Electronic device and method for transmitting and receiving signals in a multiple-input multiple-output system

Abstract

An embodiment provides a radio unit (RU) device. The device may include: a radio frequency (RF) transceiver; a fronthaul transceiver; a memory storing instructions; and a processor. When executed by the processor, the instructions cause the device to: acquire a reference signal via the RF transceiver; acquire a transformed covariance matrix of the reference signal for noise and interference via random embedding for dimensionality reduction; acquire an extraction matrix via subspace decomposition of the transformed covariance matrix; use the extraction matrix to pre-combine uplink signals acquired via the RF transceiver; and transmit the pre-combine uplink signal data to a digital unit (DU) via the fronthaul transceiver. The random embedding can be used to transform a first dimension of the RU's receiver layer number into a second dimension fewer than the number of receiver layers.

Description

Technical Field

[0001] This disclosure relates to multiple-input multiple-output (MIMO). For example, this disclosure relates to electronic devices and methods for transmitting and receiving signals in a MIMO system. Background Technology

[0002] Multiple-input multiple-output (MIMO) technology is used to improve signal transmission / reception performance. Both transmitters and receivers using MIMO technology can use multiple antennas. Compared to single-antenna technology, this significantly improves the channel capacity of wireless communication systems using MIMO technology.

[0003] The information described above is provided as relevant technology to aid in understanding the purposes of this disclosure. No dispute or decision is made regarding whether any of the above descriptions can be applied as prior art in connection with this disclosure. Summary of the Invention

[0004] Technical solution In one embodiment, a device for a radio unit (RU) is provided. The device may include: a radio frequency (RF) transceiver; a fronthaul transceiver; a memory storing instructions; and a processor. When executed by the processor, the instructions can cause the device to: obtain a reference signal via the RF transceiver; obtain a transformed covariance matrix of the reference signal for noise and interference through random embedding for dimensionality reduction; obtain an extraction matrix through subspace decomposition of the transformed covariance matrix; perform pre-combining on the uplink signal obtained via the RF transceiver using the extraction matrix; and transmit data of the pre-combined uplink signal to a digital unit (DU) via the fronthaul transceiver. The random embedding can be used to transform a first dimension having a number of receiver layers with the RU into a second dimension fewer than the number of receiver layers.

[0005] In one embodiment, a device for a digital unit (DU) is provided. The device may include: a transceiver; a memory storing instructions; and a processor. When executed by the processor, the instructions can cause the device to: obtain a reference signal from a radio unit (RU) via the transceiver; obtain a transformed covariance matrix of the reference signal for noise and interference via random embedding for dimensionality reduction; obtain an extraction matrix via subspace decomposition of the transformed covariance matrix; and obtain a transmitted signal by performing equalization on the uplink signal from the RU using the extraction matrix. The random embedding can be used to transform a first dimension having a number of receiver layers with RUs into a second dimension having fewer than the number of receiver layers.

[0006] In one embodiment, a device for a radio unit (RU) is provided. The device may include: a radio frequency (RF) transceiver; a fronthaul transceiver; a memory storing instructions; and a processor. When executed by the processor, the instructions can cause the device to: obtain a covariance matrix of noise and interference from a reference signal received via the RF transceiver; obtain a noise vector matrix by performing eigenvalue decomposition of the covariance matrix; obtain a filter matrix using the noise vector matrix and a channel matrix for the reference signal; obtain a transformed signal by applying the filter matrix to an uplink signal obtained via the RF transceiver; and transmit data of the transformed signal to a digital unit (DU) via the fronthaul transceiver. The number of columns in the noise vector matrix may be less than the number of receiver layers of the RU.

[0007] In an embodiment, a device for a radio unit (RU) may include: a radio frequency (RF) transceiver; a fronthaul transceiver; a memory storing instructions; and a processor. When executed by the processor, the instructions can cause the device to: obtain a reference signal via the RF transceiver; obtain a transformed covariance matrix of noise and interference via the reference signal; obtain an extraction matrix having column vectors corresponding to a specified number of column vectors in the orthogonal matrix of the QR decomposition by performing QR decomposition of the covariance matrix; obtain a filter matrix using a projection matrix obtained from the extraction matrix and a channel matrix for the reference signal; obtain a transformed signal by applying the filter matrix to an uplink signal obtained via the RF transceiver; and transmit data of the transformed signal to a digital unit (DU) via the fronthaul transceiver. Attached Figure Description

[0008] Figure 1 A wireless communication system is shown.

[0009] Figure 2a The fronthaul interface is shown.

[0010] Figure 2b The fronthaul interface of the Open (O) Radio Access Network (RAN) is shown.

[0011] Figure 3a The components of a distributed unit (DU) are shown.

[0012] Figure 3b The components of a radio unit (RU) are shown.

[0013] Figure 4 An example of functional splitting between DU and RU is shown.

[0014] Figure 5 An example of a transmitter and receiver is shown.

[0015] Figure 6 An example of subspace decomposition using random embeddings is shown.

[0016] Figure 7 An example of subspace decomposition using random embeddings is shown.

[0017] Figure 8 An example of precoding or precomposition using subspace decomposition is shown.

[0018] Figure 9 An example of a receiving operation using the eigenvalue decomposition of the covariance matrix of noise and interference is shown.

[0019] Figure 10 An example of receiving operation using the null space of the covariance matrix of noise and interference is shown.

[0020] Figure 11 An example of receiving operation using the null space of the covariance matrix of noise and interference is shown.

[0021] Figure 12 An example of a receive operation using weighted interpolation is shown.

[0022] Figure 13a and Figure 13b An example of the performance of the receive operation using random embedding and null space is shown. Detailed Implementation

[0023] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the scope of another embodiment. Singular expressions may include plural expressions unless the context clearly indicates otherwise. The terms used herein (including technical or scientific terms) may have the same meanings as commonly understood by one of ordinary skill in the art as described in this disclosure. Among the terms used in this disclosure, unless explicitly defined herein, terms defined in general dictionaries may be interpreted as having the same or similar meaning as in the context of related art and not as having an ideal or overly formal meaning. In some cases, even the terms defined in this disclosure may not be construed as excluding embodiments of this disclosure.

[0024] In the various embodiments of this disclosure described below, hardware solutions will be described as examples. However, because the various embodiments of this disclosure include techniques using both hardware and software, software-based solutions are not excluded.

[0025] The terms used in the following description, referring to signals (e.g., signals, information, messages, and signaling), data types (e.g., lists, sets, and subsets), states used for calculation (e.g., steps, operations, and procedures), data (e.g., packets, user streams, information, bits, symbols, and codewords), resources (e.g., symbols, time slots, subframes, radio frames, subcarriers, resource elements (REs), resource blocks (RBs), bandwidth portions (BWPs), and timings), channels, network entities, and components of devices, are illustrative for ease of description. Therefore, this disclosure is not limited to the terms described below, and other terms with equivalent technical meanings may be used.

[0026] The terms used in the following description that refer to signals (e.g., signal, information, message, or signaling), resources (e.g., symbol, time slot, subframe, radio frame, subcarrier, resource element (RE), resource block (RB), bandwidth portion (BWP), or timing), terms used for calculating states (e.g., step, operation, or process), data (e.g., packet, user stream, information, bit, symbol, or codeword), channels, network entities, and components of devices are illustrative for ease of description. Therefore, this disclosure is not limited to the terms described below, and other terms with equivalent technical meanings may be used.

[0027] Additionally, in this disclosure, the terms "greater than" or "less than" may be used to determine whether a particular condition is met or fulfilled; however, this is merely a description for illustrative purposes and does not exclude descriptions of "greater than or equal to" or "less than or equal to". A condition described as "greater than or equal to" may be replaced with "greater than", a condition described as "less than or equal to" may be replaced with "less than", and a condition described as "greater than or equal to and less than" may be replaced with "greater than and less than or equal to". Furthermore, in the following, "A" to "B" refers to at least one element from A (inclusive) to B (inclusive). In the following, "C" and / or "D" means including at least one of "C" or "D", i.e., {"C", "D", and "C" and "D"}.

[0028] Although this disclosure uses terms used in some communication standards (e.g., 3GPP, ETSI, xRAN, ORAN) to describe various embodiments, these are merely illustrative examples. The various embodiments of this disclosure can be readily modified and applied to other communication systems.

[0029] Figure 1 A wireless communication system is shown.

[0030] refer to Figure 1 , Figure 1 Base station 110 and terminal 120 are shown as part of a plurality of nodes utilizing a wireless channel in a wireless communication system. Figure 1 Only one base station is shown, but the wireless communication system may also include another base station that is the same as or similar to base station 110.

[0031] Base station 110 is a network infrastructure that provides wireless access to terminal 120. Base station 110 has a coverage area defined by the distance from which signals can be transmitted. In addition to "base station", base station 110 may also be referred to as "access point (AP)", "eNodeB (eNB)", "fifth generation node", "next generation node B (gNB)", "wireless point", "transmit / receive point (TRP)" or other terms with equivalent technical meanings.

[0032] Terminal 120, as a device used by a user, communicates with base station 110 via a wireless channel. The link from base station 110 to terminal 120 is called the downlink (DL), and the link from terminal 120 to base station 110 is called the uplink (UL). Additionally, although in Figure 1 Not shown, but terminal 120 and another terminal can communicate with each other via a wireless channel. In this case, the link between terminal 120 and the other terminal (device-to-device link (D2D)) is called a side link, and the side link can be used interchangeably with the PC5 interface. In some other embodiments, terminal 120 can be operated without user intervention. According to embodiments, terminal 120, as a device performing machine-type communication (MTC), may not be carried by the user. Additionally, according to embodiments, terminal 120 may be a narrowband (NB) Internet of Things (IoT) device.

[0033] In addition to “terminal”, terminal 120 may also be referred to as “user equipment (UE)”, “customer premises equipment (CPE)”, “mobile station”, “subscriber station”, “remote terminal”, “wireless terminal”, “electronic device”, “user equipment” or other terms with equivalent technical meaning.

[0034] Base station 110 can perform beamforming with terminal 120. Base station 110 and terminal 120 can transmit and receive radio signals in relatively low frequency bands (e.g., NR frequency range 1 (FR 1)). Additionally, base station 110 and terminal 120 can transmit and receive radio signals in relatively high frequency bands (e.g., FR 2 (or FR 2-1, FR 2-2, FR 2-3) or FR 3) and millimeter-wave frequency bands (e.g., 28 GHz, 30 GHz, 38 GHz, 60 GHz). Base station 110 and terminal 120 can perform beamforming to improve channel gain. In this document, beamforming can include transmit beamforming and receive beamforming. Base station 110 and terminal 120 can provide directionality to the transmitted or received signals. To this end, base station 110 and terminal 120 can select a serving beam through a beam search or beam management process. After selecting a serving beam, subsequent communication can be performed using resources that are in a QCL relationship with the resources of the transmit serving beam.

[0035] If the large-scale characteristics of the channel carrying symbols on the first antenna port can be inferred from the channel carrying symbols on the second antenna port, then the first and second antenna ports can be evaluated as being in a QCL relationship. For example, large-scale characteristics may include at least one of delay spread, Doppler spread, Doppler shift, average gain, average delay, and spatial receiver parameters.

[0036] although Figure 1 The description states that both base station 110 and terminal 120 perform beamforming, but the embodiments of this disclosure are not necessarily limited to this. In some embodiments, the terminal may or may not perform beamforming. Additionally, the base station may or may not perform beamforming. That is, either only one of the base station and the terminal may perform beamforming, or neither the base station nor the terminal may perform beamforming.

[0037] In this disclosure, a beam refers to the spatial flow of a signal in a wireless channel and is formed by one or more antennas (or antenna elements), and this forming process may be referred to as beamforming. Beamforming may include at least one of analog beamforming or digital beamforming (e.g., precoding). Reference signals transmitted based on beamforming may include, for example, demodulation reference signals (DM-RS), channel state information-reference signals (CSI-RS), synchronization signal / physical broadcast channel (SS / PBCH), and sounding reference signals (SRS). Additionally, IEs such as CSI-RS resources or SRS resources may be used as configurations for each reference signal, and this configuration may include beam-associated information. Beam-associated information may refer to whether the corresponding configuration (e.g., a CSI-RS resource) uses the same or different spatial domain filters as another configuration (e.g., another CSI-RS resource within the same CSI-RS resource set), or which reference signal it is quasi-co-located (QCL) with, and if so, what type it is (e.g., QCL type A, B, C, D).

