Access point control device, communication system and communication method

Abstract

To improve spectral efficiency and user fairness in a cell-free MIMO system.SOLUTION: An access point control device according to an embodiment includes a computing device that controls a plurality of access points that perform wireless communication with a plurality of terminals, and an interface connected to the plurality of access points. The computing device calculates a channel matrix representing each state of a plurality of communication channels through which uplink communication and / or downlink communication is performed between the plurality of access points and the plurality of terminals, calculates a first matrix, a second matrix, and a third matrix to decompose the channel matrix into the product of the first, second, and third matrices, calculates a transmit beamformer matrix and a receive beamformer matrix on the basis of the first matrix and the third matrix, and prevents divergence of the second matrix by using a divergence prevention parameter to calculate the second matrix.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to an access point control device, a communication system using this access point control device, and a communication method used in this communication system. [Background technology]

[0002] In recent years, cell-free MIMO (Multiple-Input and Multiple-Output) systems have been attracting attention in order to realize the large-capacity communications required for the so-called "5G+" which is an advanced version of the 5th generation mobile communications system, and the 6th generation mobile communications system also referred to as "6G". Cell-free MIMO makes it possible to utilize spatial wireless resources to the maximum extent possible by destroying the existing cell structure. In addition, "5G+" and "6G" are required to provide flexible mobile communications services that can be applied to environments where various communication requirements coexist.

[0003] However, no effective study has been conducted yet on the cell-free MIMO system in an environment where various communication requirements coexist, i.e., an environment in which each user terminal has a different number of antennas, due to the complexity of the communication path structure.

[0004] In relation to the above, Patent Document 1 (JP Patent Publication 2016-521482A) discloses a method for designing sparse transmit beamforming for a network multiple-input multiple-output (MIMO) system. The method includes a step in which a cloud central processing unit dynamically forms a cluster of transmission points (TPs) for use in transmit beamforming for each of a plurality of user equipments (UEs) in the system by optimizing a network utility function and system resources. The method further includes a step in which the cloud central processing unit determines a sparse beamforming vector for each UE according to the optimization. The method further includes a step in which the cloud central processing unit transmits a message and a first beamforming coefficient to each TP in the formed cluster associated with a first UE in the plurality of UEs, where each TP in the formed cluster associated with the first UE corresponds to a non-zero entry in the first beamforming vector corresponding to the first UE.

[0005] Also, Patent Document 2 (JP 2020-188462 A) discloses a method for providing an adaptive beamforming antenna for an OFDM-based communication system. The method includes forming a cyclic prefix value matrix (A) and a tail value matrix (B) from a received orthogonal frequency division multiplexing (OFDM) symbol, and forming an addition matrix (S) and a difference matrix (D) from the matrix (A) and the matrix (B). The method further includes a multiplication step of multiplying a beamformer preset matrix (W) by the addition matrix (S) and the difference matrix (D) to determine a matrix (P) and a matrix (Q), and a step of determining a beam identifier from the matrix (P) and the matrix (Q).

[0006] In addition, Non-Patent Document 1 (TL Marzetta, "Noncooperative cellular wireless with unlimited numbers of base station antennas", IEEE Trans. Wireless Commun., vol. 9, no. 11, pp. 3590-3600, 2010, doi: 10.1109 / TWC.2010.092810.091092.) describes massive MIMO technology.

[0007] Also, Non-Patent Document 2 (HQ Ngo, A. Ashikhmin, H. Yang, EG Larsson, and TL Marzetta, “Cell-free massive MIMO versus small cells”, IEEE Trans. Wireless Commun., vol. 16, no. 3, pp. 1834-1850, 2017, doi: 10.1109 / TWC.2017.2655515.) describes cell-free MIMO.

[0008] In addition, Non-Patent Document 3 (AJ Goldsmith and PP Varaiya, "Capacity of fading channels with channel side information", IEEE Trans. Inform. Theory, vol. 43, no. 6, pp. 1986-1992, 1997, doi: 10.1109 / 18.641562.) shows that in point-to-point MIMO communications in which channel information is known at the transmitter and receiver, beamformer design based on singular value decomposition using water filling is optimal.

[0009] In addition, Non-Patent Document 4 (D. Senaratne and C. Tellambura, "GSVD beamforming for two-user MIMO downlink channel", IEEE Trans. Veh. Technol., vol. 62, no. 6, pp. 2596-2606, 2013, doi: 10.1109 / TVT.2013.2241091.) shows that for one-to-two communications, a beamformer design utilizing generalized singular value decomposition, which expands the definition of singular value decomposition from a single matrix to two matrices, is effective.

[0010] In addition, non-patent document 5 (L. Khamidullina, ALF de Almeida, and M. Haardt, "Multilinear generalized singular value decomposition (ml-gsvd) with application to coordinated beamforming in multi-user MIMO systems", Proc. IEEE ICASSP, Barcelona, ​​Spain, 2020, pp. 4587-4591, doi: 10.1109 / ICASSP40776.2020.9053691.) discloses a computational method for expanding the definition of generalized singular value decomposition to high-dimensional tensors in one-to-many communications.

[0011] Furthermore, Non-Patent Document 6 (ITU, "Guidelines for evaluation of radio interface technologies for IMT-advanced", ITU, M Series Mobile, radiodetermination, amateur and related satellites services Report ITU-R M.2135-1, December 2009) shows propagation attenuation related to urban macro cells.

[0012] In addition, non-patent document 7 (H. Bolcskei, M. Borgmann, and AJ Paulraj, "Impact of the propagation environment on the performance of space-frequency coded MIMO-OFDM", IEEE J. Sel. Areas Commun., vol. 21, no. 3, pp. 427-439, 2003, issn: 1558-0008, doi: 10.1109 / JSAC.2003.809723.) shows a spatial correlation matrix between an access point and a terminal.

[0013] In addition, Non-Patent Document 8 (PH Schoenemann, "A generalized solution of the orthogonal procrustes problem", Psychometrika, vol. 31, no. 1, 1966) shows an update formula in a closed form expression of individual matrices that can be used in ML-GSVD.

[0014] Also, Non-Patent Document 9 (W. Rhee and J.M. Cioffi, “Increase in capacity of multiuser OFDM system using dynamic subchannel allocation”, Proc. IEEE VTC-Spring, Tokyo, Japan, vol. 2, 2000, pp. 1085-1089, doi: 10.1109 / VETECS.2000.851292.) describes Max Min resource allocation. [Prior art documents] [Patent documents]

[0015] [Patent Document 1] Special Publication No. 2016-521482 [Patent Document 2] JP 2020-188462 A [Non-patent literature]

[0016] [Non-Patent Document 1] By T. L. Marzetta, "Noncooperative cellular wireless with unlimited numbers of base station antennas", IEEE Trans. Wireless Commun., vol. 9, no. 11, pp. 3590 - 3600, 2010, doi: 10.1109 / TWC.2010.092810.091092. [Non-Patent Document 2] By H. Q. Ngo, A. Ashikhmin, H. Yang, E. G. Larsson, and T. L. Marzetta, "Cell-free massive MIMO versus small cells", IEEE Trans. Wireless Commun., vol. 16, no. 3, pp. 1834 - 1850, 2017, doi: 10.1109 / TWC.2017.2655515. [Non-Patent Document 3] By A. J. Goldsmith and P. P. Varaiya, "Capacity of fading channels with channel side information", IEEE Trans. Inform. Theory, vol. 43, no. 6, pp. 1986 - 1992, 1997, doi: 10.1109 / 18.641562. [Non-Patent Document 4] By D. Senaratne and C. Tellambura, "GSVD beamforming for two-user MIMO downlink channel", IEEE Trans. Veh. Technol., vol. 62, no. 6, pp. 2596 - 2606, 2013, doi: 10.1109 / TVT.2013.2241091. [Non-Patent Document 5] L. Khamidullina, A. L. F. de Almeida, and M. Haardt, "Multilinear generalized singular value decomposition (ml-gsvd) with application to coordinated beamforming in multi-user MIMO systems", Proc. IEEE ICASSP, Barcelona, Spain, 2020, pp. 4587-4591, doi: 10.1109 / ICASSP40776.2020.9053691. [Non-Patent Document 6] ITU, "Guidelines for evaluation of radio interface technologies for IMT-advanced", ITU, M Series Mobile, radiodetermination, amateur and related satellites services Report ITU-R M.2135-1, December 2009 [Non-Patent Document 7] H. Bolcskei, M. Borgmann, and A. J. Paulraj, "Impact of the propagation environment on the performance of space-frequency coded mimo-ofdm", IEEE J. Sel. Areas Commun., vol. 21, no. 3, pp. 427-439, 2003, issn: 1558-0008, doi: 10.1109 / JSAC.2003.809723. [Non-Patent Document 8] P. H. Schoenemann, "A generalized solution of the orthogonal procrustes problem", Psychometrika, vol. 31, no. 1, 1966 [Non-Patent Document 9] W. Rhee and JM Cioffi, "Increase in capacity of multiuser OFDM system using dynamic subchannel allocation", Proc. IEEE VTC-Spring, Tokyo, Japan, vol. 2, 2000, pp. 1085~1089, doi: 10.1109 / VETECS.2000.851292. Summary of the Invention [Problem to be solved by the invention]

[0017] In a cell-free MIMO system, the present invention improves frequency utilization efficiency and fairness to users. Other objects and novel features will become apparent from the description of this specification and the accompanying drawings. [Means for solving the problem]

[0018] The means for solving the problems will be explained below using the numbers used in the (Mode for carrying out the invention). These numbers are added to clarify the correspondence between the description in the (Claims) and the (Mode for carrying out the invention). However, these numbers should not be used to interpret the technical scope of the invention described in the (Claims).

