WO2020221126A1 - 提升多用户复用性能的方法、装置、设备和存储介质 - Google Patents

提升多用户复用性能的方法、装置、设备和存储介质 Download PDF

Info

Publication number
WO2020221126A1
WO2020221126A1 PCT/CN2020/086711 CN2020086711W WO2020221126A1 WO 2020221126 A1 WO2020221126 A1 WO 2020221126A1 CN 2020086711 W CN2020086711 W CN 2020086711W WO 2020221126 A1 WO2020221126 A1 WO 2020221126A1
Authority
WO
WIPO (PCT)
Prior art keywords
terminal
srs
base sequence
channel
frequency domain
Prior art date
Application number
PCT/CN2020/086711
Other languages
English (en)
French (fr)
Inventor
郭森宝
李桂宝
乐春晖
王勃
秦恒
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to EP20799301.5A priority Critical patent/EP3952544A4/en
Publication of WO2020221126A1 publication Critical patent/WO2020221126A1/zh
Priority to US17/513,666 priority patent/US12101205B2/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/0224—Channel estimation using sounding signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/373—Predicting channel quality or other radio frequency [RF] parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413—MIMO systems
    • H04B7/0426—Power distribution
    • H04B7/0434—Power distribution using multiple eigenmodes
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413—MIMO systems
    • H04B7/0452—Multi-user MIMO systems
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04J—MULTIPLEX COMMUNICATION
    • H04J11/00—Orthogonal multiplex systems, e.g. using WALSH codes
    • H04J11/0023—Interference mitigation or co-ordination
    • H04J11/0026—Interference mitigation or co-ordination of multi-user interference
    • H04J11/0036—Interference mitigation or co-ordination of multi-user interference at the receiver
    • H04J11/0046—Interference mitigation or co-ordination of multi-user interference at the receiver using joint detection algorithms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/022—Channel estimation of frequency response
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/0224—Channel estimation using sounding signals
    • H04L25/0228—Channel estimation using sounding signals with direct estimation from sounding signals
    • H04L25/023—Channel estimation using sounding signals with direct estimation from sounding signals with extension to other symbols
    • H04L25/0232—Channel estimation using sounding signals with direct estimation from sounding signals with extension to other symbols by interpolation between sounding signals
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L25/00—Baseband systems
    • H04L25/02—Details ; arrangements for supplying electrical power along data transmission lines
    • H04L25/0202—Channel estimation
    • H04L25/024—Channel estimation channel estimation algorithms
    • H04L25/025—Channel estimation channel estimation algorithms using least-mean-square [LMS] method
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00—Arrangements affording multiple use of the transmission path
    • H04L5/003—Arrangements for allocating sub-channels of the transmission path
    • H04L5/0048—Allocation of pilot signals, i.e. of signals known to the receiver
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W76/00—Connection management
    • H04W76/20—Manipulation of established connections
    • H04W76/27—Transitions between radio resource control [RRC] states
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00—Arrangements affording multiple use of the transmission path
    • H04L5/003—Arrangements for allocating sub-channels of the transmission path
    • H04L5/0037—Inter-user or inter-terminal allocation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L5/00—Arrangements affording multiple use of the transmission path
    • H04L5/003—Arrangements for allocating sub-channels of the transmission path
    • H04L5/0048—Allocation of pilot signals, i.e. of signals known to the receiver
    • H04L5/0051—Allocation of pilot signals, i.e. of signals known to the receiver of dedicated pilots, i.e. pilots destined for a single user or terminal
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/08—Testing, supervising or monitoring using real traffic

Definitions

  • This application relates to communication technology, and in particular, to a method, device, device, and storage medium for improving multiple user multiplexing performance.
  • TDD Time Division Duplexing
  • MIMO Massive Multiple Input Multiple Output
  • MU Multiple User MIMO
  • This deterioration is firstly the deterioration of the performance of the mobile users themselves, and then the deterioration of the performance of the entire network.
  • the single-user weight is first estimated, and then zero-forcing is used, or a zero-forcing-based method is used for orthogonalization to obtain the multi-user weight. That is, the precoding weight.
  • the above method is very sensitive to the accuracy of the channel.
  • the single user weight calculation is based on the most recent Sounding Reference Signal (SRS) information to obtain the channel, and the time interval for sending SRS information is relatively long, generally a few
  • SRS Sounding Reference Signal
  • the user scheduling cycle is generally on the order of ms.