[0038] Conventionally, in communication systems with relatively large base station cell radii, each base station is installed to include the functions of a digital processing unit (or distributed unit (DU)) and a radio frequency (RF) processing unit (or radio unit (RU)). However, with the use of high-frequency bands in fourth-generation (4G) and / or subsequent communication systems (e.g., 5G) and the shrinking cell coverage of base stations, the number of base stations required to cover a specific area has been increasing. The installation cost burden for operators has also been increasing. To minimize the installation cost of base stations, a structure has been proposed where the DU and RU of a base station are separate, one or more RUs are connected to a DU via a wired network, and one or more RUs are deployed geographically to cover a specific area. In the following sections, through... Figure 2a and Figure 2b Describes deployment structures and extended examples of base stations according to various embodiments of this disclosure.

[0039] Figure 2a The fronthaul interface is shown. Unlike backhaul between the base station and the core network, fronthaul refers to the physical link between the wireless LAN and the base station. Figure 2a An example of a fronthaul structure between a DU 210 and a RU 220 is shown, but this is for illustrative purposes only and the disclosure is not limited thereto. In other words, embodiments of the disclosure can also be applied to a fronthaul structure between a DU and multiple RUs. For example, embodiments of the disclosure can be applied to a fronthaul structure between a DU and two RUs. Additionally, embodiments of the disclosure can also be applied to a fronthaul structure between a DU and three RUs.

[0040] refer to Figure 2a Base station 110 may include DU 210 and RU 220. Fronthaul 215 between DU 210 and RU 220 may be operated via Fx interface. For operation of fronthaul 215, interfaces such as Enhanced Universal Public Radio Interface (eCPRI) or Radio over Ethernet (ROE) may be used.

[0041] As communication technologies continue to evolve, mobile data services have increased, leading to a significant increase in bandwidth requirements for fronthaul between digital units and radio units. In deployments such as centralized / cloud radio access networks (C-RAN), the DU can be implemented to perform functions for Packet Data Convergence Protocol (PDCP), Radio Link Control (RLC), Media Access Control (MAC), and Physical (PHY), while the RU can be implemented to perform functions for the PHY layer in addition to performing radio frequency (RF) functions.

[0042] DU 210 can be responsible for the upper-layer functions of the wireless network. For example, DU 210 can perform MAC layer functions and a portion of the PHY layer functions. Here, "a portion of the PHY layer" refers to functions performed at a higher level within the PHY layer and may include, for example, channel coding (or channel decoding), scrambling (or descrambling), modulation (or demodulation), and layer mapping (or layer demapping). According to embodiments, if DU 210 conforms to the O-RAN standard, it may be referred to as an O-RAN DU (O-DU). Where necessary, in embodiments of this disclosure, DU 210 may be replaced with and represented as a first network entity for a base station (e.g., gNB).

[0043] The RU 220 can handle lower-layer functions of the wireless network. For example, the RU 220 can perform a portion of the PHY layer and RF functions. In this paper, "a portion of the PHY layer" refers to functions performed at a relatively lower level compared to the DU 210, and may include, for example, iFFT transformation (or FFT transformation), cyclic prefix (CP) insertion (or CP removal), and digital beamforming. Figure 4 The document describes in detail examples of such specific functional splitting. RU 220 may be referred to as an Access Unit (AU), Access Point (AP), Transmit / Receive Point (TRP), Remote Radio Header (RRH), Radio Unit (RU), or other terms with equivalent technical meanings. According to embodiments, if RU 220 conforms to the O-RAN standard, it may be referred to as an O-RAN RU (O-RU). Where necessary, in embodiments of this disclosure, RU 220 may be replaced with and represented as a second network entity for a base station (e.g., gNB).

[0044] although Figure 2a Base station 110 is described as including DU 210 and RU 220, but embodiments of this disclosure are not limited thereto. Base stations according to embodiments can be implemented in a distributed deployment using centralized units (CUs) configured to perform functions of the upper layers of the access network (e.g., Packet Data Convergence Protocol (PDCP), Radio Resource Control (RRC)) and distributed units (DUs) configured to perform functions of the lower layers. In this case, the distributed unit (DU) may correspond to a digital unit (DU) or at least a portion of the functional components including digital units (DUs). The base station can be implemented in a structure in which CUs, DUs, and RUs are arranged sequentially between the core network (e.g., 5G core (5GC) or next-generation core (NGC)) and the radio access network (RAN). The interface between the CU and the distributed unit (DU) may be referred to as an F1 interface.

[0045] Compared to a DU, a centralized unit (CU) can handle higher-layer functions by connecting to one or more DUs. For example, a CU can handle Radio Resource Control (RRC) and Packet Data Convergence Protocol (PDCP) layer functions, while DUs and RUs can handle lower-layer functions. A DU can perform Radio Link Control (RLC), Media Access Control (MAC), and some functions of the PHY layer (high PHY), while an RU can perform the remaining functions of the PHY layer (low PHY). Additionally, as an example, digital units (DUs) can be included within a distributed unit (DU) depending on the implementation of the base station's distributed deployment. Hereinafter, unless otherwise defined, it is described as the operation of digital units (DUs) and RUs; however, various embodiments of this disclosure can be applied to base station arrangements that include CUs or where DUs are directly connected to the core network (i.e., CUs and DUs are integrated into a base station as a single entity (e.g., an NG-RAN node)).

[0046] Figure 2b The fronthaul interface of an Open (O) Radio Access Network (RAN) is shown. An eNB or gNB is illustrated as a base station 110 in a distributed deployment.

[0047] refer to Figure 2b Base station 110 may include O-DU 251 and O-RU 253-1, ..., and 253-n. In the following text, for ease of explanation, the operation and function of O-RU 253-1 can be understood as a description of each other O-RU (e.g., O-RU 253-n).

[0048] O-DU 251 includes, in addition to the functions exclusively assigned to O-RU 253-1, the features described later. Figure 4The O-DU 251 is a logical node that controls the functions of base stations (e.g., eNB, gNB). The O-DU 251 can control the operation of O-RUs 253-1, ..., and 253-n. The O-DU 251 can be referred to as the Lower Layer Split (LLS) Central Unit (CU). The O-RU 253-1 includes components according to the description to be provided later. Figure 4 Logical nodes that are a subset of the functions of a base station (e.g., eNB, gNB). Real-time aspects of communication with the O-RU253-1's control plane (C plane) and user plane (U plane) can be controlled by the O-DU 251.

[0049] The O-DU 251 can communicate with the O-RU 253-1 via the LLS interface. The LLS interface corresponds to the fronthaul interface. The LLS interface refers to the logical interface between the O-DU 251 and the O-RU 253-1 using lower-level function decomposition (i.e., function decomposition within the PHY). The LLS-C between the O-DU 251 and the O-RU 253-1 provides the C-plane through the LLS interface. The LLS-U between the O-DU 251 and the O-RU 253-1 provides the U-plane through the LLS interface.

[0050] exist Figure 2b In this document, the entities of base station 110 have been described as O-DU and O-RU to describe O-RAN. However, these names should not be construed as limiting the embodiments of this disclosure. In the embodiments described below, the operation of DU 210 can also be performed by O-DU 251. The description of DU 210 can be applied to O-DU 251. Similarly, in the case of... Figures 3a to 12 In the embodiment described in b, the operation of RU 220 can also be performed by O-RU 253-1. The description of RU 220 can be applied to O-RU 253-1.

[0051] Figure 3a The components of a distributed unit (DU) are shown. This is part of a base station. Figure 3a The illustrated configuration can be understood as Figure 2a DU 210 (or Figure 2b The configuration of O-DU 251. In the following text, the terms “…unit” and “…device” as used below refer to a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software.

[0052] refer to Figure 3a The DU 210 includes a transceiver 310, a memory 320, and a processor 330.

[0053] Transceiver 310 can perform functions for transmitting and receiving signals in a wired communication environment. Transceiver 310 may include a wired interface for controlling direct device-to-device connections via a transmission medium (e.g., copper wire, optical fiber). For example, transceiver 310 can transmit electrical signals to another device via copper wire or perform conversion between electrical and optical signals. DU 210 can communicate with a radio unit (RU) via transceiver 310. DU 210 can connect to a core network or a distributed CU via transceiver 310.

[0054] Transceiver 310 can also perform functions for transmitting and receiving signals in a wireless communication environment. For example, transceiver 310 can perform baseband signal to bit string conversion functions according to the system's physical layer specifications. For example, when transmitting data, transceiver 310 generates complex-valued symbols by encoding and modulating the transmitted bit string. Additionally, when receiving data, transceiver 310 recovers the received bit string by demodulating and decoding the baseband signal. Furthermore, transceiver 310 may include multiple transmit / receive paths. Additionally, according to embodiments, transceiver 310 may be connected to a core network or to other nodes (e.g., integrated access backhaul (IAB)).

[0055] Transceiver 310 can send and receive signals. For example, transceiver 310 can send management plane (M-plane) messages. For example, transceiver 310 can send synchronization plane (S-plane) messages. For example, transceiver 310 can send control plane (C-plane) messages. For example, transceiver 310 can send user plane (U-plane) messages. For example, transceiver 310 can receive U-plane messages. Although only in... Figure 3a Transceiver 310 is shown, but according to another implementation, DU 210 may include two or more transceivers.

[0056] Transceiver 310 transmits and receives signals as described above. Therefore, all or some of transceiver 310 may be referred to as a "communication unit," "transmitting unit," "receiving unit," or "transmitting / receiving unit." Furthermore, in the following description, "transmission and reception performed via a wireless channel" is used to include the meaning of the processes performed by transceiver 310 as described above.

[0057] Despite Figure 3a Not shown, but transceiver 310 may also include a backhaul transceiver for connecting to the core network or another base station. The backhaul transceiver provides an interface for performing communication with other nodes in the network. In other words, the backhaul transceiver converts bit strings sent from the base station to another node (such as another access node, another base station, an upstream node, and the core network) into physical signals, and converts physical signals received from another node into bit strings.

[0058] Memory 320 stores basic programs, application programs, and data such as configuration information used for the operation of DU 210. Memory 320 may be referred to as a storage unit. Memory 320 may be configured with volatile memory, non-volatile memory, or a combination of volatile and non-volatile memory. Additionally, memory 320 provides stored data according to requests from processor 330. As a functional component, memory 320 indicates storage space. For example, memory 320 can be understood not only to indicate memory (e.g., hard disk, flash memory, or RAM) disposed as part of DU 210, but also to indicate space for storing instructions and / or programs.

[0059] Processor 330 controls the overall operation of DU 210. Processor 380 can be referred to as a control unit. For example, processor 330 sends and receives signals via transceiver 310 (or via a return communication unit). Additionally, processor 330 writes and reads data from memory 320. Furthermore, processor 330 can perform the functions of the protocol stack required in the communication standard. Although only in... Figure 3a The image shows processor 330, but according to another implementation, DU 210 may include two or more processors.

[0060] Figure 3a The configuration of DU 210 shown is merely an example, and examples of DUs performing embodiments of this disclosure are not limited to. Figure 3a The configuration shown is illustrated. In some embodiments, configurations can be added, deleted, or changed.

[0061] Figure 3b The components of a radio unit (RU) are shown. This is part of a base station. Figure 3b The illustrated configuration can be understood as Figure 2b RU 220 or Figure 2b The configuration of the O-RU 253-1. In the following text, the terms "...unit" and "...device" as used below refer to a unit that processes at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software.

[0062] refer to Figure 3b The RU 220 includes an RF transceiver 360, a fronthaul transceiver 365, a memory 370, and a processor 380.

[0063] The RF transceiver 360 performs functions for transmitting and receiving signals via a wireless channel. For example, the RF transceiver 360 up-converts a baseband signal into an RF band signal, then transmits the RF band signal through an antenna, and down-converts the RF band signal received through the antenna back into a baseband signal. The RF transceiver 360 may include, for example, a transmit filter, a receive filter, an amplifier, a mixer, an oscillator, a DAC, and an ADC.

