Because the downconverted signals contain frequency components outside the frequency band to be received, the out-of-band components are removed by filters 855-1 to 855-K. The signals from which the out-of-band components have been removed are converted into digital baseband signals by A/D converters 856-1 to 856-K. All digital baseband signals are input to FFT circuits 857-1 to 857-K, which convert the time-domain signals into frequency-domain signals (separate them into signals of each frequency component) at the specified symbol timing. The signals separated into each frequency component are input to the received signal processing circuits 858-1 to 858-L, and also to the channel information estimation circuit 861.

The channel information estimation circuit 861 estimates channel information for each frequency component between the antenna elements of each terminal device and each antenna element 851-1 to 851-K of the base station device 80 based on a known signal for channel estimation (such as a preamble signal attached to the beginning of a wireless packet) separated into each frequency component, and outputs the estimation results to the multi-user MIMO receiving weight calculation circuit 862. The multi-user MIMO receiving weight calculation circuit 862 calculates the receiving weight to be multiplied for each frequency component based on the input channel information. At this time, the receiving weights used to combine the signals received by each antenna element 851-1 to 851-K differ for each signal sequence, and are input to the receiving signal processing circuits 858-1 to 858-L corresponding to the signal sequence to be extracted.

The receiving signal processing circuits 858-1 to 858-L multiply the signals for each frequency component input from the FFT circuits 857-1 to 857-K by the receiving weights input from the multi-user MIMO receiving weight calculation circuit 862, and then add and combine the signals received by each antenna element 851-1 to 851-K for each frequency component. The receiving signal processing circuits 858-1 to 858-L demodulate the added and combined signals and output the recovered data to the MAC layer processing circuit 88.

Here, different reception signal processing circuits 858-1 to 858-L perform signal processing of different signal sequences. Furthermore, the MAC layer processing circuit 88 performs processing related to the MAC layer (e.g., conversion between data input/output to/from the interface circuit 87 and data transmitted/received over the wireless line, termination of MAC layer header information, etc.). During this processing, the scheduling processing circuit 881 performs various scheduling processes, including combinations of terminal devices that simultaneously perform spatial multiplexing in multi-user MIMO transmission, and outputs the scheduling results to the communication control circuit 820. The received data processed by the MAC layer processing circuit 88 is output to an external device or network via the interface circuit 87.

The communication control circuit 820 also manages the overall communication control, including management of the source terminal device and overall timing control. The communication control circuit 820 also inputs information indicating the source terminal device, etc., to the receive weight processing unit 860, which performs signal processing related to the calculation of the receive weights described above.

As with transmission, in wideband systems using OFDM modulation or SC-FDE, the multiplication of the above-mentioned receiving weights is performed for each frequency component during signal reception. In other words, the signals output from A/D converters 856-1 to 856-K are subjected to FFT in FFT circuits 857-1 to 857-K to separate them into individual frequency components. Each separated frequency component is then subjected to signal processing in channel information estimation circuit 861 and received signal processing in received signal processing circuits 858-1 to 858-L.

(Multi-user MIMO transmission processing)
7 is a flowchart showing the transmission process of the base station device 80 in multi-user MIMO. In multi-user MIMO, downlink channel information is periodically fed back separately from data transmission. When the channel information acquisition circuit 831 acquires downlink channel information (step S831), it stores the channel information of each frequency component for each terminal device in the channel information storage circuit 832 (step S832). The processes of steps S831 and S832 are performed sequentially.

When signal transmission processing from the base station device 80 begins (step S821), the multi-user MIMO transmission weight calculation circuit 833 reads channel information for each frequency component corresponding to the destination terminal device from the channel information storage circuit 832 (step S822).

The multi-user MIMO transmission weight calculation circuit 833 calculates the multi-user MIMO transmission weight for each frequency component using the processing described above based on the read channel information (step S823). Separately from the processing of steps S822 and S823, the transmission signal processing circuits 811-1 to 811-L generate transmission signals for each frequency component for each destination station using transmission signal processing such as various modulation processes for the data to be transmitted to each destination (step S824).

The transmit signal processing circuits 811-1 to 811-L multiply the generated transmit signals by the transmit weights calculated by the multi-user MIMO transmit weight calculation circuit 833 in step S823 (step S825). The transmit signal processing circuits 811-1 to 811-L also perform a series of signal processing, and the summing circuits 812-1 to 812-K sum and combine the transmit signals addressed to each terminal device for each frequency component for each antenna element 819-1 to 819-K. The IFFT & GI adding circuits 813-1 to 813-K then convert the signals from the frequency domain to the time domain, and perform further processing such as inserting guard intervals and shaping waveforms between OFDM symbols (between blocks in block transmission in the case of SC-FDE), before outputting them to the D/A converters 814-1 to 814-K (steps S826-1 to S826-K).

The signals output from the IFFT & GI application circuits 813-1 to 813-K are processed by the D/A converters 814-1 to 814-K and the high-power amplifiers 818-1 to 818-K, and then transmitted from the antenna elements 819-1 to 819-K (steps S827-1 to S827-K), respectively, and the process ends (steps S828-1 to S828-K).

Note that the processing in steps S827-1 to S827-K includes upconversion from baseband signals to radio frequencies, removal of frequency components using a filter, and signal amplification using a high-power amplifier.

(Multi-user MIMO Reception Processing)
8 is a flowchart showing the reception processing of the base station device 80 in multi-user MIMO. First, when the reception processing starts (step S840), signals are received by the first through Kth antenna elements 851-1 through 851-K (steps S841-1 through S841-K). Here, reception includes the processing up to and including analog-to-digital conversion of the received signal or a signal obtained by down-converting the received signal. The subsequent signal processing refers to processing of the digitized received signal.

Next, FFT circuits 857-1 to 857-K perform signal processing, such as separating the received signals corresponding to antenna elements 851-1 to 851-K into individual frequency components (steps S842-1 to S842-K). Furthermore, channel information estimation circuit 861 performs channel estimation for each frequency component based on the reception state of the preamble signal of a known pattern attached to the wireless packet (steps S843-1 to S843-K). Here, the attenuation of the signal on the propagation path and the rotation state of the complex phase are determined. In the channel estimation performed in steps S843-1 to S843-K, channel estimation must be performed individually for each spatially multiplexed signal sequence, as shown individually in steps S843-1, S843-2, ..., S843-K.