  • the embodiments of the present application provide a method, device, device, and storage medium for improving the multiplexing performance of multiple users, so as to solve the problem that the existing MU MIMO algorithm causes performance degradation of mobile users.
  • the first aspect of the present application provides a method for improving multi-user multiplexing performance, which is applied to network equipment, and the method includes:
  • Channel prediction is performed according to the channel information of each terminal to obtain the channel prediction result.
  • the method further includes:
  • the movement state of the terminal is acquired, the movement state is used to indicate the movement speed of the terminal; the movement state includes: a quasi-static state, a low-speed movement state, or a medium-high speed Mobile state
  • the weight of the terminal is calculated and acquired.
  • the channel prediction result of the terminal includes an SRS channel feature vector or weight
  • the calculation and acquisition of the weight of the terminal according to the channel prediction result and motion state of each terminal includes:
  • the motion state of the terminal is a quasi-static state, perform type-spatial and/or time-domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the motion state of the terminal is a low-speed motion state, perform Type 2 spatial and/or time domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the second step distance calculation is performed on the channel of the terminal to generate the weight of the terminal.
  • the performing type-spatial and/or time-domain filtering on the SRS channel prediction vector or weight of the terminal includes:
  • the minimum mean square filtering is performed on the SRS prediction weight of the terminal based on the subspace distance and gradient descent method.
  • the performing Type 2 spatial and/or time domain filtering on the SRS channel prediction vector or weight of the terminal includes:
  • a prediction algorithm based on autoregressive AR filtering filters the SRS channel prediction vector or weight of the terminal.
  • the performing SRS detection on the multiple terminals based on the quasi-orthogonal sequence and acquiring the channel information for each terminal to send SRS includes:
  • the pretreatment includes at least one of the following treatments:
  • Equation (3) Perform time-domain interference cancellation on the frequency-domain signal to be processed at the received SRS position; where y is the frequency-domain signal to be processed at the SRS position, and the dimension is N ⁇ 1; Is the time-domain cancellation transformation function, It is the frequency domain signal after time domain cancellation, the dimension dimension is N ⁇ 1, and N is the length of channel estimation.
  • the channel information of each terminal includes the SRS frequency domain channel estimation results of some subbands, and the channel prediction is performed according to the channel information of each terminal to obtain the channel prediction results, including:
  • the full bandwidth of the SRS is predicted based on frequency domain extrapolation, and the channel prediction result of each terminal is obtained.
  • the performing channel prediction based on the channel information of each terminal to obtain the channel prediction result includes:
  • the channel prediction is performed on the SRS based on the two-dimensional time-frequency-space mutual calibration to obtain the channel prediction result of each terminal.
  • the configuring base sequence identifiers for multiple terminals through RRC signaling includes:
  • the second RRC signaling is sent to the terminal; the second RRC signaling carries an index identifier, and the index identifier is used to instruct the terminal to use the at least one base sequence identifier The first base sequence identifier for SRS transmission.
  • the configuring base sequence identifiers for multiple terminals through RRC signaling includes:
  • the first RRC signaling is sent to each terminal, and the first RRC signaling carries a base sequence identifier, and the base sequence identifier indicates a candidate collection of all SRS base sequence identifiers.
  • the second aspect of the present application provides a method for improving multiple user multiplexing performance, which is applied to a terminal, and the method includes:
  • the sounding reference signal SRS is sent.
  • the obtaining the base sequence identifier configured by the network device through RRC signaling includes:
  • the third aspect of the present application provides an apparatus for improving multiple user multiplexing performance, including:
  • the sending module is used to configure base sequence identifiers for multiple terminals through RRC signaling, and the base sequences indicated by the base sequence identifier of each terminal do not have orthogonal characteristics;
  • a processing module configured to perform SRS detection on the multiple terminals based on the quasi-orthogonal sequence, and obtain channel information for each terminal to send SRS;
  • the processing module is also used to perform channel prediction according to the channel information of each terminal to obtain a channel prediction result.
  • processing module is also used for:
  • the movement state of the terminal is acquired, and the movement state is used to indicate the movement speed of the terminal;
  • the movement state includes: a quasi-static state, a low-speed movement state, or a medium-high speed state Mobile state
  • the weight of the terminal is calculated and acquired.
  • the channel prediction result of the terminal includes the SRS channel feature vector or weight
  • the processing module is specifically configured to:
  • the motion state of the terminal is a quasi-static state, perform type-spatial and/or time-domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the motion state of the terminal is a low-speed motion state, perform Type 2 spatial and/or time domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the second step distance calculation is performed on the channel of the terminal to generate the weight of the terminal.