[0064] RF transceiver 360 may include multiple transmit / receive paths. Furthermore, RF transceiver 360 may include antenna elements. RF transceiver 360 may include at least one antenna array consisting of multiple antenna elements. In terms of hardware, RF transceiver 360 may consist of digital and analog circuitry (e.g., radio frequency integrated circuits (RFICs)). Hereinafter, the digital and analog circuitry may be implemented as a single package. Additionally, RF transceiver 360 may include multiple RF links. RF transceiver 360 may perform beamforming. To provide directionality to signals to be transmitted and received according to settings of processor 380, RF transceiver 360 may apply beamforming weights to the signals. According to embodiments, RF transceiver 360 may include a radio frequency (RF) block (or RF unit).

[0065] According to an embodiment, the RF transceiver 360 can transmit and receive signals on a wireless access network. For example, the RF transceiver 360 can transmit downlink signals. Downlink signals may include synchronization signals (SS), reference signals (RS) (e.g., cell-specific reference signals (CRS), demodulation (DM)-RS), system information (e.g., MIB, SIB, residual system information (RMSI), other system information (OSI)), configuration messages, control information, or downlink data. Additionally, for example, the RF transceiver 360 can receive uplink signals. Uplink signals may include signals associated with random access (e.g., random access preamble (RAP)) (or message 1 (Msg1), message 3 (Msg3)), reference signals (e.g., sounding reference signals (SRS), DM-RS), or power headroom reports (PHR). Although only in... Figure 3b The image shows RF transceiver 360, but according to another implementation, RU 220 may include two or more RF transceivers.

[0066] According to an embodiment, RF transceiver 460 can transmit RIM-RS. RF transceiver 460 can transmit a first type of RIM-RS (e.g., 3GPP RIM-RS type 1) to indicate the detection of remote interference. RF transceiver 460 can transmit a second type of RIM-RS (e.g., 3GPP RIM-RS type 2) to indicate the presence or absence of remote interference.

[0067] The fronthaul transceiver 365 can send and receive signals. According to an embodiment, the fronthaul transceiver 365 can send and receive signals on the fronthaul interface. For example, the fronthaul transceiver 365 can receive management plane (M-plane) messages. For example, the fronthaul transceiver 365 can receive synchronization plane (S-plane) messages. For example, the fronthaul transceiver 365 can receive control plane (C-plane) messages. For example, the fronthaul transceiver 365 can send user plane (U-plane) messages. For example, the fronthaul transceiver 365 can receive U-plane messages. Although only in... Figure 3b The image shows a fronthaul transceiver 365, but according to another implementation, the RU 220 may include two or more fronthaul transceivers.

[0068] As described above, RF transceiver 360 and fronthaul transceiver 365 transmit and receive signals. Therefore, all or some of RF transceiver 360 and fronthaul transceiver 365 may be referred to as a "communication unit," "transmitting unit," "receiving unit," or "transmit / receive unit." Furthermore, in the following description, "transmission and reception performed via a wireless channel" is used to include the meaning of the processing described above performed by RF transceiver 360.

[0069] Memory 370 stores basic programs, application programs, and data such as configuration information for the operation of RU 220. Memory 370 may be referred to as a storage unit. Memory 370 may be configured with volatile memory, non-volatile memory, or a combination of volatile and non-volatile memory. Additionally, memory 370 provides stored data upon request from processor 380. According to embodiments, memory 370 may include memory for conditions, commands, or setpoints related to the SRS transmission scheme. Memory 370, as a functional component, indicates storage space. For example, memory 370 can be understood not only to indicate memory (e.g., hard disk, flash memory, or RAM) that is part of RU 220, but also to indicate space for storing instructions and / or programs.

[0070] Processor 380 controls the overall operation of RU 220. Processor 380 can be referred to as a control unit. For example, processor 380 transmits and receives signals via RF transceiver 360 or fronthaul transceiver 365. Additionally, processor 380 writes and reads data from memory 370. Furthermore, processor 380 can perform the functions of the protocol stack required by the communication standard. Although only in... Figure 3bProcessor 380 is shown, but according to another implementation, RU 220 may include two or more processors. Processor 380, as an instruction set or code stored in memory 370, may be instructions / code that reside at least temporarily in processor 380, in the storage space storing the instructions / code, or in a portion of the circuitry constituting processor 380. Additionally, processor 380 may include various modules for performing communications. Processor 380 may control RU 220 to perform operations according to embodiments described later.

[0071] Figure 3b The configuration of RU 220 shown is merely an example, and examples of RUs performing embodiments of this disclosure are not limited to. Figure 3b The configuration shown is illustrated. In some embodiments, configurations can be added, deleted, or changed.

[0072] Figure 4 An example of functional splitting between a DU and an RU according to an embodiment is shown. With advancements in wireless communication technologies (e.g., the introduction of fifth-generation (5G) communication systems (or new radio (NR) communication systems)), the frequency bands used have increased further. As the cell radius of base stations becomes very small, the number of RUs that need to be installed has further increased. Furthermore, in 5G communication systems, the transmission capacity of the wired network sent to the fronthaul has increased significantly, as the amount of data transmitted has increased more than tenfold. Due to the factors described above, the installation cost of the wired network in a 5G communication system may increase significantly. Therefore, in order to reduce the transmission capacity of the wired network and reduce the installation cost of the wired network, a “functional split” can be used to reduce the transmission capacity of the fronthaul by transferring some functions of the DU's modem to the RU.

[0073] To reduce the burden on the DU, the role of the RU, which is currently only responsible for existing RF functions, can be expanded to include some physical layer functions. Since the RU performs higher-layer functions, its throughput increases, which can increase transmission bandwidth in the fronthaul while reducing latency requirements due to response processing. On the other hand, virtualization gain decreases and the size, weight, and cost of the RU increase due to the higher-layer functions performed by the RU. Considering the trade-offs described above, an optimal function decomposition needs to be achieved.

[0074] refer to Figure 4This illustrates the functional breakdown in the physical layer below the MAC layer. In the downlink (DL) scenario, where signals are transmitted to a terminal via a wireless network, the base station can sequentially perform channel coding / scrambling, modulation, layer mapping, antenna mapping, RE mapping, digital beamforming (e.g., precoding), iFFT conversion / CP insertion, and RF conversion. In the uplink (UL) scenario, where signals are received from a terminal via a wireless network, the base station can sequentially perform RF conversion, FFT conversion / CP removal, digital beamforming (pre-combination), RE demapping, channel estimation, layer demapping, demodulation, and decoding / descrambling. Based on the trade-offs described above, the breakdown of uplink and downlink functions can be defined in various ways, depending on inter-vendor needs, standards discussions, etc.

[0075] In the first functional split 405, the RU performs the RF function, and the DU performs the PHY function. The first functional split essentially eliminates the PHY function within the RU, and as an example, it can be referred to as Option 8. In the second functional split 410, the RU performs iFFT transformation / CP insertion in the DL of the PHY function and FFT transformation / CP removal in the UL, while the DU performs the remaining PHY functions. As an example, the second functional split 410 can be referred to as Option 7-1. In the third functional split 420a, the RU performs iFFT transformation / CP insertion in the DL of the PHY function and FFT transformation / CP removal and digital beamforming in the UL, while the DU performs the remaining PHY functions. As an example, the third functional split 420a can be referred to as Option 7-2x Category A. In the fourth functional split 420b, the RU performs digital beamforming in both the DL and UL, and the DU performs the PHY function after digital beamforming. As an example, the fourth functional split 420b can be referred to as Option 7-2x Category B. In the fifth function split 425, the RU performs RE mapping (or RE demapping) in both DL and UL, and the DU performs PHY functions after RE mapping (or RE demapping). As an example, the fifth function split 425 may be referred to as Option 7-2. In the sixth function split 430, the RU performs functions up to modulation (or demodulation) in both DL and UL, and the DU performs PHY functions after modulation (or demodulation). As an example, the sixth function split 430 may be referred to as Option 7-3. In the seventh function split 440, the RU performs functions up to encoding / scrambling (or decoding / descrambling) in both DL and UL, and the DU performs PHY functions after modulation (or demodulation). As an example, the seventh function split 440 may be referred to as Option 6. Hereinafter, embodiments of the present disclosure are described based on the sixth function split 430, but the described examples do not indicate the exclusion of the application of other function splits.

[0076] Figure 5 An example of transmitter 500 and receiver 550 is shown. Figure 5 The document describes the transmitting and receiving nodes that use channels in communication systems (e.g., wired or wireless communication systems) or broadcasting systems.

[0077] refer to Figure 5Transmitter 500 can provide information 510 to receiver 550. Transmitter 500 can perform the conversion between baseband signals and bitstreams according to physical layer standards. In order to transmit information 510 through channel 530, transmitter 500 can perform various operations. Transmitter 500 may include channel encoder 511 and modulator 513. Channel encoder 511 can encode information 510 (e.g., polar coding, low-density parity-check (LDPC)). The encoded information can be referred to as a codeword. Modulator 513 can modulate the codeword (e.g., quadrature phase shift keying (QPSK), 16 quadrature amplitude modulation (QAM), 64 QAM). The codeword can be transformed into complex symbols by modulation (which can be referred to as data symbols to distinguish them from reference signal 515, which will be described later). Transmitter 500 can input complex symbols into resource mapping and multiplexing block 517. Resource mapping and multiplexing block 517 can map complex symbols to resource elements (REs) in resource raster. In the case of multi-antenna transmission, resource mapping and multiplexing block 517 can map multiple symbols for each antenna on a RE basis. The mapping can be determined according to the multiplexing method (e.g., Orthogonal Frequency Division Multiplexing (OFDM), Discrete Fourier Transform Spread Spectrum (DFT-S) OFDM, Code Division Multiple Access (CDMA)) determined by the Radio Access Technology (RAT) between transmitter 500 and receiver 550. Transmitter 500 can transmit RF signals on channel 530 through transmit front end 519. Transmitter 500 can generate RF signals by performing signal processing on the symbols (e.g., digital-to-analog conversion (DAC) or up-conversion). Transmitter 500 can transmit RF signals on the mapped RE.

[0078] Transmitter 500 can transmit not only information 510, but also reference signals 515 for channel estimation and demodulation (e.g., coherent demodulation). For example, reference signal 515 can be a downlink reference signal. According to the 3GPP NR standard, reference signal 515 can be an SS / PBCH block, CSI-RS, demodulation-reference signal (DM-RS), or phase tracking (PT)-reference signal (RS). For example, reference signal 515 can be an uplink reference signal. According to the 3GPP NR standard, reference signal 515 can be SRS, DM-RS, or PT-RS. Transmitter 500 can provide reference signal 515 to resource mapping and multiplexing block 517. Resource mapping and multiplexing block 517 can map symbols corresponding to reference signal 515 (which may be referred to as reference symbols to distinguish them from information 510 described above) to REs in a resource grid according to the type of reference signal 515 and predefined matters in the communication protocol. In the case of multi-antenna transmission, resource mapping and multiplexing block 517 can map symbols for each antenna on a RE-by-RE basis. Transmitter 500 can transmit RF signals on channel 530 via transmit front end 519. Transmitter 500 can generate RF signals by performing signal processing on symbols (e.g., DAC or up-conversion). Transmitter 500 can transmit RF signals on the mapped RE.

[0079] The RF signal delivered on channel 530 may be damaged or have reduced gain due to background noise, interference, fading, etc. After passing through channel 530, the RF signal can be received by one or more antennas of receiver 550. Receiver 550 can receive the RF signal through receiving front-end 551. Receiver 550 can obtain complex symbols of the RF signal through signal processing (e.g., down-conversion or analog-to-digital conversion (ADC)) of receiving front-end 551. Resource demapping and demultiplexing block 553 can send information about the complex symbols to channel estimator 555 and channel equalizer 557. Channel estimator 555 can perform channel estimation. For example, channel estimator 555 can perform channel estimation using received reference symbols. Channel equalizer 557 can perform equalization on channel 530 using the channel estimation result of channel estimator 555. Channel equalizer 557 can obtain data symbols by using the equalized channel. Data symbols can be input to demodulator 559 and channel decoder 561. Receiver 550 can estimate the bit sequence of information 510 transmitted from transmitter 500 through demodulation by demodulator 559 and decoding by channel decoder 561. Demodulation can be performed according to the modulation method used in modulator 513 of transmitter 500. Decoding can be performed according to the channel coding method used in channel encoder 511 of transmitter 500.

[0080] In the following, embodiments of this disclosure relate to a technique for processing signals in a receiver 550. References described in this disclosure are as follows.