This individual channel estimation must be performed in a state where the signals transmitted from each transmitting terminal device can be separated. Taking the OFDM modulation method as an example, a channel estimation preamble signal with the same number of symbols as the spatial multiplexing number is generally required. Each terminal device transmits a signal with the same number of symbols as the spatial multiplexing number (or more), each with a different preamble signal pattern, and the base station device 80 uses the differences in these patterns to perform individual channel estimation in steps S843-1 to S843-K.

The multi-user MIMO receiving weight calculation circuit 862 uses the channel information estimated by the channel information estimation circuit 861 to calculate appropriate receiving weights for each spatially multiplexed signal sequence and frequency component (step S844). Furthermore, the receiving signal processing circuits 858-1 to 858-L multiply the receiving weights calculated for each signal sequence and frequency component by the received signals of each antenna element separated for each frequency component (steps S845-1 to S845-K).

Here, since a receiving weight is prepared for each spatially multiplexed signal sequence, the multiplication results in steps S845-1 to S845-K are different for each spatially multiplexed signal sequence. For each signal sequence, the signals from antenna elements 851-1 to 851-K are added together for each frequency component (steps S846-1 to S846-L). The combined signal sequence is then subjected to signal processing from the first signal sequence (step S847-1) to the Lth signal sequence (step S847-L), after which the process ends (steps S848-1 to S848-L).

Note that for simplicity, an example using linear receiving weights has been shown here. However, in general, for MIMO, nonlinear signal processing such as MLD (Maximum Likelihood Detection) may be performed. In this case, the processing in steps S845-1 to S845-L, steps S846-1 to S846-L, and steps S847-1 to S847-L will be performed as a unified nonlinear signal detection process. Furthermore, linear receiving weights can be calculated using a method similar to the transmission weight calculation process shown in FIG. 3. Alternatively, receiving weights using pseudo-inverse matrices or MMSE weights can also be used. Furthermore, here, the number of spatially multiplexed signal sequences is L, relative to the number of antenna elements 851-1 to 851-K used for reception, K. However, generally, K and L do not need to be the same. As long as the value of the spatial multiplexing number L is equal to or less than the number of antennas K, signals of multiple signal sequences can be spatially multiplexed.

It is generally known that MIMO transmission performance improves as the number of antenna elements on the transmitting and receiving stations increases relative to the number of spatially multiplexed signal sequences. This improvement in performance is utilized in the form of improving the SINR (Signal to Interference and Noise Ratio) of each spatially multiplexed signal sequence and further increasing the number of spatial multiplexing sequences. In recent years, progress has been made in implementing Massive MIMO technology, such as that discussed in Non-Patent Document 2, which expands the number of antenna elements on the base station side to an ultra-large number of 100 or more. Null space expansion technology is being studied as an interference suppression technology that utilizes this ultra-large number of antenna elements on the base station side (see, for example, Non-Patent Document 3 and Patent Document 1).

In multi-user MIMO technology, a base station device grasps channel information for all terminal devices to be multiplexed, and transmits signals by performing interference suppression processing based on the grasped channel information so as not to cause interference between the terminal devices. As mentioned above, there are multiple techniques for interference suppression processing, but all of these techniques have in common the point that they add pre-processing based on channel information for terminal devices other than the destination terminal device so that interfering signals are not received by terminal devices other than the destination terminal device (so that nulls are formed). For example, focusing on the i-th terminal device, based on the partial channel matrix H _i ^sub for the other terminal devices, a weight is generated that is orthogonal to each row vector of the partial channel matrix H _j ^sub for the other j (j ≠ i)-th terminal device, and the weight is multiplied by the signal for the i-th terminal device to perform interference suppression.

On the other hand, in the null space extension technology, the partial channel matrix H _i ^sub for all multiplexed terminal devices is extended by inserting additional channel vectors, and interference suppression is performed based on the extended partial channel matrix H _i ' ^sub , thereby forming nulls over a wider range than in conventional technology. This means that in addition to the null in the direction that it should originally be directed, a new null is formed in another direction, but when the channel for terminal devices other than the i-th one fluctuates over time, even if it deviates from the original null point, as long as the destination of the fluctuation is near the additional null point, the same interference suppression effect can be obtained, and thus a high inter-user interference reduction effect can be obtained even in a time-varying environment.

As an example, consider the case where a base station device with a 100-element antenna simultaneously transmits spatially multiplexed signals to 10 terminal devices, each with a single antenna element. In this case, the channel vector for each terminal device is a 100-dimensional vector. Conventional technology uses 10 degrees of freedom for spatial multiplexing of 10 terminals, i.e., interference suppression for the 10 terminals, and the remaining 90 (= 100 - 10) degrees of freedom to improve the line gain of each terminal device through in-phase combining. Here, processing is performed assuming that there is another virtual terminal device (actually, the original terminal device has moved) in its vicinity that should suppress interference. In conventional technology, when the channel fluctuated due to the movement of each terminal device, the effectiveness of interference suppression was reduced and interference increased significantly. However, if the terminal device's time-varying location coincides with the location of the virtual terminal device, interference suppression is expected to be similarly achieved. In this case, interference suppression is also performed for the additional virtual terminal device, consuming one additional degree of freedom for each terminal device.

In other words, the total number of degrees of freedom used for interference suppression becomes 20, and the degrees of freedom available for improving line gain through in-phase combining are reduced to 80. If the line gain improvement rate is proportional to the first power of the degrees of freedom, the difference in the degrees of freedom used for line gain improvement is 10 Log (80/90) = -0.51... [dB], a difference of only about 0.5 dB. However, if the time fluctuations of the channel are contained within the range of the additional interference suppression implemented as described above, the power of inter-user interference can be significantly reduced. In other words, although the SNR (Signal to Noise Ratio) will deteriorate by approximately 0.5 dB, a significant improvement in the SIR (Signal to Interference power Ratio) is expected. While the transmission capacity of multi-user MIMO is ultimately determined by the SINR (Signal to Interference plus Noise power Ratio), this method is effective because SIR characteristics are considered to be more dominant than SNR in a time-varying environment.

Next, we will explain the key points of null space extension technology. Figure 1 shows a beam pattern generated by a base station (BS) equipped in a wireless communication system using null space extension technology, and Figure 9 shows a beam pattern generated by a base station (BS) equipped in a wireless communication system using conventional technology. In a wireless communication system, a base station and multiple terminal devices can perform spatial multiplexing transmission on the same frequency at the same time. In the wireless communication systems shown in Figures 1 and 9, a base station communicates with terminal device T1 and terminal device T2 using spatial multiplexing transmission. The time-varying predicted destination for terminal device T1 is T1', and the time-varying predicted destination for terminal device T2 is T2'.