  • processing module is specifically used for:
  • the minimum mean square filtering is performed on the SRS prediction weight of the terminal based on the subspace distance and gradient descent method.
  • processing module is specifically used for:
  • a prediction algorithm based on autoregressive AR filtering filters the SRS channel prediction vector or weight of the terminal.
  • processing module is specifically used for:
  • the pretreatment includes at least one of the following treatments:
  • Equation (3) Perform time-domain interference cancellation on the frequency-domain signal to be processed at the received SRS position; where y is the frequency-domain signal to be processed at the SRS position, and the dimension is N ⁇ 1; Is the time-domain cancellation transformation function, It is the frequency domain signal after time domain cancellation, the dimension dimension is N ⁇ 1, and N is the length of channel estimation.
  • the channel information of each terminal includes SRS frequency domain channel estimation results of some subbands
  • the processing module is further specifically configured to:
  • the full bandwidth of the SRS is predicted based on frequency domain extrapolation, and the channel prediction result of each terminal is obtained.
  • processing module is also specifically configured to:
  • the channel prediction is performed on the SRS based on the two-dimensional time-frequency-space mutual calibration to obtain the channel prediction result of each terminal.
  • the sending module is specifically used for:
  • the second RRC signaling is sent to the terminal; the second RRC signaling carries an index identifier, and the index identifier is used to instruct the terminal to use the at least one base sequence identifier The first base sequence identifier for SRS transmission.
  • the sending module is specifically used for:
  • the first RRC signaling is sent to each terminal, and the first RRC signaling carries a base sequence identifier, and the base sequence identifier indicates a candidate collection of all SRS base sequence identifiers.
  • a fourth aspect of the present application provides an apparatus for improving multiple user multiplexing performance, including:
  • the acquiring module is configured to acquire the base sequence identifier configured by the network device through RRC signaling, and the base sequence indicated by the base sequence identifier does not have orthogonality with the base sequence adopted by other terminals;
  • the sending module is used to send the sounding reference signal SRS according to the base sequence identifier.
  • the obtaining module is specifically used for:
  • the fifth aspect of the present application provides a network device, including:
  • the memory is used to store programs and data, and the processor calls the programs stored in the memory to execute the method for improving multi-user multiplexing performance provided by any implementation manner of the first aspect.
  • a sixth aspect of the present application provides a terminal, including:
  • the memory is used to store programs and data, and the processor calls the programs stored in the memory to execute the method for improving the multiplexing performance of multiple users provided by any one of the second aspects.
  • a seventh aspect of the present application provides a computer-readable storage medium, the computer-readable storage medium includes a program, and when the program is executed by a processor, the Performance method.
  • An eighth aspect of the present application provides a computer-readable storage medium.
  • the computer-readable storage medium includes a program. Performance method.
  • the network equipment configures base sequence identifiers for multiple terminals through RRC signaling, and the base sequence identifiers of each terminal indicate that there is no difference between the base sequences.
  • Orthogonal feature the terminal obtains the base sequence identifier configured by the network device through RRC signaling, and sends the sounding reference signal SRS according to the base sequence identifier.
  • the network device performs SRS detection on multiple terminals based on the quasi-orthogonal sequence, and obtains the SRS sent by each terminal According to the channel information of each terminal, channel prediction is performed to obtain the channel prediction result.
  • SRS configuration and detection based on quasi-orthogonal sequences, channel prediction, and calculation of terminal weights can effectively improve user mobility scenarios, the average throughput of the cell, and the average perception rate of users.
  • FIG. 1 is a flowchart of Embodiment 1 of a method for improving multiple user multiplexing performance provided by this application;
  • FIG. 2 is a schematic diagram of the SRS base sequence provided by this application.
  • FIG. 3 is a schematic diagram of reconstruction provided by this application.
  • FIG. 4 is a schematic diagram of the SRS IC process provided by this application.
  • FIG. 5 is a schematic diagram of filtering interference elimination provided by this application.
  • FIG. 6 is a schematic diagram of a multi-user filtering interference cancellation module A provided by this application.
  • FIG. 7 is a schematic diagram of a multi-user filtering interference cancellation module B0 provided by this application.
  • FIG. 8 is a schematic diagram of a multi-user filtering interference cancellation module B1 provided by this application.