[0081] [1] N. Halko, PG Martinsson, JA Tropp "FINDING STRUCTURE WITHRANDOMNESS: PROBABILISTIC ALGORITHMS FOR CONSTRUCTING APPROXIMATE MATRIXDECOMPOSITIONS," SIAM Review, Vol 53, Issue 2, pp217-288, May 2011, [2] Yezi Huang, Wanlu Lei, Chenguang Lu, and Miguel Berg, "FronthaulFunctional Split of IRC-based Beamforming for Massive MIMO Systems," IEEEVTC2019-Fall.

[0082] Additionally, before describing embodiments of this disclosure, the following mathematical symbols may indicate the following:

[0083] Calligraphic letters (e.g., ) is used to indicate a set.

[0084] Throughout this disclosure, unless otherwise stated, it is assumed that the index of the first element of a set, sequence, or vector starts from 0 (zero-based numbering).

[0085] symbol , and These are used to indicate the set of natural numbers, the set of integers, and the set of real numbers, respectively.

[0086] For non-negative integers , Indicates from 0 to Continuous A set of integers. That is, .

[0087] Bold lowercase letters (e.g., 'a') are used to indicate vectors, and bold uppercase letters (e.g., 'A') are used to indicate matrices. In the case of vectors, unless otherwise specified, they indicate column vectors. In this case, vector a... i Represents the first... l Column vector.

[0088] For vector a and matrix A, a H and A H These represent the multiple transpose (or conjugate transpose).

[0089] In a massively multi-input multiple-output (MIMO) system, various types of receivers (e.g., receiver 550) can be used. For example, a linear receiver such as a minimum mean square error (MMSE) receiver can be used. For example, a continuous interference cancellation (SIC) receiver or a nonlinear receiver can be used, which utilizes repetitive equalization and decoding methods to anticipate maximum likelihood (ML) performance. For example, receiver 550 may include an MMSE receiver in which interference suppression combination (IRC) against interference from another cell is used.

[0090] Resource grids, separated by time and frequency axes, can include REs configured with symbols on the time axis and subcarriers on the frequency axis. Each RE is equipped with at least one receive antenna. The received signal can be represented as follows.

[0091] [Equation 1]

[0092] Here, k is the subcarrier index, and when the number of at least one or more transmission layers is hour, It is the size of And the average power is 1 Send vector. Represents a noise vector (e.g., white Gaussian noise), and This represents the interference vector. The size is , and as an interference vector The size is When the channel vector of the l-th layer is At that time, the size was The channel matrix can be represented as .

[0093] The weight vector of a full-dimensional MMSE-IRC receiver can be represented as follows.

[0094] [Equation 2]

[0095] Let represent the weight vector at subcarrier k. Here, It is the estimation of the channel matrix, and I is The identity matrix. The covariance matrix of noise and interference can be represented as follows.

[0096] [Equation 3]

[0097] This represents the covariance matrix of noise and interference. The covariance matrix of noise and interference can be estimated using a reference symbol from which the transmitted signal is known. This is obtained from the sample average. Based on Equation 3, the MMSE-IRC equalized signal can be expressed as follows.

[0098] [Equation 4]

[0099] Let represent the equalized signal at the k-th subcarrier. At this point, assume the k-th subcarrier... l The gain of the received signal is If the actual transmitted signal is known after signal detection, and the power of the transmitted signal is assumed to be 1, the quality of signal detection can be determined as in Equation 5 by treating the difference between the equalized signal in Equation 4 and the transmitted signal as an error. This corresponds to the empirical posterior signal-to-interference-plus-noise ratio (pSINR).

[0100] [Equation 5]

[0101] N represents the number of all subcarriers.

[0102] In massively multi-input multiple-output (MIMO) systems, throughput increases with the number of antennas associated with channel 530. On the other hand, the channel estimation complexity of channel 530 increases with the number of antennas. Specifically, the complexity increases during the estimation of the inverse of the channel matrix of channel 530. Various methods for reducing this complexity have been proposed. This disclosure describes a receiver technique for reducing complexity.

[0103] Random embedding Figure 6 An example of subspace decomposition using random embeddings is shown. Figure 6The operation can be performed by receiver 550. For example, receiver 550 can receive uplink signals. At least some of these operations can be performed by base station 110, DU210, or RU 220.

[0104] refer to Figure 6 In operation 601, receiver 550 can estimate the covariance matrix. The covariance matrix indicates the channel characteristics experienced by the signal received by receiver 550. The covariance matrix can be determined based on the channel matrix. For example, the covariance matrix can be determined based on the following equation.

[0105] [Equation 6]

[0106] Let represent the covariance matrix, and This indicates the number of users being served in a time slot. Let represent the estimated channel for the u-th user. Here, This represents the number of cells accumulated in the time-frequency resources to obtain covariance, and k The index represents the corresponding time-frequency resource index between 1 and N. For example, channel estimation using SRS can be performed. The time point at which the channel is estimated and the time point at which the corresponding user is served can be different, and the users paired for each time slot can also be different. Therefore, the covariance matrix for each user can be reconstructed into the covariance matrix for the user selected for the corresponding time slot. The covariance matrix of the channel for each user can be expressed as the following equation.

[0107] [Equation 7]

[0108] B can represent the covariance matrix for the user. Matrix B is an M×M matrix. This represents the user's channel matrix.

[0109] In step 603, receiver 550 may perform random embedding. Random embedding can be used to reduce receiver complexity when performing subspace decomposition of the covariance matrix. Assume the actual rank of the channel 530 between transmitter 500 and receiver 550 is K. Rank indicates the number of linearly independent vectors in the matrix of channel 530. In large-scale MIMO systems, the number of available ranks in channel 530 between transmitter 500 and receiver 550 is finite, even if the assumed number of antennas (e.g., 64) is large. For example, assume the total number of antennas is 64. If eight layers are used in the communication protocol (e.g., a maximum rank of 8 when supporting eight terminals with one transmit antenna in the same resource area, which could be a maximum of 8 when supporting DMRS-config 1 for terminals prior to NR version 18, and can support up to 24 from NR version 18 onwards), channel estimation results for other receive antennas (e.g., 56 layers) can be costly. Therefore, random embedding can be used to provide the same receiver performance as before while reducing the computational complexity of channel estimation. The dimension of the columns in the random embedding matrix can be S. S can be greater than or equal to K. The size of the covariance matrix can be changed due to the random embedding matrix. Receiver 550 can apply the random embedding matrix to the covariance matrix. For example, receiver 550 can obtain the transformed covariance matrix based on the following equation.

[0110] [Equation 8]

[0111] Let represent the covariance matrix, and This represents a random embedding matrix. It can be the size of The matrix. The transform covariance matrix is ​​represented by a randomized embedding matrix. The size of the transform covariance matrix can be reduced by using a randomized embedding matrix. Due to this reduced size, the complexity of operations using the subspace decomposition and the subspace decompositions described later (e.g., projection, precoding, precombining, and rank estimation) can be reduced.

[0112] The receiver 550 can determine the random embedding matrix. The spectral norm of the random embedding matrix can be 1. By applying the random embedding matrix, the magnitude of the spectral norm of the covariance matrix can remain unchanged. In addition, the column vectors of the random embedding matrix can be linearly independent of each other. The random embedding matrix can be understood as rotating the second dimension of the first and second dimensions of the covariance matrix. The computational complexity can be reduced by rotating the second dimension without changing the first dimension of the covariance matrix used for channel estimation of the antenna. As a non-restrictive example, the column vectors can be orthogonal to each other. For example, the random embedding matrix can be an independent and identically distributed (iid) Gaussian random matrix. For example, the random embedding matrix can be an isotropic random matrix (e.g., a Haar matrix). For example, the random embedding matrix can be a tensor matrix. For example, the random embedding matrix can be an isotropic random matrix with a specific structure (e.g., a matrix with trigonometric functions as elements, including FFT, etc.). For the random embedding matrix, see reference [1].

[0113] In operation 605, receiver 550 may perform subspace decomposition. Subspace decomposition may refer to an operation for decomposing the transformation covariance matrix obtained by the random embedding matrix into a plurality of subspaces that are orthogonal to each other. According to an embodiment, receiver 550 may perform QR decomposition. QR decomposition refers to the computational process of decomposing a matrix into an orthogonal matrix (hereinafter, the Q matrix) and an upper triangular matrix (hereinafter, the R matrix). QR decomposition refers to an orthogonalization technique performed on basis vectors using the Q matrix and the R matrix. In the present disclosure, QR decomposition is exemplified as a technique for finding orthogonal basis vectors, but decomposition techniques other than QR decomposition for finding orthogonal basis vectors may also be understood as embodiments of the present disclosure. For example, the Q matrix may be an orthogonal canonical matrix. The row vectors of the Q matrix are perpendicular to each other and have a length of 1. For example, QR decomposition may be expressed as the following equation.

[0114] [Equation 9]

[0115] This represents the transformation covariance matrix. Indicates size is A unitary matrix is ​​a matrix whose conjugate transpose is the same as its inverse. Indicates size is The upper triangular matrix (e.g., in descending order). Indicates size is The permutation matrix used for sorting.

[0116] Receiver 550 can obtain K columns (e.g., the first K columns sorted) from the columns of the Q matrix of the QR decomposition. The matrix with K columns (hereinafter, the extracted matrix) can be represented as... According to an embodiment, receiver 550 can use an extraction matrix to perform projection. Projection can be used to reduce the computational complexity between vectors. For example, receiver 550 can obtain a projection matrix. According to an embodiment, receiver 550 can use an extraction matrix to perform precoding. For example, a matrix for precoding... It can be an extraction matrix According to an embodiment, receiver 550 can use an extraction matrix to perform pre-combination. For example, a matrix for pre-combination... It can be the conjugate transpose of the extraction matrix. Even if, depending on the type of the random embedding matrix, there exists... The additional computation, in the case of QR decomposition, can also reduce the complexity from Reduced to . This represents the maximum order of computational complexity.

[0117] To select the rank more accurately, eigenvalue decomposition can be used. Since eigenvalue decomposition is performed after the dimension is reduced from M to S, complexity can be reduced. According to an embodiment, receiver 550 can perform eigenvalue decomposition. For example, eigenvalue decomposition can be performed based on the following equation.

[0118] [Equation 10]

[0119] Indicates size is It is a unitary matrix, and can correspond to the Q matrix in Equation 9. Indicates size is The matrix represents the eigenvectors. Indicates size is The matrix is ​​denoted as , and represents a diagonal matrix with eigenvalues ​​as diagonal components (e.g., ordered by the largest eigenvalues). When approximated by rank S, the eigenvectors of the covariance matrix can be derived as follows.

[0120] [Equation 11]

[0121] Indicates size is The receiver 550 can obtain K columns (e.g., based on the sorted first K column vectors) from the eigenvector matrix of the QR decomposition. The matrix with K columns (hereinafter, the extraction matrix) can be represented as... .

[0122] According to an embodiment, receiver 550 can use an extraction matrix to perform projection. Projection can be used to reduce the computational complexity between vectors. For example, receiver 550 can obtain a projection matrix. According to an embodiment, receiver 550 can use an extraction matrix to perform precoding. For example, a matrix for precoding... It can be an extraction matrix According to an embodiment, receiver 550 can use an extraction matrix to perform pre-combination. For example, a matrix for pre-combination... (In the following text, the precombined matrix) can be the conjugate transpose of the extracted matrix. According to an embodiment, receiver 550 can use the extracted matrix to perform rank estimation. For example, receiver 550 can determine the rank based on the following equation.

[0123] [Equation 12]

[0124] This represents the matrix used for rank estimation. When the values ​​of the matrix's elements are large, it can be selected as an effective flow. Indicates size is K × K And a diagonal matrix sorted in descending order of eigenvalues. For example, in the rank selection algorithm, the complexity can be reduced from... Change to . This represents the maximum order of computational complexity.

[0125] As another example, unlike the methods used in Equations 10, 11 and 12, eigenvalue decomposition can be performed based on the following equations.

[0126] [Equation 13]

[0127] Indicates size is The transformation covariance matrix. Indicates size is It is a unitary matrix, and can correspond to the Q matrix in Equation 9. This represents a random embedding matrix, and can be referenced in Equation 8. Indicates size is The matrix represents the eigenvectors. Indicates size is The matrix is ​​denoted as , and represents a diagonal matrix with eigenvalues ​​as diagonal components (e.g., ordered by the largest eigenvalues). When approximated by rank S, the eigenvectors of the covariance matrix can be derived as follows.