As shown in Figure 9, a base station device using conventional technology transmits a signal to destination terminal device T1, located in the direction of arrow D1, using beam pattern B8, which directs a null toward terminal device T2 other than terminal device T1. However, when terminal device T2 moves to the time-variation predicted location T2', indicated by the dotted line, terminal device T2 moves away from the null in the direction of arrow D2, resulting in increased interference power at terminal device T2. In contrast, as shown in Figure 1, a base station device using null space expansion technology generates beam pattern B1 by adding a null to point T2', where time variation is expected in the direction of arrow D2', based on the expanded channel matrix. This makes it possible to suppress interference power as long as terminal device T2 falls within the range of the null formed by the above procedure, whether the actual movement of terminal device T2 is to point T2' where time variation is expected or if terminal device T2 remains in its original position.

Fig. 11 shows an example of the configuration of an extended channel matrix used in calculating weights using null space extension technology, and Fig. 10 shows an example of the configuration of a channel matrix used in calculating weights using conventional technology. For simplicity, each terminal device is assumed to have one antenna, and a case where L terminal devices are spatially multiplexed is considered. h ₁ (t) to h _L (t) are channel vectors of each terminal device at time t, with the dimension of the number of base station antenna elements. Note that channel vector h _i is (h _i1 , h _i2 , ..., h _ij ), and h _ij is channel information between the j-th antenna element of the base station device and the i-th terminal device.

As shown in Fig. 10, in the prior art, a channel matrix is used in which channel vectors corresponding to the number of terminals that simultaneously perform multiplex transmission are arranged. When multiplexing transmission is performed toward terminal device #1, interference between users is suppressed by using weights that perform processing such as orthogonalization so that when multiplied by channel vectors _h2 (t) to _hL (t) (= _H1sub ) to other terminal devices #2 to #L, ^the wireless signals cancel each other out and become null.

On the other hand, as shown in FIG. 11 , in the null space extension technology, the partial channel matrix ^{H1main is common to the conventional technology, but the partial channel matrix H1sub} _is extended by adding past channel vectors _h2 (t-nT) to _hL (t-nT) for the other terminal devices #2 to #L in addition to the current channel vectors _h2 (t) to _hL (t) for the other terminal devices #2 to #L. The past channel vector _h1 (t-nT) is the channel vector acquired before time nT for the i-th terminal device #i. Here, it is assumed that propagation path estimation is performed every time period T, and n=1... is an integer representing how many periods ago the channel vector was estimated. However, if propagation path estimation is not performed periodically, any past time may be given as an argument. Furthermore, the number of past channel vectors to be extended may differ depending on the terminal device #i.

The key point of the null space extension technology is the process of extending the partial channel matrix H ^sub . By extending the partial channel matrix H ^sub more than the weight calculation in existing technologies and then calculating the weights, interference suppression can be performed over a wider range than with existing technologies. By performing processes such as orthogonalization on the extended partial channel matrix H' ₁ ^sub extended as described above, a beam is directed toward the desired terminal device #1, and in addition to forming nulls for other terminal devices, weights for which additional nulls have been formed can be obtained, making it possible to suppress interference between users in a time-varying environment. In other words, the null space extension technology calculates weights that, when multiplied by the channel vector of the destination terminal device, align the phase of the radio signal, and that, when multiplied by the channel vectors and past channel vectors of terminal devices other than the destination, cancel each other out to form a null. Note that the past channel vector added to generate the partial channel matrix H ^sub does not necessarily have to be one for each terminal device; multiple past channel vectors may be generated and added. Furthermore, a different number of past channel vectors may be generated for each terminal device.

In addition, the weight generated by the null space expansion technique not only directs null to both the current channel vector h ₂ (t) to h _L (t) and the past channel vector h ₂ (t-nT) to h _L (t-nT) of other terminal devices #2 to #L, but also directs null to the entire subspace spanned by the linear combination of these vectors. That is, specifically, for the current channel vector h _j (t) and the past channel vector h _j (t-nT) of terminal device #j, null is also directed to all channel vectors given by the linear combination such as given by h _j (t) + γ × {h _j (t-nT) - _{h j} (t)} for any complex coefficient γ. For this reason, there is no need to predict pinpoint propagation path fluctuations represented by highly accurate "points", and it is sufficient to appropriately extract the subspace to which null should be directed.

In other words, if the channel varies over time within the subspace, interference will be suppressed at any time δt. Conventional channel prediction techniques used to address time variations predict the channel at the desired transmission time δt and pinpoint weights that suppress interference for that predicted channel. This requires either predicting the channel and generating weights for each transmission request, or predicting all channels at the expected transmission times in advance and storing the generated weights. On the other hand, using weights based on null space expansion technology makes it possible to obtain weights that suppress interference by forming nulls once in accordance with updates to channel information.

JP 2016-136706 A

Yasushi Takatori et al., "Application of Downlink Multi-User MIMO Technology to Next-Generation High-Speed Wireless Access Systems," Institute of Electronics, Information and Communication Engineers, IEICE Transactions on Wireless Access Systems, Vol. J93-B, No. 9, pp. 1127-1139, September 2010. Kazuki Maruta et al., "Proposal for a Large-Scale Antenna Wireless Entrance System - Characteristic Evaluation by Computer Simulation -," Institute of Electronics, Information and Communication Engineers, Technical Report of the Institute of Electronics, Information and Communication Engineers, RCS2013-6, vol. 113, no. 8, pp. 31-36, April 2013 T.Iwakuni, et. al, "Null-Space Expansion for Multiuser Massive MIMO Inter-User Interference Suppression in Time Varying Channels," IEICE TRANSACTIONS on Communications, Vol.E100-B, No.5, pp.865-873 Taromaru, M. et al., "A Study on Beam Patterns of Multi-User MIMO Base Stations Formed by Null Space Expansion Method - Consideration of the Principle of Improving Channel Fluctuation Tolerance by Orthogonalizing to Past Interference Channels," Institute of Electronics, Information and Communication Engineers, Technical Report of the Institute of Electronics, Information and Communication Engineers, RCS2017-226, vol. 117, no. 284, pp. 123-127, November 2017.