[0019] According to one embodiment, the access point control device (2) includes a calculation device (22) and an interface (24). The calculation device (22) controls a plurality of access points (3) that perform wireless communication of a cell-free MIMO system with a plurality of terminals (5). The interface (24) is connected to the plurality of access points (3). The calculation device (22) calculates a communication channel matrix (H kThe calculation device (22) calculates the channel matrix (H k ) into the first matrix (B k ), the first matrix (B) to decompose it into the product of the second matrix (C) and the third matrix (A) k ), the second matrix (C) and the third matrix (A). The calculation device (22) calculates the first matrix (B k ) and the third matrix (A), a transmit beamformer matrix (U k , U k u ) and the receive beamformer matrix (V k , V k u The calculation device (22) prevents the second matrix (C) from diverging by using a divergence prevention parameter (ε) to calculate the second matrix (C).

[0020] According to one embodiment, a communication system (1) includes an access point control device (2), a plurality of terminals (5), and a plurality of access points (3). The access point control device (2) includes a calculation device (22) and an interface (24). The calculation device (22) controls a plurality of access points (3) that perform wireless communication of a cell-free MIMO system with the plurality of terminals (5). The interface (24) is connected to the plurality of access points (3). The calculation device (22) calculates a communication channel matrix (H k The calculation device (22) calculates the channel matrix (H k ) into the first matrix (B k ), the first matrix (B) to decompose it into the product of the second matrix (C) and the third matrix (A) k ), the second matrix (C) and the third matrix (A). The calculation device (22) calculates the first matrix (B k ) and the third matrix (A), a transmit beamformer matrix (Uk , U k u ) and the receive beamformer matrix (V k , V k u The calculation device (22) prevents the second matrix (C) from diverging by using a divergence prevention parameter (ε) to calculate the second matrix (C).

[0021] According to one embodiment, the communication method includes performing wireless communication in a communication system (1) including an access point control device (2) including a calculation device (22) that controls a plurality of access points (3) that perform wireless communication between a plurality of terminals (5) in a cell-free MIMO system, a plurality of terminals (5), and a plurality of access points (3). The communication method includes: calculating a communication channel matrix (H k The communication method includes calculating the channel matrix (H k ) into the first matrix (B k ), the first matrix (B) to decompose it into the product of the second matrix (C) and the third matrix (A) k ), a second matrix (C) and a third matrix (A) (S11, S12, S13, S14). k ) and the third matrix (A), a transmit beamformer matrix (U k , U k u ) and the receive beamformer matrix (V k , V k u The communication method further includes calculating (S16) the second matrix (C). The communication method further includes preventing divergence of the second matrix (C) by using (S13) an anti-divergence parameter (ε) to calculate the second matrix (C). Effect of the Invention

[0022] According to one embodiment, it is possible to improve frequency utilization efficiency and fairness among users in a cell-free MIMO system. [Brief description of the drawings]

[0023] [Figure 1] FIG. 1 is a diagram showing an example of a configuration of a communication system according to an embodiment. [Diagram 2] FIG. 2 is a block circuit diagram showing an example of a configuration of an access point control device according to an embodiment. [Diagram 3] FIG. 3 is a block circuit diagram showing an example of a configuration of an access point according to an embodiment. [Figure 4] FIG. 4 is a block circuit diagram showing an example of a configuration of a terminal according to an embodiment. [Figure 5A] FIG. 5A is a diagram illustrating an outline of the calculation content of ML-GSVD in downlink communication according to one embodiment. [Figure 5B] FIG. 5B is a diagram illustrating an outline of the calculation content of ML-GSVD in uplink communication according to one embodiment. [Figure 6] FIG. 6 is a flowchart showing an example of a configuration of a communication method according to an embodiment. [Figure 7A] FIG. 7A is a diagram illustrating a schematic diagram of calculation contents in downlink communication of a beamforming method according to one embodiment. [Figure 7B] FIG. 7B is a diagram illustrating a schematic diagram of calculation contents in uplink communication of a beamforming method according to one embodiment. [Figure 8A] FIG. 8A is a flow chart illustrating an example of a configuration of a communication method according to an embodiment. [Figure 8B] FIG. 8B is a flow chart illustrating another exemplary configuration of a communication method according to an embodiment. [Figure 9] FIG. 9 is a table showing an example of simulation parameters used when simulating a communication method according to an embodiment. [Figure 10] FIG. 10 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 11] FIG. 11 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 12] FIG. 12 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 13] FIG. 13 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 14] FIG. 14 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 15] FIG. 15 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 16] FIG. 16 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 17] FIG. 17 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 18] FIG. 18 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 19] FIG. 19 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 20] FIG. 20 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 21] FIG. 21 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 22] FIG. 22 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Diagram 23] FIG. 23 is a graph showing an example of a simulation result of a communication method according to an embodiment. [Figure 24]FIG. 24 is a graph showing an example of a simulation result of a communication method according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An access point control device, a communication system, and a communication method according to the present invention will be described below with reference to the accompanying drawings.

[0025] (Embodiment) 1, a communication system 1 according to an embodiment includes an access point control device 2, a plurality of access points 3, a plurality of fronthauls 4, and a plurality of terminals 5. The communication system 1 is arranged in a square area having a side length D.

[0026] As an example, the multiple access points 3 may be arranged in a grid. The multiple terminals 5 may be mobile terminals such as smartphones, in-vehicle terminals mounted on automobiles, or wireless communication terminals mounted on moving objects such as robots and drones.

[0027] Each terminal 5 performs wireless communication with at least one of the multiple access points 3. This wireless communication includes uplink communication in which a signal is transmitted from the terminal 5 to the access point 3, and downlink communication in which a signal is transmitted from the access point 3 to the terminal 5. This wireless communication is based on a so-called cell-free MIMO (Multiple-Input and Multiple-Output) system, and by destroying the existing cell structure, it is possible to utilize spatial wireless resources to the maximum extent possible.

[0028] The multiple access points 3 perform wired communication with the access point control device 2 via a fronthaul 4. The access point control device 2 controls the operations of the multiple access points 3.

[0029] As shown in Fig. 2, the access point control device 2 according to one embodiment includes a bus 20, an antenna 21, a computing device 22, a storage device 23, and an interface 24. The antenna 21, the computing device 22, the storage device 23, and the interface 24 are connected to each other via the bus 20 so as to be able to communicate with each other. The antenna 21 is optional. The storage device 23 stores a control program 231. The control program 231 may be received from the outside via the interface 24 and stored in the storage device 23, or may be read from the recording medium 230 and stored in the storage device 23. The recording medium 230 may be non-transitory and tangible.

[0030] The access point control device 2 may include a so-called computer. That is, at least a part of the functions of the access point control device 2 may be realized by the arithmetic device 22 executing the control program 231.

[0031] As shown in FIG. 3, the access point 3 according to the embodiment includes a bus 30, an antenna 31, a computing device 32, a storage device 33, and an interface 34. The antenna 31, the computing device 32, the storage device 33, and the interface 34 are connected to each other via the bus 30 so as to be able to communicate with each other. The antenna 31 may be an array antenna having a plurality of antenna elements. The storage device 33 stores a communication program 331. The communication program 331 may be received from the outside via the antenna 31 or the interface 34 and stored in the storage device 33, or may be read from the recording medium 330 and stored in the storage device 33. The recording medium 330 may be non-transitory and tangible.