[0128] [Equation 14]

[0129] Indicates size is The receiver 550 can obtain K columns (e.g., based on the sorted first K column vectors) from the eigenvector matrix of the QR decomposition. The matrix with K columns (hereinafter, the extraction matrix) can be represented as... .

[0130] According to an embodiment, receiver 550 can use an extraction matrix to perform projection. Projection can be used to reduce the computational complexity between vectors. For example, receiver 550 can obtain a projection matrix. According to an embodiment, receiver 550 can use an extraction matrix to perform precoding. For example, a matrix for precoding... It can be an extraction matrix According to an embodiment, receiver 550 can use an extraction matrix to perform pre-combination. For example, a matrix for pre-combination... It can be the conjugate transpose of the extraction matrix. According to an embodiment, receiver 550 can use the extracted matrix to perform rank estimation. For example, receiver 550 can determine the rank based on the following equation.

[0131] [Equation 15]

[0132] This represents the matrix used for rank estimation. When the values ​​of the matrix's elements are large, it can be selected as an effective flow. Indicates size is And a diagonal matrix sorted in descending order of eigenvalues. For example, in the rank selection algorithm, the complexity can be reduced from... Change to . This represents the maximum order of computational complexity.

[0133] Figure 7 An example of subspace decomposition using random embeddings is shown. Figure 7 The operation can be performed by receiver 550. For example, receiver 550 can receive uplink signals. At least some of these operations can be performed by base station 110, DU210, or RU 220.

[0134] refer to Figure 7 In operation 701, receiver 550 can perform random embedding. Random embedding can be used to reduce receiver complexity when performing subspace decomposition of the covariance matrix. Assume the actual rank of the channel 530 used between transmitter 500 and receiver 550 is K. Rank indicates the number of linearly independent vectors in the matrix of channel 530. In large-scale MIMO systems, although the assumed number of antennas (e.g., 64) is large, the number of available ranks in channel 530 between transmitter 500 and receiver 550 is finite. For example, assume the total number of antennas is 64. Because a maximum rank of 8 is defined in the communication protocol, channel estimation results for other receiving antennas (e.g., 56 layers) may incur overhead. Therefore, random embedding can be used to provide the same receiver performance as before while reducing the computational complexity of channel estimation. In the random embedding matrix, the column dimension can be S. S can be greater than or equal to K. For example, receiver 550 can perform random embedding based on the following equation.

[0135] [Equation 16]

[0136] Represents the conjugate transpose of the channel matrix. This represents a random embedding matrix. This represents the conjugate transpose of the transformed channel vector. It can be the size of The size of the channel vector can be reduced by using a random embedding matrix. Due to this reduced size, the complexity of using the subspace decomposition and the subspace decompositions described later (e.g., projection, precoding, precombining, and rank estimation) can be reduced.

[0137] The spectral norm of the random embedding matrix can be 1. By applying the random embedding matrix, the magnitude of the spectral norm of the covariance matrix can remain unchanged. In addition, the column vectors of the random embedding matrix can be linearly independent of each other. The random embedding matrix can be understood as rotating the second dimension of the first and second dimensions of the covariance matrix. The computational complexity can be reduced by rotating the second dimension without changing the first dimension of the covariance matrix used for channel estimation of the antenna. As a non-restrictive example, the column vectors can be orthogonal to each other. For example, the random embedding matrix can be an independent and identically distributed (iid) Gaussian random matrix. For example, the random embedding matrix can be an isotropic random matrix (e.g., a Haar matrix). For example, the random embedding matrix can be a tensor matrix. For example, the random embedding matrix can be an isotropic random matrix with a specific structure (e.g., a matrix with trigonometric functions as elements, including FFT, etc.). For the random embedding matrix, see reference [1].

[0138] In operation 703, receiver 550 can estimate the covariance matrix of the random embedding. For example, receiver 550 can estimate the covariance matrix of the random embedding based on the following equation.

[0139] [Equation 17]

[0140] It represents the covariance matrix of random embedding, and can be called the transformation covariance matrix.

[0141] In operation 705, receiver 550 may perform subspace decomposition. Subspace decomposition may refer to operations used to decompose the covariance matrix of a random embedding into multiple subspaces that are orthogonal to each other. According to an embodiment, receiver 550 may perform QR decomposition. For example, QR decomposition may be expressed as the following equation.

[0142] [Equation 18]

[0143] This represents the covariance matrix of the random embedding. Indicates size is A unitary matrix is ​​a matrix whose conjugate transpose is identical to its inverse. Indicates size is The upper triangular matrix (e.g., in descending order). Indicates size is The permutation matrix used for sorting.

[0144] Receiver 550 can obtain the first K columns from the columns of the Q matrix of the QR decomposition. The matrix with K columns (hereinafter, the extraction matrix) can be represented as... According to an embodiment, receiver 550 can use an extraction matrix to perform projection. Projection can be used to reduce the computational complexity between vectors. For example, receiver 550 can obtain a projection matrix. According to an embodiment, receiver 550 can use an extraction matrix to perform precoding. For example, a matrix for precoding... It can be an extraction matrix According to an embodiment, receiver 550 can use an extraction matrix to perform pre-combination. For example, a matrix for pre-combination... It can be the conjugate transpose of the extraction matrix. .

[0145] To select the rank more accurately, eigenvalue decomposition can be used. Since eigenvalue decomposition is performed after the dimension is reduced from M to S, complexity can be reduced. According to an embodiment, receiver 550 can perform eigenvalue decomposition. For example, eigenvalue decomposition can be performed based on the following equation.

[0146] [Equation 19]

[0147] Indicates size is The transformation covariance matrix. Indicates size is It is a unitary matrix, and can correspond to the Q matrix in Equation 9. This represents a random embedding matrix, and can be referenced in Equation 8. Indicates size is The matrix represents the eigenvectors. Indicates size is The matrix is ​​denoted as , and represents a diagonal matrix with eigenvalues ​​as diagonal components (e.g., ordered by the largest eigenvalues). When approximated by rank S, the eigenvectors of the covariance matrix can be derived as follows.

[0148] [Equation 20]

[0149] Indicates size is The eigenvector matrix is ​​derived from the eigenvalue decomposition. The receiver 550 can obtain K column vectors from the column vectors of the eigenvector matrix of the QR decomposition. The matrix with K columns (hereinafter, the extraction matrix) can be represented as... .

[0150] According to an embodiment, receiver 550 can use an extraction matrix to perform projection. Projection can be used to reduce the computational complexity between vectors. For example, receiver 550 can obtain a projection matrix. According to an embodiment, receiver 550 can use an extraction matrix to perform precoding. For example, a matrix for precoding... It can be an extraction matrix According to an embodiment, receiver 550 can use an extraction matrix to perform pre-combination. For example, a matrix for pre-combination... It can be the conjugate transpose of the extraction matrix. According to an embodiment, receiver 550 can use the extracted matrix to perform rank estimation. For example, receiver 550 can determine the rank based on the following equation.

[0151] [Equation 21]

[0152] This represents the matrix used for rank estimation. When the values ​​of the matrix's elements are large, it can be selected as an effective flow. Indicates size is And a diagonal matrix sorted in descending order of eigenvalues. For example, in the rank selection algorithm, the complexity can be reduced from... Change to .

[0153] exist Figure 6 In [the paper], a receiver operation for performing random embedding on the covariance matrix was proposed; however, in [the context of this paper], [the paper] proposed a receiver operation for performing random embedding on the covariance matrix. Figure 7 In this paper, a receiver operation is proposed to obtain the covariance matrix after directly performing random embedding on the channel matrix. Because subspace decomposition is performed for each covariance matrix, each covariance matrix is ​​obtained as the number of covariance matrix calculations increases, and therefore, as the number of accumulated covariance matrix calculations increases, Figure 7 The receiving operation shown can be advantageous in terms of reducing complexity.

[0154] In this disclosure, complexity can be reduced and stability of subspace decomposition can be improved by performing subspace decomposition techniques such as QR decomposition and / or eigenvalue decomposition together with random embedding. Specifically, the computational complexity of the covariance matrix can be reduced because embedding is performed on instantaneous vectors in addition to embedding the covariance matrix. The subspace obtained according to the method described above can have a reduced dimension based on the number of supported layers and a rank based on the amount of strong interference. The dimension-reduced subspace can be used for projection onto the space of the signal or the space of noise and interference / downlink MIMO precoding / uplink MIMO combiner for port reduction / rank approximation of the covariance matrix, etc., and the computational complexity in each operation can be reduced. Compared to using existing maximum ratio combining (MRC) and maximum ratio transmission (MRT) of instantaneous channels, the complexity reduction technique using random embedding can support various dimensions. Furthermore, this technique can utilize the advantages of subspace-based transceivers, which can maximize the average signal-to-noise ratio (SNR) or signal-to-interference-to-noise ratio (SINR) in the supported dimensions as is. For example, when isotropic matrices with a specific structure are used for random embedding, the complexity reduction can be represented as shown in the table below.

[0155] [Table 1]

[0156] For example, if the following conditions are met ,and I If the bit error rate is fixed by the number of repetitions, then I may be related to the rank K of the channel (e.g., channel 530).

[0157] Figure 8 An example of precoding or precomposition using subspace decomposition is shown.

[0158] refer to Figure 8 RU 220 can perform channel estimation 810. RU 220 can receive a reference signal 801. For example, the reference signal 801 can be a sounding reference signal (SRS). For example, the reference signal 801 can be a DMRS (e.g., a frontload DMRS). RU 220 can perform channel estimation 810 using the reference signal 801. For example, RU 220 can obtain a channel matrix H. As an example, the size of the channel matrix H can be M×N (e.g., M is the number of receive layers and N is the number of transmit layers).

[0159] RU 220 can perform subspace decomposition 820. For subspace decomposition 820, please refer to [reference needed]. Figures 6 to 7 The description includes random embeddings (e.g., operations 603 and 703) and subspace decompositions (e.g., operations 605 and 705). RU220 can determine the covariance matrix based on the channel matrix as a result of channel estimation 810. According to an embodiment, RU 220 can determine the transformation covariance matrix based on the randomly embedded channel matrix after performing random embedding on the channel matrix. According to another embodiment, RU 220 can determine the transformation covariance matrix by performing random embedding on the covariance matrix. RU220 can perform subspace decomposition 820 on the covariance matrix or the transformation covariance matrix. For example, RU 220 can perform QR decomposition as subspace decomposition 820. For example, RU 220 can perform eigenvalue decomposition as subspace decomposition 820. RU 220 can obtain an extraction matrix through subspace decomposition 820. The extraction matrix represents the first K columns of the columns of the Q matrix configured with QR decomposition or the columns of the eigenvector matrix. K represents the rank. For example, RU 220 can obtain the extraction matrix... For example, RU 220 can obtain the extraction matrix. RU 220 can obtain a precombined matrix through subspace decomposition 820. Precombined matrix It can be an extraction matrix Conjugate transpose or extraction matrix The conjugate transpose of the matrix. For example, the size of the precombined matrix can be K×M.

[0160] RU 220 can perform RS port reduction 830. RU 220 can receive a reference signal 803. For example, the reference signal 803 can be a sounding reference signal (SRS). For example, the reference signal 803 can be a DMRS (e.g., a preload DMRS). RU 220 can perform pre-combining for port reduction against the reference signal 803. RU 220 can multiply a vector of the reference signal 803 by a pre-combining matrix for pre-combining.

[0161] [Equation 22]

[0162] This represents the reference signal to which pre-combination is performed, and y represents reference signal 803. This represents the pre-combined matrix. RU220 can send information to DU 210 about the reference signal to which it performs pre-combination. RU 220 can provide information to reduce M to K according to the subspace decomposition 820 in RU220.

[0163] The DU 210 can perform channel estimation 850. The DU 210 can perform channel estimation 850 by applying a pre-combined reference signal to it. Assuming the number of ports is L, the DU 210 can estimate a channel matrix of size K×L.

[0164] RU 220 can perform data port reduction 840. RU 220 can receive data signal 805. For example, data signal 805 may include PUSCH. RU 220 can perform pre-combining for port reduction on data signal 805. RU 220 can multiply a vector of data signal 805 by a pre-combining matrix for pre-combining.

[0165] [Equation 23]

[0166] This indicates the data signal to be pre-combined, where y represents data signal 805. This represents the pre-combined matrix. RU220 can send information to DU 210 regarding the data signals used to perform pre-combination. RU 220 can provide information to reduce M to K based on the subspace decomposition 820 in RU220.