The generation of weights in null space expansion technology poses the following challenge. When attempting to form multiple nulls when the amount of propagation path variation is small, the null formation consumes a significant amount of freedom, resulting in a decrease in the desired signal gain. Furthermore, the null formation calculations require a large amount of calculation, and the interference suppression effect is limited.

The present invention was made in light of these circumstances, and aims to provide a base station apparatus, wireless communication method, and wireless communication system that can extract appropriate channel vectors and generate weights to efficiently suppress inter-user interference caused by channel fluctuations.

The present invention was made in light of these circumstances, and aims to provide a base station device, a weight generation method, and a wireless communication system that can efficiently suppress interference with a low amount of calculation.

One aspect of the present invention is a base station apparatus in a wireless communication system that includes a base station apparatus equipped with multiple antenna elements and multiple terminal devices, and that allows the base station apparatus and the terminal devices to perform spatial multiplexing transmission on the same frequency at the same time. The base station apparatus includes a weight calculation unit that calculates weight vectors for performing spatial multiplexing transmission for the multiple terminal devices based on a channel matrix that arranges channel vectors generated from channel information between the antenna elements of the terminal devices or a virtual antenna element obtained by combining the antenna elements and the antenna elements of the base station apparatus, and additional channel vectors that are different from the channel vectors and associated with the antenna elements of the terminal devices. The weight calculation unit selects, as the additional channel vector for the terminal device, a second channel vector that has a low degree of similarity to the first channel vector from among one or more second channel vectors acquired before a first channel vector, which is the channel vector of the terminal device, and calculates the weight vector.

Another aspect of the present invention is a weight generation method executed by a base station device equipped with multiple antenna elements and multiple terminal devices in a wireless communication system capable of performing spatial multiplexing transmission on the same frequency at the same time, the weight generation method comprising the steps of: calculating weight vectors for performing spatial multiplexing transmission for the multiple terminal devices based on a channel matrix in which channel vectors generated from channel information between the antenna elements of the terminal devices or a virtual antenna element obtained by combining the antenna elements and the antenna elements of the base station device are arranged; and selecting, as the additional channel vector for the terminal device, a second channel vector that has a low degree of similarity to the first channel vector from one or more second channel vectors acquired before the first channel vector, which is the channel vector of the terminal device, and calculating the weight vector.

Another aspect of the present invention is a wireless communication system comprising a base station device equipped with multiple antenna elements and multiple terminal devices, capable of performing spatial multiplexing transmission between the base station device and the terminal devices at the same time on the same frequency. The base station device comprises a weight calculation unit that calculates weight vectors for performing spatial multiplexing transmission for the multiple terminal devices based on a channel matrix that arranges channel vectors generated from channel information between the antenna elements of the terminal devices or a virtual antenna element obtained by combining the antenna elements and the antenna elements of the base station device, and additional channel vectors that are different from the channel vectors and are associated with the antenna elements of the terminal devices. The weight calculation unit calculates the weight vectors by selecting, as the additional channel vector for the terminal device, from one or more second channel vectors acquired before a first channel vector, which is the channel vector of the terminal device, the second channel vector that has a low degree of similarity to the first channel vector.

The present invention makes it possible to perform interference suppression efficiently with a low computational load.

10A and 10B are diagrams illustrating examples of beam patterns generated by a base station device according to an embodiment of the present invention. 1 is a schematic diagram showing an example of the configuration of a multi-user MIMO system according to the prior art; 1 is a flowchart showing a procedure for calculating a transmission weight matrix W in a multi-user MIMO system according to the prior art. FIG. 1 is a schematic block diagram showing an example of the configuration of a base station device 80 in a conventional multi-user MIMO system. 1 is a schematic block diagram showing an example of the configuration of a transmitter 81 provided in a base station device 80 in a multi-user MIMO system according to the prior art. 1 is a schematic block diagram showing an example of the configuration of a receiver 85 provided in a base station device 80 in a multi-user MIMO system according to the prior art. 10 is a flowchart showing a transmission process of a base station device 80 in multi-user MIMO according to the prior art. 10 is a flowchart showing a receiving process of a base station device 80 in multi-user MIMO according to the prior art. FIG. 1 is a diagram illustrating an example of a beam pattern generated by a base station device according to the prior art. FIG. 1 is a diagram illustrating an example of the configuration of a channel matrix used for weight calculation in the prior art. FIG. 10 is a diagram illustrating an example of the configuration of an extended channel matrix used for weight calculation in an embodiment of the present invention. FIG. 2 is a schematic block diagram showing an example of the configuration of a transmission weight processing unit 130 according to the embodiment of the present invention. 10 is a flowchart showing the operation of a transmission weight processing unit 130 according to the embodiment of the present invention. 10 is a flowchart showing the operation of a transmission weight processing unit 130 according to the embodiment of the present invention.

Embodiments of the present invention will be described below with reference to the drawings.

The basic principle of an embodiment of the present invention will now be explained. It is known that the shape of the nulls formed in null space extension technology results in null formation in propagation path element waves that cause channel fluctuations at interfering terminals (see, for example, Non-Patent Document 4). However, it is difficult to estimate the direction of arrival of the element waves themselves. Therefore, null space extension technology can be said to be a technology that automatically suppresses interference in the direction of element waves by adding past channel vectors for interfering terminals as an extended channel matrix.

At this time, null formation for a propagation path with small channel fluctuation results in null formation for adjacent elementary waves, and therefore the interference suppression effect is reduced. That is, when null formation is performed for the current channel vector h _i (t) and past channel vector h i (t-nT) of other terminal device #i in the partial channel matrix H ₁ ^sub , if h _i (t) and h _i (t-nT) are similar, it can be said that there is no need to perform null formation for each of them. Therefore, in an embodiment of the present invention, an index of similarity is introduced between the current channel vector and past channel vector related to these other terminal devices, and only those with low similarity are used in the weight calculation of the null space expansion technology.

This makes it possible to perform multi-user MIMO transmission using only channel vectors that are effective in suppressing interference.

The difference between the configuration of the base station device 80 in this embodiment of the present invention and the configuration of the conventional base station device 80 shown in FIG. 4 is that the configuration of the transmission weight processing unit 830 of the transmitter 81 shown in FIG. 5 is the configuration of the transmission weight processing unit 130 described below. Below, components having the same configuration as the configuration of the conventional base station device 80 are assigned the same reference numerals, and their description will be omitted.