  • FIG. 9 is a schematic diagram of SRS time-frequency resources provided by this application.
  • FIG. 10 is an architecture diagram of an SRS channel prediction solution based on two-dimensional time-frequency-space mutual calibration provided by this application;
  • FIG. 11 is a schematic diagram of the position and polarization mode of the base station side antenna provided by this application.
  • Figure 12 is a schematic diagram of the structure of the statistical covariance matrix of TDD LTE field acquisition channel samples provided by this application;
  • FIG. 13 is a schematic diagram of the TBT structure of the statistical covariance matrix provided by this application.
  • FIG. 14 is a schematic diagram of the approximate method of Toeplitz block array compression provided by this application.
  • FIG. 15 is a schematic diagram of the approximate method of Toeplitz matrix compression provided by this application.
  • FIG. 16 is a schematic diagram of the comparison before and after compression of the single-polarization TBT matrix provided by this application.
  • FIG. 17 is a schematic structural diagram of Embodiment 1 of an apparatus for improving multiple user multiplexing performance provided by this application;
  • FIG. 18 is a schematic structural diagram of Embodiment 2 of an apparatus for improving multi-user multiplexing performance provided by this application.
  • the channel information used in calculating the single user (Single User, SU) beamforming (BF) weights comes from the most recent sounding reference signal (Sounding Reference Signal, SRS), and the time intervals for sending SRS information are compared Long, generally on the order of tens of ms, and the user scheduling cycle is generally on the order of ms.
  • SRS Sounding Reference Signal
  • the usage information does not match the current real channel information.
  • the zero-forcing method is used to obtain the precoding weight, which is very sensitive to the accuracy of the channel.
  • the channel information used to calculate the weight comes from the most recent SRS information. When the user moves, This information does not match the current real channel, which is the main reason for the performance degradation of mobile users.
  • this application provides a method that can improve the performance of multi-user multiplexing.
  • Specific improvements include the following: SRS configuration and detection based on quasi-orthogonal sequences, and SRS channel prediction based on frequency domain extrapolation , SRS channel prediction based on time-frequency-space two-dimensional mutual calibration, user movement status and prediction accuracy discrimination, adaptive user weight calculation, including the choice of three weights based on the user status discrimination result: 1. Based on the smallest sub Weight prediction of spatial distance, 2. Weight prediction based on autoregressive theory, 3. After pairing based on statistical covariance matrix, the signal to interference plus noise ratio (SINR) is corrected.
  • SINR signal to interference plus noise ratio
  • FIG. 1 is a flowchart of Embodiment 1 of the method for improving the multiple-user multiplexing performance provided by this application.
  • the method for improving the multiple-user multiplexing performance is mainly applied to network equipment to configure base sequence identifiers for terminals
  • the terminal sends the SRS according to the base sequence identifier, and the network equipment performs SRS detection and channel prediction, which specifically includes the following steps:
  • S101 Configure base sequence identifiers for multiple terminals through RRC signaling, and the base sequences indicated by the base sequence identifier of each terminal do not have orthogonal characteristics.
  • the network side that is, the network equipment, can choose to configure the terminal SRS base sequence identifier through terminal RRC signaling, and the terminal obtains the SRS base sequence identifier by receiving the RRC signaling sent by the base station side.
  • the SRS base sequence identifier is used to generate the SRS base sequence, and at the same time, there is no orthogonality between the base sequences of each terminal in the solution. In the concrete realization of this scheme, at least include the following two situations:
  • the network device sends the first RRC signaling to each terminal, and the first RRC signaling carries at least one base sequence identifier, and then when the terminal needs to change the SRS base sequence, the network device sends to the terminal again
  • the second RRC signaling; the second RRC signaling carries an index identifier, and the index identifier is used to instruct the terminal to use the first base sequence identifier in the at least one base sequence identifier for SRS transmission.
  • the first RRC signaling sent by the network device is received, and when the SRS base sequence needs to be changed, the second RRC signaling sent by the network device is received, and the second RRC signaling carries an index identifier ; Determine according to the index identifier to use the first base sequence identifier in the at least one base sequence identifier for SRS transmission.
  • the base station can pre-configure a set of SRS base sequence identifiers to the terminal through RRC signaling 0, including at least one SRS base sequence identifier, and then every time the terminal needs to change the SRS base sequence
  • the base station configures an index identifier of the terminal through RRC signaling 1, and the identifier is used to instruct the terminal to use one of the multiple SRS base sequence identifiers in the above-mentioned set of SRS base sequence identifiers to send SRS.