[0167] The DU 210 can perform equalization 860. The DU 210 can perform equalization 860 on the pre-combined data signals based on the results of channel estimation 850. The DU 210 can obtain the signals of the independent paths of the channel through equalization 860.

[0168] The DU 210 can also maximize channel power based on channel estimation. Although in Figure 8 Not shown, but if interference estimation is possible, a receiver can also be configured to maximize the signal-to-interference-plus-noise ratio (SINR). For example, if maximum ratio combination (MRC) is performed, the number of ports can be reduced to L. Therefore, additional ports can be used for interference estimation. Port reduction techniques using subspace decomposition 820 facilitate the selection of a number of ports greater than the supported number of layers (e.g., rank).

[0169] In the operations described above, an example of pre-combining uplink signals (e.g., reference signal 801, reference signal 803, or data signal 805) has been described. The information obtained through subspace decomposition 820 can be applied not only to pre-combining but also to precoding. According to an embodiment, DU 210 can perform precoding 870. DU 210 can perform precoding 870 on downlink data based on the results of subspace decomposition 820. For example, DU 210 can transmit downlink data via RU 220. DU 210 can determine the precoder to be applied to the downlink data. DU 210 can extract the matrix... Or extract the matrix It has been identified as a pre-encoder.

[0170] Eigendecomposition Figure 9 An example of a receiving operation using the eigenvalue decomposition of the covariance matrix of noise and interference is shown.

[0171] refer to Figure 9 RU 220 can perform noise and interference estimation 910. RU 220 can receive a reference signal 901. For example, the reference signal 901 can be an SRS. For example, the reference signal 901 can be a DMRS. RU 220 can perform noise and interference estimation 910 using the reference signal 901. For example, RU 220 can obtain the covariance matrix R of noise and interference. nn The covariance matrix R of noise and interference nn The size can be M×M (e.g., M is the number of transmitting layers and the number of receiving layers). The covariance matrix R of noise and interference. nn The sample vectors can be used for channel and interference estimation. In embodiments of this disclosure, the covariance matrix R of noise and interference is not used as is. nn Instead of using a sample vector, eigenvalue decomposition can be performed. Through eigenvalue decomposition, the number of samples delivered from RU 220 to DU 210 can be reduced. According to an embodiment, the covariance matrix R of noise and interference... nnThe sample vector can be replaced with a noise vector using eigenvectors and eigenvalues. As the number of samples delivered to the DU 210 decreases, the number of ports considered for channel estimation and equalization in the DU 210 can be reduced in large-scale MIMO systems.

[0172] The RU 220 can perform eigenvalue decomposition. For example, the RU 220 can perform eigenvalue decomposition on the covariance matrix (e.g., the covariance matrix R of noise and interference). nn The eigenvalues ​​of the random embeddings are subjected to eigenvalue decomposition 930 (e.g., operation 605). For example, RU 220 can perform eigenvalue decomposition 930 (e.g., operation 705) on the covariance matrix of the random embeddings. For example, the covariance matrix R of noise and interference can be performed according to equations 10 to 15. nn 930.

[0173] [Equation 24]

[0174] This represents a vector consisting of the first K columns of eigenvectors configured according to the eigenvalue decomposition of the covariance matrix (e.g., the covariance matrix of noise and interference or the covariance matrix of random embeddings). This represents the feature values ​​of column K.

[0175] It can be a noise and interference vector. It can be a matrix of size M×K. With With the size reduced, the size of the information about the samples sent from RU 220 to DU 210 can be reduced. This size reduction allows for stable communication performance.

[0176] RU 220 can perform channel estimation 920. RU 220 can receive reference signal 901. RU 220 can perform channel estimation 920 using reference signal 901. For example, RU 220 can obtain channel matrix H through channel estimation 920. For example, the size of channel matrix H can be M×L (e.g., M is the number of receive layers and L is the number of transmit layers).

[0177] RU 220 can perform channel projection 940. RU 220 can perform channel projection 940 based on the results of channel estimation 920 and eigenvalue decomposition 930. For example, RU 220 can perform channel projection 940 based on the following equation.

[0178] [Equation 25]

[0179] This represents the projected channel matrix based on channel projection 940, and can have a size of M×(L+K). This represents the estimated channel vector as a result of channel estimation 920. It can have a size of M×L. This represents the estimated noise vector as a result of eigenvalue decomposition 930. It can have a size of M×K.

[0180] RU 220 can transform the data signal 903 using a matched filter 950. The transformed result can be referred to as the transformed signal. The transformed result can be provided to DU 210. For example, RU 220 can transform the data signal 903 based on the following equation.

[0181] [Equation 26]

[0182] The vector representing the transformed data signal, i.e., the transformed signal. This represents the filter matrix based on channel projection 940. y represents the received vector of the received data signal 903. RU 220 can send information about the transformed data signal to DU 210.

[0183] The DU 210 can perform channel estimation 960. For example, the DU 210 can perform channel estimation 960 based on the following equation.

[0184] [Equation 27]

[0185] This represents the estimated projection channel matrix in DU 210, and can have a size of M×(L+K). This represents the zero vector estimated in DU 210. It can have a size of M×L. This represents the noise vector estimated in DU 210. It can have a size of M×K.

[0186] The DU 210 can perform channel inversion 970. For example, the DU 210 can perform calculations based on the following equation.

[0187] [Equation 28]

[0188] This represents the inverse matrix. This represents the estimated projection channel matrix in DU 210, and can have a size of M×(L+K).

[0189] The DU 210 can perform equalization 980. For example, the DU 210 can obtain the transformed result from the RU 220 through a matched filter 950 (e.g., DU 210 can perform equalization 980 based on the following equation.

[0190] [Equation 29]

[0191] This represents the vector of the transmitted signal obtained through equalization 980. This represents the estimated projection channel matrix in DU 210, and can have a size of M×(L+K).

[0192] exist Figure 9 An example of compressing samples of the covariance matrix of noise and interference based on eigenvalue decomposition has been described. The signal received by eigenvalue decomposition can be separated into the space of the signal and the space of noise and interference. Since the column vectors obtained by eigenvalue decomposition are projected onto the channel separately, the computational complexity in DU 210 can be reduced. For example, the number of ports of data provided from RU 220 to DU 210 can be reduced compared to the method of extracting samples from the column vectors of the covariance matrix of noise and interference in reference [2]. When performing eigenvalue decomposition, it is possible to use Figures 6 to 7 The eigenvalue decomposition method for random embeddings described in the figure (e.g., the method described in Equations 10 to 15) further reduces computational complexity. Meanwhile, if more signal processing due to noise and interference is performed in the RU 220, a more simplified computation can be performed in the DU 210. For simplified computation, the null space of noise and interference can be used. The channel is projected onto the null space of noise and interference, and the projected channel can be used to reduce the number of receiver ports in the DU or for equalization. In the following, through... Figures 10 to 12 This will describe channel projection using null space.

[0193] Channel projection using null space Figure 10 An example of receiving operation using the null space of the covariance matrix of noise and interference is shown.

[0194] refer to Figure 10RU 220 can perform noise and interference estimation 910. RU 220 can receive a reference signal 901. For example, the reference signal 901 can be an SRS. For example, the reference signal 901 can be a DMRS. RU 220 can perform noise and interference estimation 910 using the reference signal 901. For example, RU 220 can obtain the covariance matrix R of noise and interference. nn The covariance matrix R of noise and interference nn The size can be M×M (for example, M is the number of transmitting layers and the number of receiving layers).

[0195] RU 220 can perform channel estimation 920. RU 220 can receive reference signal 901. RU 220 can perform channel estimation 920 using reference signal 901. For example, RU 220 can obtain channel matrix H through channel estimation 920. For example, the size of channel matrix H can be M×L (e.g., M is the number of receive layers and L is the number of transmit layers).

[0196] RU 220 can obtain a null space of 1030. The null space can indicate the null space of noise and interference. The null space of noise and interference can be obtained by subspace decomposition of the covariance matrix (or transformed covariance matrix) of noise and interference. For example, the noise and interference vectors can be defined based on the following equation.

[0197] [Equation 30]

[0198] z represents the noise and interference vector. This represents the vector of the received signal. This indicates the channel estimated based on channel estimation (e.g., channel estimation 920). This represents the vector of the transmitted signal.

[0199] Figures 6 to 7 The noise and interference vectors of the random embedding described in the text can be represented as follows.

[0200] [Equation 31]

[0201] z represents the noise and interference vector. Ω represents the random embedding matrix. Ω can be of size 1. The matrix. It can be the size of The matrix.

[0202] The covariance matrix of randomly embedded noise and interference can be represented as follows.

[0203] [Equation 32]

[0204] Let represent the covariance matrix of random embedding. z represents the noise and interference vectors. Ω represents the random embedding matrix. In the following, although the covariance matrix of random embedding is described as an example, the use of a null space without random embedding can also be understood as an embodiment of this disclosure.

[0205] To obtain the null space of the covariance matrix of a random embedding, a QR decomposition can be performed. For example, the null space can be obtained based on the following equation.

[0206] [Equation 33]

[0207] Indicates size is The unitary matrix. Indicates size is The upper triangular matrix (e.g., in descending order). Indicates size is The permutation matrix used for sorting.

[0208] [Equation 34]

[0209] This represents the projection matrix used for the null space. This represents a vector of K columns from the column vectors of the Q matrix configured with QR decomposition. K can be a pre-specified number. For example, K can be adjusted based on the number and complexity of the main interferences. When performing a QR decomposition in descending order, a null space orthogonal to the space configured with K main interferences can be configured. The projection matrix can be used to achieve interference attenuation at the full antenna resolution dimension.

[0210] The RU 220 can perform channel projection 1040. This allows for the acquisition of a projected channel where the channel is projected onto the null space of noise and interference. The number of ports for receiving signals can be reduced to the same number of receiving layers.

[0211] RU 220 can transform the data signal 903 using the matched filter 1050. The transformed result can be referred to as the transformed signal. The transformed result can be provided to DU 210. For example, RU 220 can transform the data signal 903 based on the following equation.

[0212] [Equation 35]

[0213] The vector representing the transformed data signal, i.e., the transformed signal. This represents the filter matrix based on channel projection 1040. y represents the received vector of the received data signal 903. RU 220 can send information about the transformed data signal to DU 210.

[0214] The DU 210 can perform channel estimation 1060. For example, the DU 210 can perform channel estimation 1060 based on the following equation.

[0215] [Equation 36]

[0216] This represents the estimated projection channel matrix in DU 210, and can have a size of M×L. This represents the projection matrix used for the null space. It can have a size of M×M. This represents the zero vector estimated in DU 210. It can have a size of M×L.

[0217] The DU 210 can perform channel inversion 1070. For example, the DU 210 can perform calculations according to the following equation.

[0218] [Equation 37]

[0219] This represents the inverse matrix. This represents the estimated projection channel matrix in DU 210, and can have a size of M×L.

[0220] DU 210 can perform equalization 1080. For example, DU 210 can obtain the transformed result from RU 220 through matched filter 1050 (e.g., ). DU 210 can perform equilibration 1080 based on the following equation.

[0221] [Equation 38]

[0222] This represents the vector of the transmitted signal obtained through equalization at 1080°. This represents the estimated projection channel matrix in DU 210, and can have a size of M×L.

[0223] Figure 11 An example of receiver operation using the null space of the covariance matrix of noise and interference is shown. Figure 11 The figure describes an example in which all received signal processing is performed in receiver 550 without separating DU 210 and RU 220. The same reference numerals may indicate the same description.

[0224] refer to Figure 11 Receiver 550 can perform noise and interference estimation 910. Receiver 550 can perform channel estimation 920. Receiver 550 can obtain the null space 1030. Receiver 550 can perform channel projection 1040. Receiver 550 can transform the data signal 903 using a matched filter 1050. Figure 10 Unlike other devices, receiver 550 does not need to perform delivery to another node or perform channel estimation at another node.

[0225] Receiver 550 can perform channel inversion 1170. For example, receiver 550 can perform the calculation according to the following equation.

[0226] [Equation 39]

[0227] This represents the inverse matrix. It is the result of channel projection 1040. This represents the channel projected onto the null space, i.e., the projection channel matrix. It can have a size of M×L.

[0228] Receiver 550 can perform equalization 1180. For example, receiver 550 can obtain the transformed result through matched filter 1050 (e.g., ). DU 210 can perform equilibration 1080 based on the following equation.