[0032] The access point 3 may include a so-called computer. That is, at least a part of the functions of the access point 3 may be realized by the computing device 32 executing the communication program 331.

[0033] As shown in FIG. 4, the terminal 5 according to the embodiment includes a bus 50, an antenna 51, a computing device 52, a storage device 53, and an interface 54. The antenna 51, the computing device 52, the storage device 53, and the interface 54 are connected to each other via the bus 50 so as to be able to communicate with each other. The antenna 51 may be an array antenna having a plurality of antenna elements. The storage device 53 stores a communication program 531. The communication program 531 may be received from the outside via the antenna 51 or the interface 54 and stored in the storage device 53, or may be read from the recording medium 530 and stored in the storage device 53. The recording medium 530 may be non-transitory and tangible.

[0034] The terminal 5 may include a so-called computer. That is, at least a part of the functions of the terminal 5 may be realized by the computing device 52 executing the communication program 531.

[0035] (System Model) A communication channel model according to an embodiment will be described below. In a communication system 1 according to an embodiment, a communication channel between an access point 3 and a terminal 5 is modeled as a communication channel model defined by the following "Mathematical Expression 1".

number

[0036] In the formula "1", "H l,k ” is a channel matrix representing the state of the communication channel between the l-th access point 3 and the k-th terminal 5, and M k It is a matrix consisting of complex numbers with N rows and N columns. "l" is the number of the access point 3 and is an integer from 1 to L. "L" is the total number of access points 3 included in the communication system 1. "k" is the number of the terminal 5 and is an integer from 1 to K. "K" is the total number of terminals 5 included in the communication system 1. "M k " is the total number of antenna elements that the kth terminal 5 has. "δ(d l,k) represents the propagation attenuation between the lth access point 3 and the kth terminal 5, and is approximated, for example, by the following "Equation 2".

number

[0037] In the formula "1", "R l,k " is M k Row M k It is a matrix composed of complex numbers in the columns, is a spatial correlation matrix between the l-th access point 3 and the k-th terminal 5, and is defined as in the following "Equation 3".

number

[0038] In the formula "1", "G l,k " is a fading coefficient matrix between the l-th access point 3 and the k-th terminal 5, and satisfies the following "Mathematical Expression 4".

number

[0039] In the formula "1", "T l,k " is a matrix composed of N rows and N columns of complex numbers, and is the spatial correlation matrix between the l-th access point 3 and the k-th terminal 5, and is defined as shown in the following "Equation 5".

number

[0040] In "Formula 1" and "Formula 2", "d l,k " represents the distance between the l-th access point 3 and the k-th terminal 5. In "Equation 2", "f c " represents the carrier frequency.

[0041] In "Equation 3" and "Equation 5", "θ l,k t " represents the emission angle between the l-th access point 3 and the k-th terminal 5. Also, "θ l,k r " represents the angle of incidence between the l-th access point 3 and the k-th terminal 5. "θ l,k t " and "θ l,k r " angle "θ l,k In "Equation 3" and "Equation 5", "a(θ l,k )" represents the array response vector in a uniform linear array antenna, and the angle "θ l,k " is defined as in the following "Equation 6".

number

[0042] A baseband complex received signal in downlink communication in one embodiment will now be described. In one embodiment, the received signal received by the k-th terminal 5 from the access point 3 is expressed by the following "Equation 7".

number

[0043] In the formula "7", "y k " is Q k It is a matrix consisting of complex numbers with one row and one column, and represents the received signal received by the kth terminal 5, and is expressed as the sum of the desired signal of the kth terminal 5, the inter-user interference, and colored noise. k " represents the number of transmission streams corresponding to the kth terminal 5.

[0044] In the formula "7", "U k " is Q k Row M k It is a matrix composed of complex numbers of columns, and is a receiving beamformer matrix for forming a receiving beam for reception by the kth terminal 5. k " is M k It is a matrix composed of complex numbers with rows and columns LN, and is a channel matrix between all access points 3 and the k-th terminal 5, and is expressed as the following "Equation 8".

number

[0045] In the formula "7", "V k " is the LN row Q k The matrix is ​​a complex matrix of columns, and is a transmit beamformer matrix for forming a transmit beam transmitted by the k-th terminal 5. k " is Q k It is a matrix consisting of complex numbers with one row and one column, and represents a transmission signal to the kth terminal 5. k " represents AWGN (Additive White Gaussian Noise) of the kth terminal 5 and satisfies the following "Equation 9".

number

[0046] The spectral efficiency of a downlink signal according to an embodiment will now be described. The spectral efficiency of a downlink signal is expressed by the following formula (10).

number

[0047] In the formula "10", "η k " represents the frequency utilization efficiency of the kth terminal 5. "E k " is defined as in the following "Equation 11".

number

[0048] In the formulas "10" and "11", "F k TX " is defined as the following "Formula 12". Also, "F k RX " is defined as in the following "Equation 13".

number

number

[0049] An uplink baseband complex received signal in one embodiment will now be described. In one embodiment, the received signal received by the access point 3 from the k-th terminal 5 is expressed by the following formula 14.

number

[0050] In the formula "14", "y u" is a matrix consisting of LN rows and 1 column of complex numbers, and represents the signals received by all the access point control devices 2 from the terminals 5. k u " is LN row M k It is a matrix composed of complex numbers in the columns, represents a channel matrix between all access points 3 and the kth terminal 5, and is defined as shown in the following "Equation 15".

number

[0051] In the formula (14), "V k u " is M k row Q k The matrix is ​​a complex matrix of columns, and is a transmit beamformer matrix for forming a transmit beam transmitted by the k-th terminal 5. k " is Q k It is a matrix consisting of complex numbers with one row and one column, and represents the transmission signal transmitted by the kth terminal 5. u " is the AWGN that the access point control device 2 receives from the terminal 5, and satisfies the following "Equation 16".

number

[0052] The transmission signal from the k-th terminal 5 can be estimated as shown in the following formula (17).

number

[0053] In the formula (17), "s k u ” (specifically, with a hat over the “s”) is Q k It is a matrix consisting of complex numbers with one row and one column, and represents the estimated transmission signal from the kth terminal 5. k u " is Q kIt is a matrix consisting of complex numbers with rows and columns LN, and is a receiving beamformer matrix corresponding to the kth terminal 5. u " is a matrix made up of LN rows and 1 column of complex numbers, and represents a signal received by the access point control device 2 from the terminal 5.

[0054] Based on "Formula 17" and "Formula 7", the following "Formula 18" is obtained.

number

[0055] In the formula (18), the estimated transmission signal s k u (To be precise, there is a hat symbol above the "s") is expressed as the sum of the signal of the kth terminal 5, the inter-user interference, and colored noise. "H k u " is LN row M k It is a matrix composed of complex numbers in the column, and represents the communication channel matrix between all the access points 3 and the k-th terminal 5. k u " is M k row Q k is a matrix composed of complex numbers in the k-th column, and is the transmit beamformer matrix of the k-th terminal 5. k u " is Q k It is a matrix consisting of one row and one column of complex numbers, and represents the transmission signal transmitted by the kth terminal 5 .

[0056] The spectral efficiency of the uplink signal according to one embodiment will be described below. The spectral efficiency of the uplink signal is expressed by the following formula (19).

number

[0057] In the formula "19", "η k u " represents the uplink frequency utilization efficiency of the kth terminal 5. "E ku " is defined as in the following equation "Number 20".

number

[0058] In "Equation 19" and "Equation 20", "F k UTX " is defined as the following "Equation 21". Also, in "Equation 20", "F URX " is defined as in the following "Number 22".

number

number

[0059] In one embodiment, based on the system model of the communication channel described above, a calculation called MultiLinear Generalized Singular Value Decomposition (ML-GSVD) is performed.

[0060] A description will be given of downlink communication in improved ML-GSVD according to one embodiment. In this embodiment, the channel matrix H k is decomposed into a product of three matrices as shown in the following "Number 23".

number

[0061] In the formula "23", "B k " is M k A matrix consisting of complex numbers with rows and columns LN, which is a separate matrix for the dimensions of each terminal 5, and the separate matrix B k "diag{C(k,:)}" is a matrix of LN rows and LN columns of real numbers, and is called the scaling matrix C k"A" is a matrix consisting of complex numbers with LN rows and LN columns, and is a common matrix for all communication channel matrices, and is called the common matrix A.