  • the network device sends the first RRC signaling to each terminal.
  • the first RRC signaling carries a base sequence identifier, and the base sequence identifier indicates a candidate collection of all SRS base sequence identifiers.
  • the first RRC signaling sent by the network device is received, and the first RRC signaling has a base sequence identifier.
  • the base station can configure an SRS base sequence identifier to the terminal through RRC signaling 0.
  • the range indicated by the sequence identifier includes all candidate sets of SRS base sequence identifiers, and then each When the terminal needs to change the SRS base sequence for the second time, the base station configures another SRS base sequence identifier of the terminal through RRC signaling 0, and the range indicated by the sequence identifier includes all candidate sets of SRS base sequence identifiers.
  • FIG. 2 is a schematic diagram of the SRS base sequence provided by this application.
  • the base sequence noun is only used to distinguish different SRS sequences caused by SRS cyclic shifts of different terminals. If different users generate SRS sequences by configuring cyclic shifts, the sequence conversion to the time domain is windowed and separable, that is, different users occupy different time domain segments. For example, users 0 to 7 use different cyclic shift SRS sequences. The domain is separable. If different users generate SRS sequences by configuring different base sequences, the sequence conversion to the time domain is windowed and inseparable, such as between user 0 and user 8.
  • S102 Send the SRS according to the base sequence identifier.
  • the corresponding terminal obtains the base sequence identifier configured by the network device through RRC signaling, and then obtains the corresponding base sequence according to the above description, and then sends it SRS.
  • S103 Perform SRS detection on multiple terminals based on the quasi-orthogonal sequence, and acquire channel information for each terminal to send SRS.
  • the network device can perform SRS detection.
  • the SRS of multiple terminals is detected based on a quasi-orthogonal sequence to obtain Send SRS channel information to each terminal.
  • the traditional channel estimation algorithm base station uses the received reference signal sequence to remove the reference signal sequence, and then performs time-domain windowing of different users, thereby separating multiple multiplexed users of different SRS cyclic shift sequences.
  • the network equipment configures multiple base sequences for multiple multiplexed terminals, the traditional channel estimation method cannot separate terminals with different base sequences (also called users), resulting in sequence interference between terminals with different base sequences, resulting in The channel estimation performance is degraded.
  • the following PDP stands for the abbreviation of power delay distribution.
  • the SRS detection method based on the quasi-orthogonal sequence provided in this application can preprocess the frequency domain information to be processed at the received SRS position, and then send the SRS channel to each terminal according to the preprocessed frequency domain signal.
  • the pre-processing process includes at least one of the following operations, and may also include a combination of multiple operations:
  • the network equipment performs frequency domain filtering on the frequency domain signal to be processed at the received SRS position:
  • y(n) is the frequency domain signal at the SRS position
  • w(n) is the frequency domain window coefficient
  • Is the frequency domain filtered signal
  • n 0,...
  • N is the channel estimation position index
  • N is the length of the channel estimation.
  • w(n) can be pre-stored and pre-calculated for storage or obtained through online calculation of some parameters.
  • the same w(n) can be applied to multiple antennas and multiple beams.
  • the network equipment performs frequency domain filtering on the frequency domain signal to be processed at the received SRS position:
  • y is the frequency domain signal at the SRS position, and the dimension is N ⁇ 1.
  • w is the frequency domain window coefficient, and the dimension is N ⁇ N. It is the signal after frequency domain filtering, the dimension is N ⁇ 1, and N is the length of channel estimation.
  • w can be pre-calculated and stored or obtained through online calculation of some parameters, including but not limited to: SRS sequence, signal-to-noise ratio information, time-domain window correlation coefficient information (including the number of sub-paths, the quantization delay of sub-paths , The amplitude and phase of the sub-path, etc.).
  • the same w(n) can be applied to multiple antennas and multiple beams.
  • the network equipment performs time-domain interference cancellation on the frequency-domain signal at the received SRS position:
  • y is the frequency domain signal to be processed at the SRS position, and the dimension is N ⁇ 1.
  • Is the time-domain cancellation transformation function (can also be expressed as It is the frequency domain signal after time domain cancellation, the dimension dimension is N ⁇ 1, and N is the length of channel estimation.
  • the function of y is to eliminate the mutual interference between different base sequence channels in the time domain, to obtain each independent base sequence channel that minimizes the interference between the base sequences, and then perform time domain windowing and noise reduction on each base sequence. Field to obtain channel values corresponding to different SRS sequences.