As mentioned above, the multi-user MIMO system of the embodiment described below is a wireless communication system that includes a base station device 80 equipped with multiple antenna elements and multiple terminal devices 802, and is capable of performing spatial multiplexing transmission between the base station device 80 and the terminal devices 802 at the same time on the same frequency.

[Configuration of transmission wait processing unit]
The transmission weight processing unit 130 (weight calculation unit) in the embodiment calculates weight vectors for spatial multiplexing transmission to multiple terminal devices 802 based on a channel matrix that arranges channel vectors generated from channel information between the antenna elements of the terminal device 802 or virtual antenna elements obtained by combining the antenna elements and the antenna elements provided by the base station device 80, and additional channel vectors associated with the antenna elements of the terminal device 802 that are different from the channel vectors.

Here, the transmission weight processing unit 130 selects, as an additional channel vector for the terminal device 802, a second channel vector that has a low degree of mutual similarity with the first channel vector from among one or more second channel vectors acquired before the first channel vector, which is the channel vector of the terminal device 802, and calculates a weight vector.

A block diagram of the transmission weight processing unit 130 in an embodiment of the present invention is shown in Figure 12, and weight calculation will be explained in detail. The following describes the case where transmission weights are calculated.

Figure 12 is a block diagram showing the functional configuration of the transmission weight processing unit 130 in an embodiment of the present invention. As shown in Figure 12, the transmission weight processing unit 130 includes a channel information acquisition circuit 831, a channel information storage circuit 132, a MU-MIMO transmission weight calculation circuit 133, a channel similarity calculation circuit 135, and a channel information selection circuit 136.

The channel similarity calculation circuit 135 calculates the mutual similarity between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i based on the absolute value of the dot product of the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i. For example, the channel similarity calculation circuit 135 calculates the correlation coefficient between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i using the following equation (12).

Here, H represents a conjugate transpose vector. If |γ _i (n)|, which is the absolute value of this correlation coefficient γ _i (n), is equal to or less than a certain value, the channel similarity calculation circuit 135 determines that the similarity between the propagation paths is low, and records that the similarity is low in h _i (t−nT).

The correlation coefficient may be normalized by the norm of each channel vector. The channel similarity calculation circuit 135 may calculate the mutual similarity between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i based on the cross-correlation value obtained by dividing the absolute value of the inner product of the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i by the product of the norm of the current channel vector h _i (t) and the past channel vector h _i (t-nT). In this case, equation (12) is rewritten as the following equation (13).

Furthermore, in equation (13), each channel vector is normalized using the norm of each channel vector, but the norm of the channel vector is expected to have a certain magnitude depending on the number of elements of the base station antenna, etc. Therefore, the channel similarity calculation circuit 135 may calculate the mutual similarity between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i based on an approximate correlation value obtained by dividing the absolute value of the inner product of the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i by the norm of either the current channel vector h _i (t) or the past channel vector h _i (t-nT) of the terminal device #i, or by the square of either norm.

That is, for example, instead of equation (13), the approximate norm expressed in the following equation (14) or (15) may be used.

Furthermore, the difference vector between the channel vectors used for null formation represents the difference in the direction of the null to be formed for that channel vector, and by acquiring the norm, it becomes possible to evaluate it as the similarity between the vectors. Therefore, the channel similarity calculation circuit 135 may calculate the mutual similarity between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i based on the norm of the difference vector between the current channel vector h _i (t) and the past channel vector h _i (t-nT) of the terminal device #i, or based on the value obtained by dividing the norm of the difference vector by the norm of either the current channel vector h _i (t) or the past channel vector h _i (t-nT) of the terminal device #i.

That is, for example, the norm of the difference vector d _i (n) expressed by the following equation (16) can be used as the similarity.

After calculating the similarity between channel vectors using these methods, information on past channel vectors h _i (t−nT) that have been recorded as having low similarity is sent to a channel information selection circuit 136 .

The channel information selection circuit 136 reads out only past channel vectors h _i (t−nT) that have been recorded as having low similarity from the channel information storage circuit 132, and transfers them to the MU-MIMO transmission weight calculation circuit 133. This enables the MU-MIMO transmission weight calculation circuit 133 to obtain the extended partial channel matrix H _i ′ ^sub for the i-th terminal device, and to calculate a transmission weight based on it.

In addition, the channel information selection circuit 136 may determine the weight vector by excluding weight vectors calculated before the similarity between the channel vectors having a value higher than a predetermined value appears more than a predetermined number of times.

[Operation of the transmission wait processing unit]
The operation of the transmission weight processing unit 130 is shown as a flowchart in, for example, FIGS.

First, the channel similarity calculation circuit 135 calculates the similarity of the above-mentioned channel vectors for each terminal that performs spatial multiplexing transmission. That is, for terminal #k, the similarity between the most recent channel vector _hk (t) and the past channel vector _hk (t-nT) obtained in the nth channel estimation is calculated using one of equations (12) to (16) (step S050).

If the similarity obtained in this way is equal to or less than a certain value (step S050, YES), the channel similarity calculation circuit 135 records the past channel vector h _k (t-nT) as one to be used in the partial channel matrix (step S060). If n reaches a predetermined upper limit or if there are no more past channel vectors (e.g., they are not stored), the process ends. Otherwise (step S080, YES), n is incremented by 1 (step S040), and the similarity is calculated in the same way.

When the calculation of the similarity of past channel vectors is complete, the channel similarity calculation circuit 135 performs the same similarity calculation for all terminals that simultaneously transmit via spatial multiplexing (steps S020 to S090). When the above processing is complete for all terminals (step S090, YES), the channel similarity calculation circuit 135 calculates the partial channel matrices Hmain and H'sub for each terminal device #k (steps S120 to S130).

The MU-MIMO transmission weight calculation circuit 133 calculates the orthogonal basis vectors of the subspace spanned by each row ^vector of the extended partial channel matrix H'sub, and sets this as the basis vector { _e'j } (step S135). Next, the MU-MIMO transmission weight calculation circuit 133 cancels the components related to the basis vector { _e'j } calculated in step S135 from the partial channel matrix ^Hmain for the terminal device 802-k of interest, and sets this as the matrix ~ ^H'main (step S140). Furthermore, the MU-MIMO transmission weight calculation circuit 133 calculates any orthogonal basis vector of the subspace spanned by the row vectors of the matrix ~ ^H'main , and sets this as the basis vector { _e'k } (step S145).