[0062] FIG. 5A is a diagram illustrating the outline of the calculation contents of ML-GSVD in downlink communication according to one embodiment. In the example of FIG. 5A, LN=6, M1=1, M2=3, and M3=2, and the downlink channel tensor H corresponding to all terminals 5 is shown divided into H1 corresponding to the first terminal 5, H2 corresponding to the second terminal 5, and H3 corresponding to the third terminal 5. Similarly, the individual tensor B is shown divided into B1 corresponding to the first terminal 5, B2 corresponding to the second terminal 5, and B3 corresponding to the third terminal 5. In addition, the scaling matrix C is shown divided into C1 corresponding to the first terminal 5, C2 corresponding to the second terminal 5, and C3 corresponding to the third terminal 5, and elements other than the shaded parts are 0.

[0063] From "Equation 7" and "Equation 23", the following "Equation 24" can be obtained.

number

[0064] In the formula "24", "C k " is defined as in the following equation "Number 25".

number

[0065] An embodiment of the ML-GSVD in uplink communication will be described. In the embodiment, the channel matrix H k is decomposed into a product of three matrices as shown in the following "Number 26".

number

[0066] 5B is a diagram illustrating the outline of the calculation contents of ML-GSVD in uplink communication according to one embodiment. In the example of FIG. 5B, LN=6, M1=1, M2=3, and M3=2, and the uplink channel tensor H u H1 corresponding to the first terminal 5 u and H2 corresponding to the second terminal 5 u and H3, which corresponds to the third terminal 5 u Similarly, the transpose of an individual tensor B is T B1 corresponding to the first terminal 5 T and B2 corresponding to the second terminal 5 T and B3, which corresponds to the third terminal 5 T In addition, the scaling matrix C is shown divided into C1 corresponding to the first terminal 5, C2 corresponding to the second terminal 5, and C3 corresponding to the third terminal 5, and elements other than the shaded parts are 0.

[0067] From "Equation 18" and "Equation 26", the following "Equation 27" is obtained.

number

[0068] A method for calculating the ML-GSVD according to one embodiment will be described with reference to the flowchart of Fig. 6. The calculation of the ML-GSVD according to one embodiment may be realized by the calculation device 22 of the access point control device 2 executing the control program 231.

[0069] 6 starts, step S11 is executed. In step S11, the calculation device 22 calculates the initial values ​​of the common matrix A and the scaling matrix C.

[0070] The initial value of the common matrix A is obtained as a left singular matrix of the following formula (28).

number

[0071] The initial value of the scaling matrix C is obtained as a random variable whose elements follow a uniform distribution in the interval [0,1].

[0072] After step S11, step S12 is executed. In step S12, the calculation device 22 calculates an individual matrix B based on the common matrix A and the scaling matrix C. k When step S12 is executed for the first time, the initial value of the individual matrix Bk is calculated based on the initial value of the common matrix A and the initial value of the scaling matrix C. Here, the communication channel matrix H k The channel tensor H, which is an integration of the above, is defined as shown in the following equation (29).

number

[0073] Individual matrix B k In order to optimize the update of k is selected based on the following formula (30).

number

[0074] Individual matrix "B" of "Number 30" k " satisfies the following "Number 31".

number

[0075] The decomposition error of "Equation 30" RBk " is defined as in the following equation "32".

number

[0076] In the formula "32", "H k ” (Actually, “H” is a script character) represents the component corresponding to the kth terminal 5 in the channel tensor H before decomposition, and “B k C k A T " That is, "B k H k ” (Actually, “H” is a script letter, and there is a tilde symbol above “H”) represents the component of the decomposed channel tensor H corresponding to the k-th terminal 5, and the decomposition error “E RBk " is LN row M k It is a matrix composed of complex numbers in the columns, and represents the difference between the channel tensor H before and after decomposition.

[0077] Individual matrix B k The Lagrangian for updating is given by the following equation (33).

number

[0078] In Equation 33, "L" (to be precise, "L" is a script letter) is the Lagrangian. RBk " can be expressed as "Number 34" below based on "Number 32".

number

[0079] In the formula "33", "L m " is M k Row M k A matrix composed of complex numbers in the columns, which are the Lagrange multipliers.

[0080] In addition, the individual matrix B kAs an update equation in a closed form expression of the above, the following “Equation 35” is described in Non-Patent Document 8 (PH Schoenemann, “A generalized solution of the orthogonal procrustes problem”, Psychometrika, vol. 31, no. 1, 1966).

number

[0081] After step S12, step S13 is executed. In step S13, the arithmetic unit 22 performs a calculation based on the following "Formula 36" when updating the scaling matrix C, and then performs a flipping operation for normalization using the divergence prevention parameter ε and for adjusting the number of dimensions to realize the update of the scaling matrix C.

number

[0082] In the formula (36), the operator "◇" represents the KR (Khatri-Rao) product. In addition, in order to compensate for the effect of normalization of the scaling matrix C, "A(:,l n ) is updated as shown in the following formula (37).

number

[0083] In the formula "37", l n The th column "C(:,l n )”, that is, the l of the normalized scaling matrix C n The sequence is expressed as in the following formula (38).

number

[0084] The "ε" in the denominator of "Equation 38" is a divergence prevention parameter that prevents divergence during normalization. By using this divergence prevention parameter ε and the "max" function in the denominator of "Equation 38", n The th column "C(:,l n In addition, the inventors have confirmed through simulations described below that by setting the divergence prevention parameter ε to an appropriate value that is greater than 0 and sufficiently smaller than 1, and combining it with a flipping operation described below, the improved ML-GSVD according to one embodiment can achieve better performance than the ML-GSVD according to the related art.

[0085] The flipping operation of the scaling matrix C in the improved ML-GSVD according to one embodiment will be described. In the improved ML-GSVD according to one embodiment, the channel matrix H k To express this, a dimension equal to the number of non-zero elements in the k-th row of the scaling matrix C is used. According to the dimension theorem, if there is an excess or deficiency in the number of non-zero elements in the k-th row of the scaling matrix C, this must be eliminated; this operation is called a flipping operation and will be described in detail later.

[0086] In the k-th row C(k,:) of the scaling matrix C, k The set of indices of the smallest elements is called Z k " The elements C(k,l) of the scaling matrix C are n ), the index l n Set Z k If the element C(k,l) is included in the formula (39), more nonzero elements than necessary will be included. n This will be rectified by updating the .

number

[0087] In addition, if the number of non-zero elements in the k-th row C(k,:) included in the scaling matrix C is insufficient, it is necessary to allocate an extra dimension. Here, the set of indices of the extra dimensions that can be allocated to the k-th terminal 5 is denoted as "Z' k " The elements C(k,l) of the scaling matrix C are n ), the index l n is the set Z' k If it is included in the element C(k,l) as shown in the following "Number 40" n ) to achieve flipping motion.

number

number

number

[0088] After step S13, step S14 is executed. In step S14, the arithmetic unit 22 updates the common matrix A using the full rank parameter ξ below. The common matrix A is updated based on the following formula 43.

number

[0089] In the formula "Number 43", "[·] (j) " represents the unfolding in the jth mode. "Δ" is a parameter for preventing divergence. "ξ" is a parameter for full rank.

[0090] After step S14, step S15 is executed. In step S15, the arithmetic unit 22 calculates the decomposition error E RBk Determine whether the average value of the decomposition error E is smaller than a given threshold. RBkIf the average value of is not smaller than the threshold (No), the calculation device 22 executes step S12 again after step S15. RBk If the average value of is smaller than the threshold (Yes), the calculation device 22 executes step S16 after step S15. In other words, the calculation device 22 calculates the decomposition error E RBk Steps S12 to S14 are repeated until becomes smaller than the threshold value.

[0091] In step S16, the calculation device 22 calculates the individual matrix B k and the common matrix A to calculate the beamformer matrix.

[0092] The design concept of the beamformer according to one embodiment will be described. At the time of transmission, the channel matrix H k The common matrix A is diagonalized to put the transmitted signal into the desired dimension. k Individual matrix B of k Due to the unitary nature of this portion, the signal from each terminal 5 is extracted.