[0229] [Equation 40]

[0230] This represents the vector of the transmitted signal obtained through equalization 980. This represents the projection channel matrix and can have a size of M×L. Regarding the data signal 903, the operation of the receiver 550 can be understood as applying a specific filter. For example, the specific filter can be expressed as the following equation.

[0231] [Equation 41]

[0232] Indicates a specific filter.

[0233] Figure 12An example of a receive operation using weighted interpolation is shown. Figure 10 and Figure 11 An example in which a matched filter (e.g., matched filter 950 or matched filter 1050) is applied to the received data signal 903 has already been described. However, projecting the null space onto the estimated channel for each of all received symbols can increase the computational load. Figure 12 The diagram describes the operation of applying weights (or filters) to the received signal via weight interpolation without null space calculations for the received signal. At least a portion of the operation of receiver 550 can be performed by DU 210 or RU 220. The same reference numerals can denote the same description.

[0234] refer to Figure 12 Receiver 550 can perform noise and interference estimation 910. RU 220 can receive a reference signal 901. For example, reference signal 901 can be an SRS. For example, reference signal 901 can be a DMRS. RU 220 can perform noise and interference estimation 910 using reference signal 901. Receiver 550 can perform channel estimation 920. RU 220 can receive reference signal 901. RU 220 can perform channel estimation 920 using reference signal 901. For example, RU 220 can obtain the channel matrix H through channel estimation 920. As an example, the channel matrix H can have a size of M×L (e.g., M is the number of receive layers, and L is the number of transmit layers). Receiver 550 can obtain a null space 1030. The null space can be obtained through subspace decomposition of the covariance matrix (or transformed covariance matrix) of noise and interference. For example, noise and interference vectors can be defined based on the following equation. Receiver 550 can perform channel projection 1040. The projection channel can be obtained by projecting the null space of noise and interference onto the channel. The number of ports for receiving signals can be reduced to the same number of receiving layers. As a non-limiting example, in the case where channel projection 1040 is performed only for DMRS symbols, channel interpretation (e.g., channel interpolation) can be performed for other symbols. Receiver 550 can obtain information about the channel or projected channel at each time-frequency resource through channel interpretation (e.g., channel interpolation).

[0235] Receiver 550 can perform channel inversion (1220). For example, receiver 550 can obtain the inverse matrix. .

[0236] Receiver 550 can perform weight calculation 1230. Receiver 550 can calculate weights based on the inverse matrix and the projected channel matrix. For example, receiver 550 can calculate weights based on the following equation.

[0237] [Equation 42]

[0238] Indicates the weight. This represents the projection matrix used for the null space. It represents the projection channel matrix and can have a size of M×L.

[0239] Receiver 550 can perform weighted interpolation 1240. Through weighted interpolation 1240, receiver 550 can obtain information about the channel experienced by the actual received data by performing interpolation in the frequency domain or time domain based on information about the channel experienced by the reference signal 901. For example, linear interpolation methods or high-dimensional interpolation methods can be used based on a time-frequency resource grid. For example, receiver 550 can obtain a weight matrix through weighted interpolation 1240 with respect to the weights in Equation 42. .

[0240] Receiver 550 can perform equalization 1250.

[0241] [Equation 43]

[0242] This represents the vector of the transmitted signal obtained through equalization 1250. This represents the weight matrix obtained by interpolating the weights by 1240. y represents the data signal 903.

[0243] exist Figure 12 An example of using a channel or weight interpolation method has been described where the filter that projects the channel only onto the null space of noise and interference is known only for some symbols or some subcarriers, or where the receive weight filter for only some symbol subcarriers is known. For example, the channel may change relatively slowly. The received signal may change for each symbol (time domain unit) and for each subcarrier (frequency domain unit). Therefore, if filtering is performed for all received signals, filtering is performed multiple times even when the amount of channel change is small, and thus the complexity may increase. Therefore, the computational complexity can be improved by weight interpolation 1240. In addition, the receiver 550 only needs to interpolate the channel filter projected onto the null space of noise and interference, without interpolating the channel separately to obtain the weights. The properties of the projected channel matrix can satisfy the following equation.

[0244] [Equation 44]

[0245]

[0246] This represents the projection matrix used for the null space.

[0247] Because samples with noise and interference are used in reference [2], performance degradation is required to reduce the number of receiver ports of DU210. The number of receiver ports can represent the number of streams effective for processing the received signal and can be associated with the complexity of the equalizer of DU210 or receiver 550. In embodiments of this disclosure, the number of receiver ports of DU210 can be reduced to the number of transmit layers and interference layers through eigenvalue decomposition. In addition, in embodiments of this disclosure, the number of receiver ports of DU210 can be reduced to the number of transmit layers through null space projection. When the channel of the signal is projected onto the null space and a filter according to this is used, the computational complexity of receiver 550 or DU210 can be reduced. Specifically, it is not just about increasing the power of the received signal, but about improving the reception performance (e.g., SINR) across the entire dimension of the receiving antenna. It is possible to reduce complexity while achieving the performance of minimum mean square error interference suppression combination (MMSE-IRC) across the entire dimension. In addition, the characteristics of the projection matrix (e.g., This reduces complexity in weight calculation. Since channel projection of the null space is not required for each of the received signals, the complexity of weight calculation for the received signals can be further reduced.

[0248] Figure 13a and Figure 13b An example of the performance of the receive operation using random embedding and null space is shown.

[0249] refer to Figure 13a and Figure 13bGraph 1300 indicates the cumulative distribution function (CDF) based on pSINR. The horizontal axis of graph 1300 indicates pSINR (in dB), and the vertical axis indicates CDF. Graph 1350 indicates CDF based on pSINR. The horizontal axis of graph 1350 indicates pSINR (in dB), and the vertical axis indicates CDF. Graph 1350 may correspond to region 1310 of graph 1300. The first line 1301 indicates the reception performance of the MMSE-IRC receiver across the entire dimension based on the ideal noise and interference matrix. The second line 1302 indicates the reception performance of the MMSE-IRC receiver across the entire dimension based on the estimated noise and interference matrix. The third line 1303 indicates the reception performance according to the method in reference [2] when the number of noise and interference samples is 8 (e.g., in reference [2], the receiver (RU and / or DU) obtains noise and interference samples by extracting samples from the column vectors of the noise and interference covariance matrix, changes the channel by the noise and interference samples, and performs MMSE reception operation according to the changed channel). The fourth line 1304 indicates the reception performance according to the method in reference [2] when the number of noise and interference samples is 5. The fifth line 1305 indicates the reception performance according to the method in reference [2] when the number of noise and interference samples is 2. The sixth line 1306 indicates the reception performance according to the method in reference [2] when the number of noise vectors is 2. Figure 9 The receiving performance of the method shown. The seventh line, 1307, indicates the performance when the number of noise vectors is 2. Figure 10 The receiving performance of the method shown.

[0250] The techniques in reference [2] (e.g., methods for extracting samples from the column vectors of the covariance matrix of noise and interference) require a sufficiently large number of samples relative to the number of interference streams. For example, for two interference streams, the technique using 8 samples (e.g., the third line 1303) shows approximately 1 dB of performance degradation compared to the proposed techniques (e.g., the sixth line 1306 and / or the seventh line 1307). According to the proposed techniques, only two paths can be selected for two strong interference streams. Therefore, the computational complexity in DU 210 can be reduced because only the null space form or two large eigenvectors are considered. Improved performance is shown compared to existing MMSE IRC (e.g., the second line 1302) as the accuracy of the estimation of the noise and interference covariance is improved through the eigenvalue decomposition of the noise and interference covariance matrix. In addition, the complexity can be further reduced because the path of the interference stream is not selected in the receiving techniques using random embedding and null space (e.g., the seventh line 1307). It can be confirmed that the receiving technique demonstrates performance comparable to that of MMSE techniques based on eigenvalue decomposition (e.g., the sixth line 1301).

[0251] Random embedding can be performed on the covariance matrix itself, or on the channel or noise vector used to obtain the covariance matrix. Spatial decomposition techniques can be used to separate the spatial distribution of interference and noise from the spatial distribution of the signal, or filters can be used to project the estimated channel onto a null space orthogonal to the spatial distribution of interference and noise. With this filtering, the equalizer complexity can be reduced while decreasing the number of ports on the DU 210 by as much as the number of transmit layers, thus reducing the impact of interference.

[0252] The effects that can be obtained from this disclosure are not limited to those described above, and any other effects not mentioned herein will be clearly understood by those skilled in the art to which this disclosure pertains, based on the following description.

[0253] In one embodiment, a device for a radio unit (RU) is provided. The device may include a radio frequency (RF) transceiver, a fronthaul transceiver, a memory storing instructions, and a processor. When executed by the processor, the instructions enable the device to: obtain a reference signal via the RF transceiver; obtain a transformed covariance matrix of the reference signal for noise and interference via random embedding for dimensionality reduction; obtain an extraction matrix via subspace decomposition of the transformed covariance matrix; perform pre-combining on the uplink signal obtained via the RF transceiver using the extraction matrix; and transmit data of the pre-combined uplink signal to a digital unit (DU) via the fronthaul transceiver. The random embedding can be used to transform a first dimension having a number of receiver layers with the RU into a second dimension fewer than the number of receiver layers.

[0254] According to an embodiment, when executed by a processor, the instructions enable the device to obtain the noise and interference covariance matrix of the channel of the reference signal, and to obtain the transform covariance matrix by multiplying the random embedding matrix used for random embedding with the noise and interference covariance matrix. Each column vector of the random embedding matrix may be linearly independent.

[0255] According to an embodiment, when the instructions are executed by the processor to obtain the transform covariance matrix, the device can obtain a channel matrix for the reference signal, obtain a randomly embedded channel matrix by multiplying the random embedding matrix for random embedding by the channel matrix, and obtain the noise and interference covariance matrix of the channel matrix for random embedding as the transform covariance matrix. Each column vector of the random embedding matrix can be linearly independent.

[0256] According to an embodiment, when the instructions are executed by the processor to obtain the extraction matrix, the device can perform a QR decomposition on the transform covariance matrix and extract at least one column vector from the column vectors of the orthogonal matrix, as many as a specified number K. The extraction matrix may correspond to the extracted at least one column vector.

[0257] According to an embodiment, when the instructions are executed by the processor to obtain the extraction matrix, the device can perform eigenvalue decomposition on the transform covariance matrix and extract at least one column vector as many as a specified number K from the column vectors of the eigenvector matrix according to the eigenvalue decomposition. The extraction matrix may correspond to the extracted at least one column vector.

[0258] According to an embodiment, random embedding may include multiplication of random embedding matrices. The random embedding matrix may be an independent and identically distributed (iid) Gaussian random matrix, an isotropic random matrix, or a tensor matrix.

[0259] According to an embodiment, the uplink signal may include at least one of the Physical Uplink Shared Channel (PUSCH) signal, the Sound Reference Signal (SRS) or the Uplink Demodulation Reference Signal (DMRS).

[0260] According to an embodiment, when the instructions are executed by the processor, the device can: determine a precoding matrix for the downlink signal based on the extraction matrix, generate a transmit signal by applying the precoding matrix to the downlink signal, and transmit the transmit signal via an RF transceiver.

[0261] According to an embodiment, when the instructions are executed by the processor, the device can: obtain a projection matrix for the null space based on the extraction matrix, obtain a filter matrix by applying the projection matrix to a channel matrix for a reference signal, and perform pre-combination by multiplying the filter matrix with an uplink signal.

[0262] In one embodiment, a device for a digital unit (DU) is provided. The device may include a transceiver, a memory storing instructions, and a processor. When executed by the processor, the instructions enable the device to: obtain a reference signal from a radio unit (RU) via the transceiver; obtain a transformed covariance matrix of the reference signal for noise and interference through random embedding for dimensionality reduction; obtain an extraction matrix through subspace decomposition of the transformed covariance matrix; and obtain a transmitted signal by performing equalization on the uplink signal from the RU using the extraction matrix. The random embedding can be used to transform a first dimension having a number of receiver layers with RUs into a second dimension having fewer than the number of receiver layers.