Here, the arbitrary basis vector may be, for example, a vector that constitutes a right singular matrix when singular value decomposition is performed on the matrix ~H ^main . Thereafter, the MU-MIMO transmission weight calculation circuit 133 determines a transmission weight vector {w' _k } for the signal of the terminal device 802-k as the Hermitian conjugate vector (column vector obtained by transposing the complex conjugate vector) of each vector in the basis vector {e' _k } (step S150).

The MU-MIMO transmission weight calculation circuit 133 determines whether transmission weight vectors have been determined for all destination terminal devices 802 (step S155). If the MU-MIMO transmission weight calculation circuit 133 determines that there is an unprocessed terminal device 802 (step S155: NO), it repeats the processes from step S120 to step S150. If the MU-MIMO transmission weight calculation circuit 133 determines that transmission weight vectors have been determined for all destination terminal devices 802 (step S155: YES), it determines a transmission weight matrix W' as a matrix with the transmission weight vector {w' _k } as each column vector (step S160), and ends the process. The k-th column of the transmission weight matrix W' is the transmission weight vector w' _k .

In this way, by performing orthogonalization processing, the base station device 80 of this embodiment can extract only channel vectors that are effective in interference suppression and perform null formation, thereby efficiently generating multi-user MIMO weights.

The transmitter 81 of the base station device 80 performs transmission processing using the transmission weight matrix W' determined in the transmission weight calculation processing shown in Figures 13 and 14 described above. As a result, the transmission signal s from the base station device 80 becomes the transmission weight matrix W' x the transmission signal t from the base station device to each terminal. However, the transmission signal s is a column vector whose elements are the transmission signals s ₁ , s ₂ , ..., s _K from each of the antenna elements 819-1 to 819-K, and the transmission signal t is a column vector whose elements are the transmission signals t ₁ , t ₂ , ..., t _L from the base station device 80 to each of the terminal devices 802-1 to 802-L.

As explained above, when multiple channel vectors used in null space extension technology are similar, the interference suppression effect of null formation using both of them decreases. Therefore, the base station device 80 in an embodiment of the present invention calculates the similarity between channel vectors and uses only channel vectors whose similarity is below a certain level for weight formation. This allows the base station device 80 to efficiently suppress inter-user interference. According to the base station device 80 in an embodiment of the present invention, by extracting only channel vectors that are effective for interference suppression and generating weights in null space extension technology, it becomes possible to perform efficient interference suppression with a low amount of calculation.

As described above, the present invention relates to null formation technology using multiple antenna elements that is applicable to multi-user MIMO wireless communications and the like. Null formation technology has been studied in the past to suppress interference that occurs between users, etc., by generating weights for signals from other users, other beams, and other streams and performing matrix calculations. However, forming nulls consumes degrees of freedom, i.e., the number of antenna elements used, and while interference can be suppressed, there is an issue of reduced gain for the desired signal. Furthermore, conventional methods have the issue of the enormous amount of calculation required for null generation. In response to this issue, the present invention devise a new formula for calculating the similarity between channel vectors of each channel, and uses those channel vectors for weight generation only when the similarity is high, enabling efficient interference suppression and resolving the above-mentioned issues.

[Other supplementary points regarding the embodiment of the present invention]
Below, some supplementary points regarding the embodiments of the present invention are summarized.

In embodiments of the present invention, any method can be used for downlink channel estimation. For example, when acquiring downlink channel information, a method (explicit feedback method) can be used in which a base station device transmits a predetermined training signal on the downlink, a terminal device performs channel estimation based on the training signal, and the channel information is directly fed back to the base station device using a wireless packet containing predetermined control information. Similarly, a method (implicit feedback method) can be used in which a terminal device transmits a predetermined training signal on the uplink, a base station device performs uplink channel estimation based on the training signal, and the downlink channel information is estimated after a predetermined calibration process based on the obtained uplink channel information.

As such, there are many variations in how the channel information acquisition circuit 831 in Figure 12 can be implemented, but in the embodiment of the present invention, this processing occurs after the channel information acquisition circuit 831 has completed acquiring channel information, so there is no impact no matter what channel information acquisition method is used in the channel information acquisition circuit 831 of the prior art. In other words, the embodiment of the present invention is applicable to any channel information acquisition method.

In the above explanation, for simplicity's sake, subscripts representing frequency components have been omitted, and explanations of individual frequency components have also been omitted. However, generally, all signal processing, including channel information, transmit and receive weights, and transmit and receive signals, is defined and processed individually for each frequency component on the frequency axis. Within each signal processing circuit, for example, all signal processing up to the stage preceding IFFT processing on the transmitting side (bit sequence interleaving, signal point mapping, signal modulation, transmit weight multiplication, etc.) is performed for each frequency component. Similarly, all signal processing from FFT processing on the receiving side (receive weight multiplication, signal detection, signal demapping, deinterleaving, etc.) is also performed for each frequency component. For this reason, acquisition of downlink channel information is also performed for each frequency component, and similarly, prediction of future channel information is also performed for each frequency component.

However, while downlink channel information is typically obtained individually for all frequency components, it is expected that the accuracy of future channel predictions will be lower than when simple channel information is obtained. In this case, it is not necessary to perform channel prediction for all subcarriers; it is expected that characteristics will not deteriorate significantly even if channel prediction is performed by thinning out frequency components to a certain extent and using predicted channel vectors for nearby frequency components. For example, if channel prediction is performed once for three frequency components, it is acceptable to use predicted channel vectors for frequency components before and after the frequency component for which prediction was performed. If the correlation in the frequency direction is even stronger, it is acceptable to perform channel prediction by thinning out even more.

In terms of circuit configuration, individual circuits may be provided for each frequency component, or, since the same processing is performed, processing may be performed serially for each frequency component in order, with the circuit being shared among all frequency components. Furthermore, as an intermediate step, multiple circuits may be provided, the frequency components may be divided appropriately, and parallel processing may be performed serially across the multiple circuits. These are common to all embodiments.

When a terminal device is equipped with a plurality of antenna elements and the dimension of the partial channel matrix H ^main is two or more, it is not essential to perform orthogonalization of the partial channel matrix H ^main , which is a channel matrix for a desired terminal, using transmission weights on the base station device side. For example, if signal separation between different terminal devices is possible using a transmission weight generation method such as block diagonalization, desired signal separation within the same terminal device (suppression of inter-stream interference) can be handled by signal processing on the terminal device side.