[0093] 7A is a schematic diagram illustrating the calculations in downlink communication of a beamforming method according to one embodiment. FIG. 7A includes seven matrices arranged from left to right. The leftmost matrix is ​​Q k Row LN column receive stream synthesis matrix J k T The received stream synthesis matrix J k T The elements of are either 0 or 1. In the example of FIG. 7A, the elements shown with hatching are "1" and the other elements are "0". The second matrix from the left is the individual matrix B k Hermitian matrix B of k H and the received stream synthesis matrix J k T The receive beamformer matrix U k Also, the individual matrix B k By the unitary nature of k Hermitian matrix B of kH and the channel matrix H k Individual matrix B in k and the matrix B k H B k As shown in Figure 7A, is a partial identity matrix. The fourth matrix from the left is the k-th scaling matrix C k The third matrix from the right is the transpose matrix A of the common matrix A for the channel tensor H. T Transpose matrix A T and the normalized inverse matrix (A T ) † and the matrix A T (A T ) † is an identity matrix, as shown in Figure 7A. The rightmost matrix has LN rows and Q k Column transmit stream selection matrix J k The transmission stream selection matrix J k The elements of are 0 or 1, and (A T ) † By multiplying with this, the transmit beamformer matrix V k In the example of FIG. 7A, the hatched elements are "1" and the other elements are "0". k is defined as shown in the following equation (44).

number

[0094] In the formula "44", "j k,i " is a stream selection variable for the i-th stream of the k-th terminal 5 and takes the value 0 or 1.

[0095] As shown in FIG. 7A, the received stream synthesis matrix J k T and individual matrix B k Hermitian matrix B of k H The product of this and the receive beamformer matrix U k And Q k Row Mk A matrix consisting of complex numbers in each column. Also, the individual matrix B k and the scaling matrix C k and the transposed matrix A T As explained with reference to FIG. 5A and “Equation 23”, the product of k In addition, the normalized inverse matrix (A T ) † and the transmission stream selection matrix J k The product of this and the transmit beamformer matrix V k And LN rows Q k In this way, the arithmetic unit 22 of the access point control device 2 according to the embodiment calculates the individual matrix B k Based on the receive beamformer matrix U k The transmit beamformer matrix V can be calculated based on the common matrix A. k can be calculated.

[0096] FIG. 7B is a schematic diagram illustrating the calculations in uplink communication of a beamforming method according to one embodiment. FIG. 7B includes seven matrices arranged from left to right. The leftmost matrix is ​​Q k Row LN column receive stream synthesis matrix J k T The received stream synthesis matrix J k T The elements of are either 0 or 1. In the example of FIG. 7A, the elements shown with hatching are "1" and the other elements are "0". The second matrix from the left is the normalized inverse matrix A of the common matrix A. † The normalized inverse matrix A of the common matrix A † and the product of the common matrix A, i.e., matrix A † A is the identity matrix as shown in Figure 7B. The fourth matrix from the left is the scaling matrix C k The third matrix from the right is the transpose matrix B of the individual matrix Bk. k T Individual matrix B k Transpose of matrix B k T and individual matrix B kThe complex conjugate matrix B of k * and the matrix B k T B k * is a partial identity matrix, as shown in Figure 7B. The rightmost matrix has LN rows and Q k Column transmit stream selection matrix J k The transmission stream selection matrix J k The elements of the transmission stream selection matrix J are either 0 or 1. In the example of FIG. 7B, similarly to the example of FIG. 7A, the elements shown with hatching are "1" and the other elements are "0". k is defined as in Equation 44.

[0097] As shown in FIG. 7B, the received stream synthesis matrix J k T and the normalized inverse matrix A † The product of this and the receive beamformer matrix U k u And M k It is a matrix consisting of complex numbers with rows and columns LN. Also, the common matrix A and the scaling matrix C k and the transposed matrix B k T As explained with reference to FIG. 5B and “Equation 26”, the product of k u In addition, the complex conjugate matrix B k * and the transmission stream selection matrix J k The product of this and the transmit beamformer matrix V k u And M k row Q k In this way, in uplink communication, the calculation device 22 of the access point control device 2 according to the embodiment calculates the receiving beamformer matrix U based on the common matrix A. k u The individual matrix B can be calculated. k Based on the transmit beamformer matrix V k u can be calculated.

[0098] In downlink communication, the transmit beamformer matrix V k qth column of <V k > q is designed to satisfy the following formula (45).

number

[0099] In the formula "Number 45", "q" is a number between 1 and Q. k It is an integer up to p k,i " represents the transmission power to be allocated to the i-th stream of the k-th terminal 5.

[0100] In uplink communication, the transmit beamformer matrix V k u qth column of <V k u > q is designed to satisfy the following formula (46).

number

[0101] In the formula "Number 46", "q" is a number between 1 and Q. k It is an integer up to p k,i u " represents the transmission power to be allocated to the i-th stream of the k-th terminal 5.

[0102] In downlink communication, the receive beamformer matrix U k As described with reference to FIG. 7A, is designed to satisfy the following formula (47).

number

[0103] In uplink communication, the receiving beamformer matrix U k u As described with reference to FIG. 7B, is designed to satisfy the following formula (48).

number

[0104] In the formula "48", the normalized inverse matrix A † is defined as shown in the following equation (49).

number

[0105] In equation (49), the common matrix A satisfies equation (50) below by singular value decomposition.

number

[0106] In equation (49), “Λ” (more precisely, there is a bar symbol above “Λ”) is defined as in equation (51) below.

number

[0107] When step S16 is completed, the flowchart in FIG. 6 ends.

[0108] Allocation of streams (antenna dimensions) in uplink and downlink communications and allocation of transmission power in downlink communications performed by the access point control device 2 according to one embodiment will be described with reference to Figs. 8A and 8B.

[0109] FIG. 8A is a flowchart showing an example of a configuration of a communication method according to an embodiment. FIG. 8B is a flowchart showing another example of a configuration of a communication method according to an embodiment. In the flowchart of FIG. 8A, streams and / or transmission power are allocated by a so-called "greedy method" that maximizes frequency utilization efficiency in the communication system 1. In the flowchart of FIG. 8B, streams and / or transmission power are allocated by a so-called "Max Min resource allocation" that ensures fairness in the allocation of radio resources among multiple terminals 5. By ensuring fairness in the allocation of radio resources, a minimum transmission rate can be guaranteed for all terminals 5. In the communication system 1 according to an embodiment, the arithmetic device 22 of the access point control device 2 can select and switch between the so-called "greedy method" and the so-called "Max Min resource allocation" at any timing. Non-Patent Document 9 (W. Rhee and J.M. Cioffi, “Increase in capacity of multiuser OFDM system using dynamic subchannel allocation”, Proc. IEEE VTC-Spring, Tokyo, Japan, vol. 2, 2000, pp. 1085-1089, doi: 10.1109 / VETECS.2000.851292.) describes Max Min resource allocation.

[0110] When the flowchart in FIG. 8A starts, step S21 is executed. In step S21, the calculation device 22 of the access point control device 2 calculates the column C(:, l) included in the scaling matrix C. n ) is the maximum value. n ) is the vector C(:,l n ) may also be called.

[0111] After step S21, step S22 is executed. In step S22, the arithmetic unit 22 transmits the k-th terminal 5 n Assign the stream dimensions.

[0112] After step S22, step S23 is executed. In step S23, the arithmetic unit 22 judges whether the number of times steps S21 and S22 have been repeated reaches a predetermined threshold value. This threshold value is the number Q of transmission streams corresponding to each of all terminals 5. k and is expressed as the following equation (52).

number

[0113] A numerical value representing the number of times steps S21 and S22 are repeated is preferably initialized before step S21 is executed for the first time. In addition, this number is preferably incremented at any point in time between steps S21 to S23, each time these steps are repeated. If the number of times of repetition has not reached the threshold (No), the calculation device 22 executes step S21 again. Conversely, if the number of times of repetition has reached the threshold (Yes), the flowchart in FIG. 8A ends.

[0114] 8B starts, step S31 is executed. In step S31, the calculation device 22 assigns to all terminals 5 one stream that has the maximum value in row C(k,:) included in the scaling matrix C. Row C(k,:) may be called vector C(k,:).

[0115] After step S31, step S32 is executed. In step S32, the calculation device 22 searches for the terminal 5 with the smallest allocated resources. This minimum value is expressed by the following formula (53).

number

[0116] In "Number 53", "q k “(i)” represents the index of the i-th stream assigned to the k-th terminal 5 .

[0117] After step S32, step S33 is executed. In step S33, the calculation device 22 assigns, to the searched terminal 5, one of the streams having the maximum value among the streams to which row C(k,:) is not assigned.