[0263] In one embodiment, a device for a radio unit (RU) is provided. The device may include a radio frequency (RF) transceiver, a fronthaul transceiver, a memory storing instructions, and a processor. When executed by the processor, the instructions enable the device to: obtain a covariance matrix of noise and interference from a reference signal received via the RF transceiver; obtain a noise vector matrix by performing eigenvalue decomposition of the covariance matrix; obtain a filter matrix using the noise vector matrix and a channel matrix for the reference signal; obtain a transformed signal by applying the filter matrix to an uplink signal obtained via the RF transceiver; and transmit the transformed signal data to a digital unit (DU) via the fronthaul transceiver. The number of columns in the noise vector matrix may be less than the number of receiver layers in the RU.

[0264] According to an embodiment, when the instructions are executed by the processor, the device can: obtain an eigenvector matrix and eigenvalues ​​by performing eigenvalue decomposition of the covariance matrix, and obtain a noise vector matrix by extracting a specified number of column vectors from the column vectors of the eigenvector matrix and the eigenvalues ​​corresponding to the extracted column vectors.

[0265] According to an embodiment, the noise vector matrix can be determined based on the following equation.

[0266] [Equation 45]

[0267] Represents the noise vector matrix, This represents a matrix configured with extracted column vectors corresponding to a specified number K, and This represents a diagonal matrix containing eigenvalues ​​corresponding to column vectors extracted according to a specified number K.

[0268] According to an embodiment, the number of rows in the filter matrix can correspond to the number of receiving layers in the RU. The number of columns in the filter matrix can correspond to the sum of the number of transmitting layers and the number of columns in the noise vector matrix.

[0269] According to an embodiment, the reference signal may include an uplink demodulation reference signal (DMRS). The uplink signal may include a physical uplink shared channel (PUSCH) signal.

[0270] In an embodiment, a radio unit (RU) device may include a radio frequency (RF) transceiver, a fronthaul transceiver, a memory storing instructions, and a processor. When executed by the processor, the instructions can cause the device to: obtain a reference signal via the RF transceiver; obtain a transformed covariance matrix of noise and interference via the reference signal; obtain an extraction matrix with a specified number of column vectors in the column vectors of the orthogonal matrix of the QR decomposition by performing QR decomposition of the covariance matrix; obtain a filter matrix using a projection matrix obtained from the extraction matrix and a channel matrix for the reference signal; obtain a transformed signal by applying the filter matrix to an uplink signal obtained via the RF transceiver; and transmit the transformed signal data to a digital unit (DU) via the fronthaul transceiver.

[0271] According to an embodiment, the projection matrix can be determined based on the following equation.

[0272] [Equation 46]

[0273] Let I denote the projection matrix, and let I denote the identity matrix. This indicates the extraction vector.

[0274] According to an embodiment, the filter matrix can be determined based on the following equation.

[0275] [Equation 47]

[0276] Represents the filter matrix, Let represent the projection matrix, and This represents the channel matrix.

[0277] According to an embodiment, the number of rows in the filter matrix can correspond to the number of receiving layers in the RU. The number of columns in the filter matrix can correspond to the number of transmitting layers.

[0278] According to an embodiment, the reference signal may include an uplink demodulation reference signal (DMRS). The uplink signal may include a physical uplink shared channel (PUSCH) signal.

[0279] For one or more embodiments, at least one component described in one or more of the preceding figures may be configured to perform one or more operations, techniques, processes, and / or methods as described in this disclosure. For example, a processor (e.g., a baseband processor) described in association with one or more of the preceding figures may be configured to operate according to one or more examples described in this disclosure. As another example, circuitry associated with a user equipment (UE), base station, network element, etc., as described in association with one or more of the preceding figures may be configured to operate according to one or more examples described herein.

[0280] Unless otherwise explicitly stated, any of the embodiments described above may be combined with any other embodiment (or combination of embodiments). The above description of one or more implementations is provided for illustration and explanation, but is not intended to limit the scope of the embodiments or to be exhaustive to the precise form disclosed. Modifications and variations are possible in light of the above teachings or may be obtained through practice with various embodiments.

[0281] The methods described in the claims or specification of this disclosure can be implemented in hardware, software, or a combination of hardware and software.

[0282] In the case of a software implementation, a computer-readable storage medium may be provided for storing one or more programs (software modules). One or more programs stored in the computer-readable storage medium are configured to be executed by one or more processors in an electronic device. One or more programs include instructions that cause the electronic device to perform methods according to embodiments described in the claims or specification of this disclosure. One or more programs may be included and provided in a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable storage medium (e.g., a compact disc read-only memory (CD-ROM)), or distributed online (e.g., downloaded or uploaded) via an app store (e.g., Play Store™), or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product may be temporarily generated or at least temporarily stored in a machine-readable storage medium (such as the memory of a manufacturer's server, an app store's server, or a relay server).

[0283] Such programs (software modules, software) can be stored in random access memory, non-volatile memory (including flash memory, read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM)), disk storage devices, optical storage devices (e.g., compact disc-ROM (CD-ROM), digital universal disc (DVD), or other formats), or cartridges. Alternatively, they can be stored in some or all of these memories. Additionally, multiple configuration memories may be included.

[0284] Additionally, the program can be stored in an attachable storage device accessible via a communication network such as the Internet, intranet, local area network (LAN), wide area network (WAN), or storage area network (SAN), or a combination thereof. Such a storage device can be connected to a device executing embodiments of this disclosure via an external port. Furthermore, a separate storage device on the communication network can also be connected to a device executing embodiments of this disclosure.

[0285] In the specific embodiments described above in this disclosure, the components included in this disclosure are expressed in either a singular or plural form according to the presented specific embodiments. However, the singular or plural representation may be appropriately chosen depending on the circumstances presented for ease of explanation, and this disclosure is not limited to a singular or plural number of components, and may even be configured in a singular manner with components expressed in a plural manner, or may be configured in a plural manner with components expressed in a singular manner.

[0286] According to various embodiments, one or more components or operations described above may be omitted, or one or more additional components or operations may be added. Alternatively or additionally, multiple components (e.g., modules or programs) may be integrated into a single component. In such a case, the integrated component may still perform one or more functions of each of the multiple components in the same or similar manner as they were performed by their counterparts among the multiple components prior to integration. According to various embodiments, operations performed by a module, program, or other component may be run sequentially, in parallel, repeatedly, or heuristically, or they may be run in a different order, or one or more operations may be omitted, or one or more additional operations may be added.

[0287] Furthermore, specific embodiments have been described in detail in this disclosure, and of course, various modifications are possible without departing from the scope of this disclosure.

Claims

1. A device for a radio unit RU, the device comprising: RF transceiver; Fronthaul transceiver; The memory stores instructions; as well as processor, Wherein, when the instruction is executed by the processor, the device: The reference signal is obtained through the RF transceiver; The transformed covariance matrix of the noise and interference of the reference signal is obtained by random embedding for dimensionality reduction; The extraction matrix is ​​obtained by subspace decomposition of the transformation covariance matrix; The extraction matrix is ​​used to pre-combine the uplink signals obtained through the RF transceiver; and The data is transmitted to the digital unit (DU) via the fronthaul transceiver to perform the pre-combined uplink signal on it. The random embedding is used to transform a first dimension with the number of receiver layers having the RU into a second dimension with a number less than the number of receiver layers.

2. The apparatus of claim 1, wherein, When the instruction is executed by the processor, it causes the device to: Obtain the noise and interference covariance matrix of the channel of the reference signal, and The transformation covariance matrix is ​​obtained by multiplying the random embedding matrix used for the random embedding with the noise and interference covariance matrix, and In this context, each column vector of the random embedding matrix is ​​linearly independent.

3. The apparatus of claim 1, wherein, When the instructions are executed by the processor to obtain the transformation covariance matrix, the device: Obtain the channel matrix for the reference signal; The channel matrix for random embedding is obtained by multiplying the random embedding matrix used for the random embedding with the channel matrix; The noise and interference covariance matrices of the channel matrix used for the random embedding are obtained as the transform covariance matrix, and In this context, each column vector of the random embedding matrix is ​​linearly independent.

4. The apparatus of claim 1, wherein, When the instructions are executed by the processor to obtain the extraction matrix, the device: Perform QR decomposition on the transformed covariance matrix, and According to the QR decomposition, at least one column vector, equal to the specified number K, is extracted from the column vectors of the orthogonal matrix, and The extraction matrix corresponds to the extracted at least one column vector.

5. The device according to claim 1, wherein, When the instructions are executed by the processor to obtain the extraction matrix, the device: Perform eigenvalue decomposition on the transformed covariance matrix, and Based on the eigenvalue decomposition, at least one column vector, equal to the specified value K, is extracted from the column vectors of the eigenvector matrix. The extraction matrix corresponds to the extracted at least one column vector.

6. The device according to claim 1, in, The random embedding includes the multiplication operation of the random embedding matrix, and The random embedding matrix is ​​an independent and identically distributed (i.i.d.) Gaussian random matrix, an isotropic random matrix, or a tensor matrix.

7. The device according to claim 1, wherein, The uplink signal includes at least one of the Physical Uplink Shared Channel (PUSCH) signal, the Sound Reference Signal (SRS) or the Uplink Demodulation Reference Signal (DMRS).

8. The device according to claim 1, wherein, When the instruction is executed by the processor, it causes the device to: Based on the extracted matrix, a precoding matrix for the downlink signal is determined. The transmitted signal is generated by applying the precoding matrix to the downlink signal, and The transmit signal is transmitted via the RF transceiver.

9. The device according to claim 1, wherein, When the instruction is executed by the processor, it causes the device to: Based on the extracted matrix, a projection matrix for the null space is obtained. The filter matrix is ​​obtained by applying the projection matrix to the channel matrix used for the reference signal. The pre-combination is performed by multiplying the filter matrix with the uplink signal.

10. A device for a radio unit RU, the device comprising: RF transceiver; Fronthaul transceiver; The memory stores instructions; as well as processor, Wherein, when the instruction is executed by the processor, the device: The covariance matrix of noise and interference is obtained by using the reference signal received via the RF transceiver; The noise vector matrix is ​​obtained by performing eigenvalue decomposition of the covariance matrix. The filter matrix is ​​obtained using the noise vector matrix and the channel matrix for the reference signal; The transformed signal is obtained by applying the filter matrix to the uplink signal obtained through the RF transceiver; and The data of the transformed signal is transmitted to the digital unit (DU) via the fronthaul transceiver. The number of columns in the noise vector matrix is ​​less than the number of receiving layers in the RU.

11. The device according to claim 10, wherein, When the instruction is executed by the processor, it causes the device to: The eigenvector matrix and eigenvalues ​​are obtained by performing the eigenvalue decomposition of the covariance matrix; The noise vector matrix is ​​obtained through the following: The extracted column vectors corresponding to a specified number of column vectors in the feature vector matrix, and The feature values ​​corresponding to the extracted column vectors.

12. The device according to claim 10, in, The number of rows in the filter matrix corresponds to the number of receiver layers in the RU. Wherein, the number of columns in the filter matrix corresponds to the sum of the number of transmission layers and the number of columns in the noise vector matrix. The reference signal includes the uplink demodulation reference signal DMRS, and The uplink signal includes the Physical Uplink Shared Channel (PUSCH) signal.

13. A device for a radio unit RU, the device comprising: RF transceiver; Fronthaul transceiver; The memory stores instructions; as well as processor, Wherein, when the instruction is executed by the processor, the device: The reference signal is obtained through the RF transceiver; The transformed covariance matrix of noise and interference is obtained through the reference signal; By performing QR decomposition of the covariance matrix, an extraction matrix with a specified number of column vectors is obtained from the column vectors of the orthogonal matrix of the QR decomposition. The filter matrix is ​​obtained using the projection matrix obtained from the extraction matrix and the channel matrix for the reference signal; The transformed signal is obtained by applying the filter matrix to the uplink signal obtained through the RF transceiver; and The data of the transformed signal is transmitted to the digital unit (DU) via the fronthaul transceiver.

14. The device according to claim 13, wherein, The projection matrix is ​​determined based on the following equation: Let I represent the projection matrix, and let 1 represent the identity matrix. This indicates the extraction matrix, and The filter matrix is ​​determined based on the following equation: Represents the filter matrix, Let the projection matrix be represented, and This represents the channel matrix.

15. The device according to claim 13, in, The number of rows in the filter matrix corresponds to the number of receiver layers in the RU. The number of columns in the filter matrix corresponds to the number of transmission layers. The reference signal includes the uplink demodulation reference signal DMRS, and The uplink signal includes the Physical Uplink Shared Channel (PUSCH) signal.

Images