  • a function can be equivalent to perform at least one of the following operations:
  • SRSs with different base sequences and/or different sequences that is, SRSs with the same base sequence are grouped into one group, which can be divided into several groups of SRS sequences.
  • the corresponding criteria include but not limited to the average signal to interference and noise ratio of different groups, the number of SRS sequences, the average received power of SRS sequences, etc. Wait.
  • the above operation process can be executed K times, and K may optionally be equal to the number of groups of SRS, or the number of times that the preset condition has been executed.
  • L p represents the time domain corresponding point corresponding to the value added back to L p.
  • I the time domain channel coefficient of the jth user in the p group
  • W p,j is the time domain window coefficient of the jth user in the p group. Then, the time domain window coefficients of the jth user in the p group are transformed from the time domain to the frequency domain to obtain the frequency domain channel coefficients of the jth user in the p group.
  • N base is the number of base sequences, and/or the discrete Fourier transform (Inverse Discrete Fourier Transform, IDFT) or Inverse Fast Fourier Transform (IFFT) operation or equivalent after zero padding to N FFT points Operation, N FFT is the discrete Fourier transform (Discrete Fourier Transform, DFT) or IFFT operation points, the result is recorded as
  • y PDPp is y PDPp with zero padding or adding virtual subcarriers to extrapolate to N FFT points and IDFT to obtain the PDP result.
  • the method of adding virtual subcarriers to extrapolation includes using all and part of the existing subcarrier signals to obtain the signal values outside the subcarriers by multiplying them by prestored or online calculated weights.
  • the weight calculation can be obtained by using the minimum mean square error criterion and/or signal-to-noise ratio parameters and/or SRS sequence related parameters. % Is the modulo symbol.
  • Reconstructing more accurate sub-carrier user interference cancellation refers to reconstructing a more complete PDP with partial large-path time domain points, thereby improving the quality of interference cancellation.
  • the reconstruction operation can be expressed as: Or an equivalent operation, where 1 T represents a full 1-column vector of T ⁇ 1 dimensions.
  • Hypothesis Represents the Y large-path time-domain points selected by the p-th cell (the dimension of Y is the number of pre-defined or the maximum number of pre-defined time-domain points selected, or the number of radial-time-domain points that meet certain conditions), then reconstructed
  • the first step is to solve the equation (Similar to deconvolution or equivalent operation) or equivalent operation, where A is based on the number and position of major diameters Select the matrix formed by the corresponding elements in, where the dimension of A is Y ⁇ Y, x represents the time domain path fitted with these major paths, and the dimension of x is Y ⁇ 1.
  • FIG. 3 is a schematic diagram of the reconstruction provided by this application. As shown in Figure 3, the figure only contains one true path without interference or noise.
  • the dots in Figure 3 represent a group of major diameters selected by the interference cancellation scheme without reconstruction (i.e., the first group of major diameters), and the crosses represent the three major diameters for reconstruction selected by a higher threshold (that is, the second group). Large diameter), and the thinner line is the reconstructed PDP (Power Delay Profile) obtained after inversion and circular convolution (namely, reconstruction path).
  • the reconstructed PDP Power Delay Profile
  • FIG 4 is a schematic diagram of the SRS Interference Cancellation (IC) process provided by this application.
  • the abscissa from left to right is the frequency domain received signal of the three base sequences after passing the Least Squares (Least Square, LS) operation, and then transformed to the time domain PDP map marked as PDP 0 , PDP 1 , PDP 2 , and then calculate the normalized result of the signal power in the PDP map to the average power of interference noise outside the signal time domain window, recorded as the time domain Click to select the value PDP 0,NI .
  • Least Square Least Square
  • the first step for the first base sequence, select the time-domain point value PDP 0, and several time-domain point values T 0 in NI that are greater than the preset threshold, and then use the correlation between the first base sequence and other base sequences, Eliminate the interference of the time domain point value T 0 of the first base sequence in the other two PDP spectra. Update the time domain PDP map of the second base sequence and the third base sequence.
  • the second step for the second base sequence, select the time domain point value PDP 1, and several time domain point values T 1 in NI that are greater than the preset threshold, and then use the correlation between the first base sequence and other base sequences, Eliminate the interference of the time domain point value T 1 of the first base sequence in the other two PDP spectra.
  • an iterative mechanism can be introduced.