[0118] After step S33, step S34 is executed. In step S34, the arithmetic unit 22 judges whether the number of times steps S31, S32 and S33 have been repeated reaches a predetermined threshold value. This threshold value is the number of transmission streams Q corresponding to each of all terminals 5, as in the case of step S23 of the flowchart in FIG. 8A. k This is the sum of these and can be expressed as shown in Equation 52.

[0119] A numerical value representing the number of times steps S31, S32, and S33 are repeated is preferably initialized before step S31 is executed for the first time. This number is also preferably incremented each time these steps are repeated at any point in time from step S31 to step S34. If the number of times of repetition has not reached the threshold (No), the calculation device 22 executes step S31 again. Conversely, if the number of times of repetition has reached the threshold (Yes), the flowchart in FIG. 8B ends.

[0120] As a method for maximizing the frequency utilization efficiency in the communication system 1 according to the embodiment described with reference to the flowchart of FIG. 8A, there is a "water-filling method." In this case, the transmission power p k,q is expressed by the following equation (54).

number

[0121] As described with reference to the flowchart in FIG. 8B, the communication system 1 according to the embodiment combines the "water-filling" and "reverse water-filling" methods to ensure fairness among the terminals 5. That is, power is allocated to each terminal 5 based on the "reverse water-filling" method, and to each stream based on the "water-filling" method. In this case, the transmission power pk,q is expressed by the following equation (55).

number

[0122] In “Equation 54” and “Equation 55”, “P” represents the total transmission power constraint for all access points 3 .

[0123] (Simulation results) The results of simulating a communication method according to an embodiment will be described with reference to Fig. 9 to Fig. 24. Fig. 9 is a table showing an example of simulation parameters used when simulating a communication method according to an embodiment.

[0124] In the simulation results shown in Figs. 10 to 24, the improved ML-GSVD according to one embodiment is compared with Maximal Ratio Combining (MRC), Minimum Mean Square Error (MMSE) and ML-GSVD according to related techniques.

[0125] In the related art MRC-based beamformer, the transmit beamformer matrix V k is defined as the following equation (56), and the receive beamformer matrix U k is defined as the following equation (57).

number

number

[0126] In addition, in the MRC beamformer of the related technology, the transmit beamformer matrix V k u is defined as the following equation (58), and the receive beamformer matrix U k u is defined as the following equation (59).

number

number

[0127] In the MMSE beamformer of the related art, the transmit beamformer matrix V k is defined as the following equation (60), and the receive beamformer matrix U k is defined as the following equation (61).

number

number

[0128] In addition, in the MMSE beamformer of the related technology, the transmit beamformer matrix V k u is defined as the following equation (62), and the receive beamformer matrix U k u is defined as the following equation (63).

number

number

[0129] In the case of ML-GSVD, the transmission power p k,i is expressed by the following equation (64).

number

[0130] In addition, in the ML-GSVD according to the related art, the transmission power p k,i is expressed by the following formula (65).

number

[0131] 10 to 12 are graphs showing an example of a simulation result relating to the decomposition characteristics of the communication method according to one embodiment. In both of FIG. 10 to FIG. 12, the horizontal axis indicates the channel matrix H k To decompose the individual matrices B k , the number of iterations (iteration) i of the process of updating the scaling matrix C and the common matrix A, and the vertical axis represents the channel matrix H k Normalized Error E in the decomposition of (i) In addition, the total number L of the access points 3 is 25, the total number K of the terminals 5 is 25, and the total number M of the antenna elements of the k-th terminal 5 is 1. k is an integer between 2 and 6.

[0132] FIG. 10 shows a simulation result in a relatively low load state, and the total number N of antenna elements of each access point 3 is 9. FIG. 10 includes graphs G011 and G012. Graph G011 shows a simulation result of ML-GSVD according to the related art, and graph G012 shows a simulation result of improved ML-GSVD according to one embodiment. As shown in FIG. 10, when the number of iterations i exceeds a certain threshold, graph G012 is located below graph G011. This means that when the number of iterations i exceeds this threshold, the improved ML-GSVD according to one embodiment has less error than the ML-GSVD according to the related art, that is, the decomposition accuracy is higher.

[0133] FIG. 11 shows a simulation result in an intermediate load state, where the total number N of antenna elements of each access point 3 is 4. FIG. 11 includes graphs G021 and G022. Graph G021 shows a simulation result of ML-GSVD according to the related art, and graph G022 shows a simulation result of improved ML-GSVD according to an embodiment. As shown in FIG. 11, when the number of iterations i exceeds a certain threshold, graph G022 is located below graph G021. This means that when the number of iterations i exceeds this threshold, the improved ML-GSVD according to an embodiment has a smaller decomposition error than the ML-GSVD according to the related art, that is, the decomposition accuracy is higher.

[0134] FIG. 12 shows a simulation result in an overload state, where the total number N of antenna elements of each access point 3 is 3. FIG. 12 includes graphs G031 and G032. Graph G031 shows a simulation result of ML-GSVD according to the related art, and graph G032 shows a simulation result of improved ML-GSVD according to one embodiment. As shown in FIG. 12, when the number of iterations i exceeds a certain threshold, graph G022 is located below graph G021. This means that when the number of iterations i exceeds this threshold, the improved ML-GSVD according to one embodiment has a smaller decomposition error than the ML-GSVD according to the related art, that is, the decomposition accuracy is higher.

[0135] As can be seen from Figures 10 to 12, regardless of the load condition, if the number of iterations i exceeds this threshold, the improved ML-GSVD according to one embodiment will have less decomposition error than the ML-GSVD according to the related technology, i.e., the decomposition accuracy will be higher.

[0136] 13 to 15 are graphs showing an example of a simulation result relating to the spectral efficiency of all terminals 5 in downlink communication in a communication method according to an embodiment. In common to Figs. 13 to 15, the horizontal axis represents the spectral efficiency, and the vertical axis represents the CDF (Cumulative Distribution Function) of the spectral efficiency. In addition, the total number L of access points 3 is 25, the total number K of terminals 5 is 25, and the total number M of antenna elements possessed by the k-th terminal 5 is 1. k is an integer between 2 and 6.

[0137] FIG. 13 shows a simulation result in a relatively low load state, and the total number N of antenna elements of each access point 3 is 9. FIG. 13 includes graphs G041, G042, G043, G044, G045, G046, G047, and G048. Graph G041 shows a simulation result of MRC according to the related art when priority is given to fairness. Graph G042 shows a simulation result of MRC according to the related art when priority is given to frequency efficiency. Graph G043 shows a simulation result of MMSE according to the related art when priority is given to fairness. Graph G044 shows a simulation result of MMSE according to the related art when priority is given to frequency efficiency. Graph G045 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G046 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency efficiency. Graph G047 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G048 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 13, graph G047 is located to the right of all of graphs G041, G043, and G045. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G048 is located to the right of all of graphs G042, G044, and G046. This means that when priority is given to frequency utilization efficiency, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art.

[0138] FIG. 14 shows a simulation result in an intermediate load state, and the total number N of antenna elements of each access point 3 is 4. FIG. 14 includes graphs G051, G052, G053, G054, G055, G056, G057, and G058. Graph G051 shows a simulation result of MRC according to the related art when priority is given to fairness. Graph G052 shows a simulation result of MRC according to the related art when priority is given to frequency efficiency. Graph G053 shows a simulation result of MMSE according to the related art when priority is given to fairness. Graph G054 shows a simulation result of MMSE according to the related art when priority is given to frequency efficiency. Graph G055 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G056 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency efficiency. Graph G057 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G058 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 14, graph G057 is located to the right of all of graphs G051, G053, and G055. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G058 is located to the right of all of graphs G052, G054, and G056. This means that when priority is given to frequency utilization efficiency, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art.

[0139] FIG. 15 shows a simulation result in an overload state, and the total number N of antenna elements of each access point 3 is 3. FIG. 15 includes graphs G061, G062, G063, G064, G065, G066, G067, and G068. Graph G061 shows a simulation result of MRC according to the related art when fairness is prioritized. Graph G062 shows a simulation result of MRC according to the related art when frequency efficiency is prioritized. Graph G063 shows a simulation result of MMSE according to the related art when fairness is prioritized. Graph G064 shows a simulation result of MMSE according to the related art when frequency efficiency is prioritized. Graph G065 shows a simulation result of ML-GSVD according to the related art when fairness is prioritized. Graph G066 shows a simulation result of ML-GSVD according to the related art when frequency efficiency is prioritized. Graph G067 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G068 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 15, graph G067 is located to the right of all of graphs G061, G063, and G065. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G068 is located to the right of all of graphs G062, G064, and G066. This means that when priority is given to frequency utilization efficiency, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art.