Furthermore, in the OFDM modulation method, all subcarriers are used for communication with the same terminal device, so the transmit and receive weights (averaged transmit and receive weight vector and real-time transmit and receive weight matrix) used at that time are transmit and receive weights for the terminal device with a common combination across all subcarriers. However, in OFDMA (Orthogonal frequency-division multiple access), different combinations of assignments to terminal devices are gathered together in a patchwork pattern on the time and frequency axes, so it is necessary to use transmit and receive weights for the assigned terminal device for each time (OFDM symbol) and frequency (subcarrier). However, apart from this difference, OFDM and OFDMA can be processed in exactly the same way, and although this specification has focused on OFDM, embodiments of the present invention can also be applied in exactly the same way to OFDMA.

While there are various operational variations of SC-FDE, the above-described configuration examples apply the same processing as conventional SC-FDE in both the received signal processing after the signals transmitted from each antenna element are spatially combined by multiplying the signals by an averaged transmit weight on the transmitting side, and the received signal processing after the signals from each antenna element are added and combined by multiplying the signals by an averaged receive weight on the receiving side, making them applicable to all variations of SC-FDE. In this case, single-carrier signal processing is performed instead of OFDM modulation signal processing, and then, in the downlink, FFT processing is performed on the single-carrier signal on the time axis to generate signal components for each frequency component. These signal components are then treated as subcarrier signals generated by OFDM modulation and multiplied by the transmit weights generated by an embodiment of the present invention.

Similarly, for the uplink, the received signal is subjected to FFT processing, and the signal is treated in the same way as in the OFDM modulation method, and is separated by multiplying it by the transmission weight generated by an embodiment of the present invention. The separated frequency component signals are then subjected to IFFT processing to convert them into single-carrier signals on the time axis. As such, while there are differences in some signal processing between the OFDM modulation method and SC-FDE, processes such as the generation of transmission and reception weights and multiplication are common, and embodiments of the present invention can be applied to either signal method.

Furthermore, while this specification is written assuming wideband systems such as OFDM and OFDMA, embodiments of the present invention are equally applicable to narrowband systems.

Furthermore, for the sake of convenience, this specification does not distinguish between "row vectors" and "column vectors." In strict mathematical notation that unifies the direction of vector arrangement, symbols such as "transpose" should be used. However, the information required to implement this invention is the value of each component of a vector, and whether the vector is a row vector or a column vector does not make much sense. Therefore, for the sake of ease of understanding, this specification does not distinguish between "row vectors" and "column vectors."

For example, in equation (3), the channel vector h _i is a row vector, and the transmission weight vector w _j is a column vector. As shown in Fig. 11, the extended partial channel matrix H' ^sub used in calculating the transmission weights in this embodiment of the present invention has channel vectors and predicted channel vectors of terminal devices not subject to transmission weight calculation arranged in each row. That is, in the case of the downlink, the predicted channel vectors of the row vectors are added to the partial channel matrix H ^sub .

On the other hand, in the case of the uplink, the channel vectors of each terminal device are column vectors, and the receiving weight vectors are column vectors. Therefore, the extended partial channel matrix H' ^sub used to calculate the receiving weights is obtained by arranging the channel vectors and predicted channel vectors of terminal devices not subject to receiving weight calculation in each column. In other words, in the case of the uplink, the predicted channel vectors, which are column vectors, are added to the partial channel matrix H ^sub .

According to the above-described embodiment, the wireless communication system includes a base station device equipped with multiple antenna elements and multiple terminal devices, and the base station device and terminal devices can perform spatial multiplexing transmission on the same frequency at the same time. For example, the base station device is base station device 80 in the embodiment, and the terminal device is terminal device 802 in the embodiment. The above-described base station device includes a weight calculation unit. For example, the weight calculation unit is transmission weight processing unit 130 in the embodiment.

The weight calculation unit calculates weight vectors for spatial multiplexing transmission to multiple terminal devices based on a channel matrix that lists channel vectors generated from channel information between the antenna elements of the terminal devices or virtual antenna elements obtained by combining these antenna elements and the antenna elements of the base station device, and additional channel vectors associated with antenna elements of the terminal devices that are different from the channel vectors.

The weight calculation unit selects, as an additional channel vector for the terminal device, a second channel vector that has a low degree of mutual similarity with the first channel vector from among one or more second channel vectors acquired before the first channel vector, which is the channel vector of the terminal device, and calculates a weight vector.

In the above base station device, the weight calculation unit may calculate the mutual similarity between the first channel vector and the second channel vector based on the absolute value of the dot product of the first channel vector and the second channel vector, or based on a cross-correlation value obtained by dividing the absolute value of the dot product by the product of the norm of the first channel vector and the norm of the second channel vector. For example, the first channel vector is the current channel vector h _i (t) of the terminal device #i in the embodiment, the second channel vector is the past channel vector h _i (t-nT) of the terminal device #i in the embodiment, the absolute value of the dot product is the absolute value of the correlation coefficient γ _i (n) expressed by equation (12) in the embodiment, and the cross-correlation value is the correlation coefficient γ _i (n) expressed by equation (13) in the embodiment.

In the above base station device, the weight calculation unit may calculate the mutual similarity between the first channel vector and the second channel vector based on an approximate correlation value obtained by dividing the absolute value of the inner product of the first channel vector and the second channel vector by the norm of either the first channel vector or the second channel vector or by the square of the norm of either the first channel vector or the second channel vector. For example, the approximate correlation value is the correlation coefficient γ _i (n) expressed by equation (14) or equation (15) in the embodiment.

In the above base station device, the weight calculation unit may calculate the mutual similarity between the first channel vector and the second channel vector based on the norm of a difference vector between the first channel vector and the second channel vector, or based on a value obtained by dividing the norm of the difference vector by the norm of either the first channel vector or the second channel vector. For example, the difference vector is the difference vector d _i (n) expressed by equation (16) in the embodiment.

In the above-mentioned base station device, the weight calculation unit may determine the weight vector by excluding weight vectors calculated before the time when mutual similarities higher than a predetermined value appear more than a predetermined number of times.