  • the three base sequence time-domain PDPs output from the previous paragraph are used as the input of the next iteration.
  • the number of iterations can be preset or meet a certain A condition stops the iteration.
  • FIG. 5 is a schematic diagram of filtering interference elimination provided by this application, as shown in FIG.
  • SRS received frequency domain received signal y
  • the dimension of y is S ⁇ 1
  • ⁇ is the frequency domain filter coefficient
  • the dimension of ⁇ is S ⁇ 1
  • S is the number of sub-carriers
  • frequency domain filtering is performed on y:
  • Fig. 4 Take the operation of Fig. 4 as module A, after the major path time domain point of each base sequence in module A is selected, time domain path reconstruction is performed before interference is eliminated.
  • the reconstruction operation can be expressed as: Or an equivalent operation, where 1 T represents a full 1-column vector of T ⁇ 1 dimensions.
  • Hypothesis Represents the Y large-path time-domain points selected by the p-th cell (the dimension of Y is the number of pre-defined or the maximum number of pre-defined time-domain points selected, or the number of radial-time-domain points that meet certain conditions), then reconstructed
  • the first step is to solve the equation (Similar to deconvolution or equivalent operation) or equivalent operation, where A is based on the number and position of major diameters Select the matrix formed by the corresponding elements in, where the dimension of A is Y ⁇ Y, x represents the time domain path fitted with these major paths, and the dimension of x is Y ⁇ 1.
  • FIG. 3 is a schematic diagram of the above reconstruction process.
  • the figure only contains a true path without interference or noise.
  • the dots in the figure represent a group of major diameters (the first group of major diameters) selected by the interference cancellation scheme without reconstruction, and the crosses represent the 3 major diameters for reconstruction (the second group of major diameters) filtered by a higher threshold.
  • the thin line is the reconstructed PDP (reconstruction path) obtained after inversion and circular convolution. After the reconstructed PDP is obtained, other weak paths that do not overlap with the first group of major diameter positions can be eliminated together (the overlap position is still based on the major diameter selected for the first time).
  • the frequency domain filtering step can be performed before module A, as shown in formula (8).
  • the time domain path reconfiguration is performed before interference is eliminated.
  • the reconstruction operation can be expressed as: Or an equivalent operation, where 1 T represents a full 1-column vector of T ⁇ 1 dimensions.
  • Hypothesis Represents the Y large-path time-domain points selected by the p-th cell (the dimension of Y is the number of pre-defined or the maximum number of pre-defined time-domain points selected, or the number of radial-time-domain points that meet certain conditions), then reconstructed
  • the first step is to solve the equation (Similar to deconvolution or equivalent operation) or equivalent operation, where A is based on the number and position of major diameters Select the matrix formed by the corresponding elements in, where the dimension of A is Y ⁇ Y, x represents the time domain path fitted with these major paths, and the dimension of x is Y ⁇ 1.
  • FIG. 3 is a schematic diagram of the above reconstruction process.
  • the figure only contains a true path without interference or noise.
  • the dots in the figure represent a group of major diameters (the first group of major diameters) selected by the interference cancellation scheme without reconstruction, and the crosses represent the 3 major diameters for reconstruction (the second group of major diameters) filtered by a higher threshold.
  • the thin line is the reconstructed PDP (reconstruction path) obtained after inversion and circular convolution. After the reconstructed PDP is obtained, other weak paths that do not overlap with the first group of major diameter positions can be eliminated together (the overlap position is still based on the major diameter selected for the first time).
  • Fig. 6 is a schematic diagram of the multi-user filtering interference cancellation module A provided by this application.
  • the input is x0.
  • the input x0 is subjected to matrix frequency domain filtering, where the filter matrix A -1 dimension is TT, after filtering
  • the output is y0, y0.
  • FIG. 7 is a schematic diagram of the multi-user filtering interference cancellation module B0 provided by this application.
  • the module B0 integrates the operation of the module A, and uses the output of the module A For further signal processing, as shown in Figure 7, update Then right Perform noise reduction processing (for example, transform to the time domain for windowing and noise reduction or time-domain point selection noise reduction algorithm), output Then the n+1th base sequence uses the first n updated Update Output sequence
  • the frequency domain data output of each user is different cyclically shifting the same base sequence after the user separation process of the time domain window.
  • FIG. 8 is a schematic diagram of the multi-user filtering interference cancellation module B1 provided by this application.