[0140] As can be seen from Figures 13 to 15, regardless of the load condition, whether fairness is prioritized or frequency efficiency is prioritized, the improved ML-GSVD according to one embodiment has better frequency efficiency than any of the related technologies, MRC, MMSE, and ML-GSVD.

[0141] 16 to 18 are graphs showing an example of a simulation result relating to the spectral efficiency of a terminal 5 in the worst communication state among a plurality of terminals 5 in downlink communication of a communication method according to an embodiment. In common to Figs. 13 to 15, the horizontal axis represents the minimum spectral efficiency, and the vertical axis represents the CDF (Cumulative Distribution Function) of the minimum spectral efficiency. In addition, the total number L of access points 3 is 25, the total number K of terminals 5 is 25, and the total number M of antenna elements possessed by the kth terminal 5 is 1. k is an integer between 2 and 6.

[0142] FIG. 16 shows a simulation result in a relatively low load state, and the total number N of antenna elements of each access point 3 is 9. FIG. 16 includes graphs G071, G072, G073, G074, G075, G076, G077, and G078. Graph G071 shows a simulation result of MRC according to the related art when priority is given to fairness. Graph G072 shows a simulation result of MRC according to the related art when priority is given to frequency efficiency. Graph G073 shows a simulation result of MMSE according to the related art when priority is given to fairness. Graph G074 shows a simulation result of MMSE according to the related art when priority is given to frequency efficiency. Graph G075 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G076 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency efficiency. Graph G077 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G078 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 16, graph G077 is located to the right of all of graphs G071, G073, and G075. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G078 is located to the right of all of graphs G072, G074, and G076. This means that when priority is given to frequency utilization efficiency, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art.

[0143] FIG. 17 shows a simulation result in an intermediate load state, and the total number N of antenna elements of each access point 3 is 4. FIG. 17 includes graphs G081, G082, G083, G084, G085, G086, G087, and G088. Graph G081 shows a simulation result of MRC according to the related art when priority is given to fairness. Graph G082 shows a simulation result of MRC according to the related art when priority is given to frequency efficiency. Graph G083 shows a simulation result of MMSE according to the related art when priority is given to fairness. Graph G084 shows a simulation result of MMSE according to the related art when priority is given to frequency efficiency. Graph G085 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G086 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency efficiency. Graph G087 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G088 shows a simulation result of the improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 17, graph G087 is located to the right of all of graphs G081, G083, and G085. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G088 is located to the right of all of graphs G082, G084, and G086. This means that when priority is given to frequency utilization efficiency, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than all of MRC, MMSE, and ML-GSVD according to the related art.

[0144] FIG. 18 shows a simulation result in an overload state, and the total number N of antenna elements of each access point 3 is 3. FIG. 18 includes graphs G091, G092, G093, G094, G095, G096, G097, and G098. Graph G091 shows a simulation result of MRC according to the related art when priority is given to fairness. Graph G092 shows a simulation result of MRC according to the related art when priority is given to frequency efficiency. Graph G093 shows a simulation result of MMSE according to the related art when priority is given to fairness. Graph G094 shows a simulation result of MMSE according to the related art when priority is given to frequency efficiency. Graph G095 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G096 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency efficiency. Graph G097 shows a simulation result of the improved ML-GSVD according to an embodiment when fairness is prioritized. Graph G098 shows a simulation result of the improved ML-GSVD according to an embodiment when frequency efficiency is prioritized. As shown in FIG. 18, in a range where the minimum frequency efficiency exceeds a certain threshold, graph G097 is located to the right of all of graphs G091, G093, and G095. This means that, in a range where fairness is prioritized and the minimum frequency efficiency exceeds this threshold, the improved ML-GSVD according to an embodiment has better frequency efficiency than any of MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G098 is located to the right of all of graphs G092, G094, and G096 in a range where the minimum frequency efficiency exceeds another threshold. This means that when frequency utilization efficiency is prioritized, in the range where the minimum frequency utilization efficiency exceeds this other threshold, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than any of the related art MRC, MMSE, and ML-GSVD.

[0145] As can be seen from Figures 16 to 18, regardless of the load condition, whether fairness is prioritized or frequency efficiency is prioritized, the improved ML-GSVD of one embodiment has better frequency efficiency than any of the related technologies, MRC, MMSE, and ML-GSVD, at least in the range where the minimum frequency efficiency exceeds a certain threshold.

[0146] 19 to 21 are graphs showing an example of a simulation result relating to the spectral efficiency of all terminals 5 in uplink communication in a communication method according to an embodiment. In common to Figs. 19 to 21, the horizontal axis represents the spectral efficiency, and the vertical axis represents the CDF of the spectral efficiency. Furthermore, the total number L of access points 3 is 25, the total number K of terminals 5 is 25, and the total number M of antenna elements possessed by the k-th terminal 5 is 1. k is an integer between 2 and 6.

[0147] FIG. 19 shows a simulation result in a relatively low load state, and the total number N of antenna elements of each access point 3 is 9. FIG. 19 includes graphs G101, G102, G103, G104, G105, and G106. Please note that in FIG. 19, graphs G105 and G106 almost overlap. Graph G101 shows a simulation result of MRC according to the related art. Graph G102 shows a simulation result of MMSE according to the related art. Graph G103 shows a simulation result of ML-GSVD according to the related art when fairness is prioritized. Graph G104 shows a simulation result of ML-GSVD according to the related art when frequency utilization efficiency is prioritized. Graph G105 shows a simulation result of improved ML-GSVD according to one embodiment when fairness is prioritized. Graph G106 shows a simulation result of improved ML-GSVD according to one embodiment when frequency utilization efficiency is prioritized. As shown in FIG. 19, the graph G105 is located to the right of all of the graphs G101, G102, and G103. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, the graph G106 is located to the right of all of the graphs G101, G102, and G104. This means that when priority is given to frequency efficiency, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art.

[0148] FIG. 20 shows a simulation result in an intermediate load state, and the total number N of antenna elements of each access point 3 is 4. FIG. 20 includes graphs G111, G112, G113, G114, G115, and G116. Graph G111 shows a simulation result of MRC according to the related art. Graph G112 shows a simulation result of MMSE according to the related art. Graph G113 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G114 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency utilization efficiency. Graph G115 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G116 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 20, the graph G115 is located to the right of all of the graphs G111, G112, and G113. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, the graph G116 is located to the right of all of the graphs G111, G112, and G114. This means that when priority is given to frequency efficiency, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art.

[0149] FIG. 21 shows a simulation result in an overload state, and the total number N of antenna elements of each access point 3 is 3. FIG. 21 includes graphs G121, G122, G123, G124, G125, and G126. Graph G121 shows a simulation result of MRC according to the related art. Graph G122 shows a simulation result of MMSE according to the related art. Graph G123 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G124 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency utilization efficiency. Graph G125 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G126 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 21, graph G125 is located to the right of all of graphs G121, G122, and G123. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better spectral efficiency than any of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, graph G126 is located to the right of any of the graphs G121, G122, and G124. This means that when priority is given to frequency efficiency, the improved ML-GSVD according to one embodiment has better spectral efficiency than any of the MRC, MMSE, and ML-GSVD according to the related art.

[0150] As can be seen from Figures 19 to 21, regardless of the load condition, whether fairness is prioritized or frequency utilization efficiency is prioritized, the improved ML-GSVD according to one embodiment has better frequency utilization efficiency than any of the related technologies, MRC, MMSE, and ML-GSVD.

[0151] 22 to 24 are graphs showing an example of a simulation result relating to the spectral efficiency of a terminal 5 in the worst communication state among a plurality of terminals 5 in downlink communication of a communication method according to an embodiment. In common to Figs. 22 to 24, the horizontal axis represents the minimum spectral efficiency, and the vertical axis represents the CDF (Cumulative Distribution Function) of the minimum spectral efficiency. In addition, the total number L of access points 3 is 25, the total number K of terminals 5 is 25, and the total number M of antenna elements possessed by the kth terminal 5 is 1. k is an integer between 2 and 6.