In addition, the above-mentioned processing may be performed by recording a program for realizing the functions of the base station device 80 in this embodiment on a computer-readable recording medium, and then loading and executing the program recorded on this recording medium into a computer system. Note that the term "computer system" as used here includes hardware such as an OS and peripheral devices. It also includes a WWW system equipped with a homepage provision environment (or display environment). Furthermore, "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, as well as storage devices such as hard disks built into computer systems. Furthermore, "computer-readable recording medium" also includes devices that retain a program for a certain period of time, such as volatile memory (RAM) within a computer system that acts as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line.

The above program may also be transmitted from a computer system that stores the program in a storage device or the like to another computer system via a transmission medium, or by transmission waves within the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The above program may also be one that realizes part of the above-mentioned functions. Furthermore, it may be a so-called differential file (differential program) that can realize the above-mentioned functions in combination with a program already recorded on the computer system.

It can be used in wireless communication systems in which a base station device with multiple antenna elements communicates with multiple terminal devices using the same frequency channel.

80...base station device, 81...transmitter, 85...receiver, 87...interface circuit, 88...MAC layer processing circuit, 130...transmission weight processing unit, 132...channel information storage circuit, 133...MU-MIMO transmission weight calculation circuit, 135...channel similarity calculation circuit, 136...channel information selection circuit, 801...base station device, 802, 802-1 to 802-k, 802-L...terminal device, 811-1, 811-L...transmission signal processing circuit, 812-1 to 812-K...addition and synthesis circuit, 813-1 to 813-K...GI assignment circuit, 814-1 to 814-K...D/A converter, 815...local oscillator, 816-1 to 816-K...mixer, 817-1 to 817-K...filter, 818-1 to 818-K...high power amplifier, 8 19-1 to 819-K...antenna elements, 820...communication control circuit, 830...transmission weight processing circuit, 831...channel information acquisition circuit, 832...channel information storage circuit, 833...multi-user MIMO transmission weight calculation circuit, 851-1 to 851-K...antenna elements, 852-1 to 852-K...low-noise amplifier, 853...local oscillator, 854-1 to 854-K...mixer, 855-1 to 855-K...filter, 856-1 to 856-K...A/D converter, 857-1 to 857-K...FFT circuit, 858-1 to 858-L...received signal processing circuit, 860...received weight processing circuit, 861...channel information estimation circuit, 862...multi-user MIMO receive weight calculation circuit, 881...scheduling processing circuit

Claims (7)

A base station apparatus in a wireless communication system including a base station apparatus having a plurality of antenna elements and a plurality of terminal apparatuses, wherein the base station apparatus and the terminal apparatuses can perform spatial multiplexing transmission on the same frequency at the same time,
a weight calculation unit that calculates weight vectors for performing spatial multiplexing transmission for the plurality of terminal devices based on a channel matrix in which first channel vectors for the plurality of terminal devices, which are generated based on channel information between the antenna element of the terminal device or a virtual antenna element obtained by combining the antenna elements, and the antenna element provided by the base station device, and second channel vectors for the plurality of terminal devices, which are associated with the antenna element of the terminal device different from the first channel vectors, are arranged;
A base station device in which the weight calculation unit uses , as the second channel vector for each of the multiple terminal devices arranged in the channel matrix, one or more second channel vectors that are channel vectors of a terminal device included in the multiple terminal devices and that were acquired before the first channel vector of the terminal device, and the second channel vector has a low mutual similarity with the first channel vector. 2. The base station device according to claim 1, wherein the weight calculation unit calculates the mutual similarity between the first channel vector and the second channel vector based on the absolute value of the dot product between the first channel vector and the second channel vector, or based on a cross-correlation value obtained by dividing the absolute value of the dot product by the product of the norm of the first channel vector and the norm of the second channel vector. 2. The base station device according to claim 1, wherein the weight calculation unit calculates the mutual similarity between the first channel vector and the second channel vector based on an approximate correlation value obtained by dividing the absolute value of the inner product of the first channel vector and the second channel vector by the norm of either the first channel vector or the second channel vector or by the square of the norm of either the first channel vector or the second channel vector. 2. The base station device according to claim 1, wherein the weight calculation unit calculates the mutual similarity between the first channel vector and the second channel vector based on a norm of a difference vector between the first channel vector and the second channel vector, or based on a value obtained by dividing the norm of the difference vector by the norm of either the first channel vector or the second channel vector. 5. The base station device according to claim 1, wherein the weight calculation unit calculates the weight vector based on the channel matrix excluding the second channel vector acquired before a point in time when the mutual similarity having a value higher than a predetermined value appears more than a predetermined number of times. A weight generation method executed by a base station device in a wireless communication system including a base station device having a plurality of antenna elements and a plurality of terminal devices, wherein the base station device and the terminal devices can perform spatial multiplexing transmission on the same frequency at the same time, the weight generation method comprising:
a step of calculating weight vectors for performing spatial multiplexing transmission for the plurality of terminal devices based on a channel matrix in which first channel vectors for the plurality of terminal devices, which are generated based on channel information between the antenna elements of the terminal devices or virtual antenna elements obtained by combining the antenna elements, and the antenna elements of the base station device, and second channel vectors for the plurality of terminal devices, which are associated with the antenna elements of the terminal devices different from the first channel vectors, are arranged;
As the second channel vector of each of the plurality of terminal devices arranged in the channel matrix , one or more of the second channel vectors , which are channel vectors of a terminal device included in the plurality of terminal devices and acquired before the first channel vector of the terminal device, are used, and the second channel vector has a low degree of similarity to the first channel vector;
A weight generation method having the following. A wireless communication system comprising a base station device having a plurality of antenna elements and a plurality of terminal devices, wherein the base station device and the terminal devices can perform spatial multiplexing transmission at the same time on the same frequency,
The base station device
a weight calculation unit that calculates weight vectors for performing spatial multiplexing transmission for the plurality of terminal devices based on a channel matrix in which first channel vectors for the plurality of terminal devices, which are generated based on channel information between the antenna element of the terminal device or a virtual antenna element obtained by combining the antenna elements, and the antenna element provided by the base station device, and second channel vectors for the plurality of terminal devices, which are associated with the antenna element of the terminal device different from the first channel vectors, are arranged;
The weight calculation unit uses, as the second channel vector of each of the plurality of terminal devices arranged in the channel matrix , one or more second channel vectors that are channel vectors of a terminal device included in the plurality of terminal devices and that were acquired before the first channel vector of the terminal device, and that second channel vector has a low degree of mutual similarity with the first channel vector.