[0152] FIG. 22 shows a simulation result in a relatively low load state, and the total number N of antenna elements of each access point 3 is 9. FIG. 22 includes graphs G131, G132, G133, G134, G135, and G136. Please note that in FIG. 22, graphs G135 and G136 almost overlap. Graph G131 shows a simulation result of MRC according to the related art. Graph G132 shows a simulation result of MMSE according to the related art. Graph G133 shows a simulation result of ML-GSVD according to the related art when fairness is prioritized. Graph G134 shows a simulation result of ML-GSVD according to the related art when frequency utilization efficiency is prioritized. Graph G135 shows a simulation result of improved ML-GSVD according to one embodiment when fairness is prioritized. Graph G136 shows a simulation result of improved ML-GSVD according to one embodiment when frequency utilization efficiency is prioritized. As shown in FIG. 22, the graph G135 is located to the right of all of the graphs G131, G132, and G133. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, the graph G136 is located to the right of all of the graphs G131, G132, and G134. This means that when priority is given to frequency efficiency, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art.

[0153] FIG. 23 shows a simulation result in an intermediate load state, and the total number N of antenna elements of each access point 3 is 4. FIG. 23 includes graphs G141, G142, G143, G144, G145, and G146. Please note that in FIG. 23, graphs G145 and G146 almost overlap. Graph G141 shows a simulation result of MRC according to the related art. Graph G142 shows a simulation result of MMSE according to the related art. Graph G143 shows a simulation result of ML-GSVD according to the related art when fairness is prioritized. Graph G144 shows a simulation result of ML-GSVD according to the related art when frequency utilization efficiency is prioritized. Graph G145 shows a simulation result of improved ML-GSVD according to one embodiment when fairness is prioritized. Graph G146 shows a simulation result of improved ML-GSVD according to one embodiment when frequency utilization efficiency is prioritized. As shown in FIG. 23, the graph G145 is located to the right of all of the graphs G141, G142, and G143. This means that when priority is given to fairness, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, the graph G146 is located to the right of all of the graphs G141, G142, and G144. This means that when priority is given to frequency efficiency, the improved ML-GSVD according to one embodiment has better spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art.

[0154] FIG. 24 shows a simulation result in an overload state, and the total number N of antenna elements of each access point 3 is 3. FIG. 24 includes graphs G151, G152, G153, G154, G155, and G156. Graph G151 shows a simulation result of MRC according to the related art. Graph G152 shows a simulation result of MMSE according to the related art. Graph G153 shows a simulation result of ML-GSVD according to the related art when priority is given to fairness. Graph G154 shows a simulation result of ML-GSVD according to the related art when priority is given to frequency utilization efficiency. Graph G155 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to fairness. Graph G156 shows a simulation result of improved ML-GSVD according to one embodiment when priority is given to frequency utilization efficiency. As shown in FIG. 24, in a range where the minimum spectral efficiency exceeds a certain threshold, the graph G155 is located to the right of all of the graphs G151, G152, and G153. This means that, when fairness is prioritized, in a range where the minimum spectral efficiency exceeds this threshold, the improved ML-GSVD according to one embodiment has a higher spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art. Similarly, the graph G156 is located to the right of all of the graphs G151, G152, and G154 in a range where the minimum spectral efficiency exceeds another threshold. This means that, when spectral efficiency is prioritized, in a range where the minimum spectral efficiency exceeds this other threshold, the improved ML-GSVD according to one embodiment has a higher spectral efficiency than all of the MRC, MMSE, and ML-GSVD according to the related art.

[0155] As can be seen from Figures 22 to 24, regardless of the load condition, whether fairness is prioritized or frequency efficiency is prioritized, the improved ML-GSVD of one embodiment has better frequency efficiency than any of the related technologies, MRC, MMSE, and ML-GSVD, at least in the range where the minimum frequency efficiency exceeds a certain threshold.

[0156] Although the invention made by the inventor has been specifically described based on the embodiment, the present invention is not limited to the embodiment, and it goes without saying that various modifications can be made without departing from the gist of the invention. Furthermore, the respective features described in the embodiment can be freely combined within the scope of technical compatibility. [Explanation of symbols]

[0157] 1. Communication Systems 2 Access point control device 20 Bus 21 Antenna 22 Arithmetic unit 23 Storage device 230 Recording media 231 Control Program 24 Interface 3. Access Points 30 Bus 31 Antenna 32 Arithmetic unit 33 Storage device 330 Recording media 331 Correspondence Programs 34 Interface 4. Fronthaul 5. Terminal 50 Bus 51 Antenna 52 Arithmetic unit 53 Storage device 530 Recording media 531 Correspondence Programs 54 Interface G011, G012 graph G021, G022 graph G031, G032 graph G041, G042, G043, G044, G045, G046, G047, G048 Graph G051, G052, G053, G054, G055, G056, G057, G058 Graph G061, G062, G063, G064, G065, G066, G067, G068 Graph G071, G072, G073, G074, G075, G076, G077, G078 Graph G081, G082, G083, G084, G085, G086, G087, G088 Graph G091, G092, G093, G094, G095, G096, G097, G098 Graph G101, G102, G103, G104, G105, G106 graphs G111, G112, G113, G114, G115, G116 Graph G121, G122, G123, G124, G125, G126 graphs G131, G132, G133, G134, G135, G136 Graphs G141, G142, G143, G144, G145, G146 Graphs G151, G152, G153, G154, G155, G156 graphs

Claims

1. A computing device that controls a plurality of access points that perform wireless communication between a plurality of terminals in a cell-free MIMO system; an interface connected to the plurality of access points; Equipped with The computing device includes: calculating a communication channel matrix representing each state of a plurality of communication channels through which uplink communication and / or downlink communication is performed between a plurality of first antenna elements included in each of the plurality of access points and a plurality of second antenna elements included in each of the plurality of terminals; calculating a first matrix, a second matrix, and a third matrix to decompose the channel matrix into a product of the first matrix, the second matrix, and a third matrix; calculating a transmit beamformer matrix for performing transmit beamforming and a receive beamformer matrix for performing receive beamforming based on the first matrix and the third matrix; Preventing divergence of the second matrix by using an anti-divergence parameter to calculate the second matrix. Access point control device.

2. 2. The access point control device according to claim 1, The computing device includes: calculating the second matrix using the anti-divergence parameter; A process of calculating the third matrix using a full rank parameter; and calculating the first matrix based on the second matrix and the third matrix. This process is repeated until a decomposition error between the communication channel matrix and the product becomes smaller than a predetermined threshold value. Access point control device.

3. 3. The access point control device according to claim 2, The arithmetic device calculates the second matrix by further using a flipping operation for eliminating an excess or deficiency in the number of dimensions expressing a channel matrix corresponding to each terminal for elements included in the second matrix. Access point control device.

4. 4. The access point control device according to claim 2, The calculation device selects and switches between a first calculation method that maximizes frequency utilization efficiency between the plurality of access points and the plurality of terminals and a second calculation method that maintains fairness in allocation of radio resources among the plurality of terminals in order to calculate the transmission beamformer matrix and the reception beamformer matrix. Access point control device.

5. an access point control device including a calculation device for controlling a plurality of access points that perform wireless communication between a plurality of terminals in a cell-free MIMO system; The plurality of terminals; the plurality of access points; Equipped with The access point control device includes: The computing device; an interface connected to the plurality of access points; Equipped with The computing device includes: calculating a communication channel matrix representing each state of a plurality of communication channels through which uplink communication and / or downlink communication is performed between a plurality of first antenna elements included in each of the plurality of access points and a plurality of second antenna elements included in each of the plurality of terminals; calculating a first matrix, a second matrix, and a third matrix to decompose the channel matrix into a product of the first matrix, the second matrix, and a third matrix; calculating a transmit beamformer matrix for performing transmit beamforming and a receive beamformer matrix for performing receive beamforming based on the first matrix and the third matrix; Preventing divergence of the second matrix by using an anti-divergence parameter to calculate the second matrix. Communication systems.

6. an access point control device including a calculation device for controlling a plurality of access points that perform wireless communication between a plurality of terminals in a cell-free MIMO system; The plurality of terminals; the plurality of access points; A communication method for performing wireless communication in a communication system comprising: calculating a communication channel matrix representing each state of a plurality of communication channels through which uplink communication and / or downlink communication is performed between a plurality of first antenna elements included in each of the plurality of access points and a plurality of second antenna elements included in each of the plurality of terminals; calculating a first matrix, a second matrix, and a third matrix to decompose the channel matrix into a product of the first matrix, the second matrix, and a third matrix; calculating a transmit beamformer matrix for performing transmit beamforming and a receive beamformer matrix for performing receive beamforming based on the first matrix and the third matrix; Preventing divergence of the second matrix by using an anti-divergence parameter to calculate the second matrix. Includes Communication methods.

Images