  • the module B1 integrates the operation of the module A, and uses the output of the module A For further signal processing, update Then right Perform noise reduction processing and time-domain window user separation process of the same base sequence with different cyclic shifts. Frequency domain data output for each user.
  • Performing SRS detection processing according to any of the above solutions can obtain channel information of each terminal.
  • S104 Perform channel prediction according to the channel information of each terminal to obtain a channel prediction result.
  • the network equipment After the network equipment obtains the channel information of each terminal, it can perform channel prediction.
  • This application provides the following prediction methods:
  • the first prediction method is SRS channel prediction based on frequency domain extrapolation.
  • the channel information of each terminal includes the SRS frequency domain channel estimation results of some subbands, and the network equipment performs the frequency domain extrapolation based on the frequency domain channel estimation results of the SRS frequency domain channel estimation results of some subbands of each terminal.
  • the bandwidth is predicted, and the channel prediction result of each terminal is obtained.
  • FIG. 9 is a schematic diagram of the SRS time-frequency resources provided by this application.
  • SRS time-frequency resources include time domain
  • the horizontal is time and the vertical is frequency domain.
  • the SRS hops over the full SRS bandwidth four times, and each user uses 1/4 SRS full bandwidth to carry the SRS signal each time.
  • the SRS frequency domain channel estimation results of some subbands can be used to extrapolate to the full bandwidth, and then the full bandwidth channel estimation results can be obtained.
  • each user only needs to use 1/4 of the SRS bandwidth to achieve channel information feedback.
  • the SRS capacity is 4 times the baseline. Under the condition that the total SRS remains unchanged, each user The feedback cycle is reduced to 1/4 of the original, and the expansion of the frequency domain is realized to achieve the purpose of improving the performance of the mobile scene.
  • the time domain sparsity is mainly used here, and it can also be expressed as the space formed by a small number of time domain sparse path bases for the frequency domain signal.
  • Using the sparsity time domain signal can be expressed as:
  • the full bandwidth frequency domain signal can be obtained, where N paths is the number of time domain sparse paths, a p is the amplitude and phase parameters of the time domain path, and W p (t- ⁇ p ) is the phase and/or amplitude change parameters of Doppler frequency shift, Doppler spread and multipath delay with time.
  • N paths is the number of time domain sparse paths
  • a p is the amplitude and phase parameters of the time domain path
  • W p (t- ⁇ p ) is the phase and/or amplitude change parameters of Doppler frequency shift, Doppler spread and multipath delay with time.
  • the dimension of W is N ⁇ S column
  • the dimension of H exist is S ⁇ 1
  • the dimension of H extended is N ⁇ 1.
  • N is the number of subcarriers of the full bandwidth
  • S is the number of subcarriers of the partial bandwidth
  • f 0,0 is the mapping function, which maps the frequency domain channels of each subcarrier of the partial bandwidth to each subcarrier of the full bandwidth.
  • the time domain related parameters can be searched in the following ways: (11)
  • Transform the frequency domain signal to the sparse basis domain through the sparse basis transform Through the 1 norm or 0 norm sparse method, iteratively select the basis that meets certain constraints one by one (group) as the candidate basis. When certain conditions are met, Stop the iteration, transform the candidate base set through the sparse domain to the frequency domain signal, and restore the full bandwidth frequency domain channel.
  • the second prediction method is based on SRS channel prediction based on time-frequency-space two-dimensional mutual calibration.
  • the network equipment performs channel prediction on the SRS based on the time-frequency-space two-dimensional mutual calibration based on the channel information of each terminal, and obtains the channel prediction result of each terminal.
  • Fig. 10 is an architecture diagram of the SRS channel prediction scheme based on time-frequency-space two-dimensional mutual calibration provided by this application, as shown in Fig. 10: the dotted line represents optional parts, and the specific operation includes at least one of the following steps:
  • the channel estimation value at time 0 is estimated through the time domain channel prediction algorithm scheme, where time 1 is the current time, time N-1 ⁇ 2 is the historical time, and time 0
  • the corresponding time-domain channel prediction algorithm solution may use Kalman filter algorithms or autoregressive algorithms.
  • the channel feature vectors or weights at time 0 are estimated through the time domain vector prediction algorithm scheme, where time 1 is the current time and N-1 ⁇ 2 Time is a historical time, and time 0 is a future time.
  • the corresponding time-domain channel prediction algorithm scheme can use Kalman filter algorithms or autoregressive algorithms.
  • time domain prediction intermediate parameters and frequency domain extrapolation parameters as well as the channel value or vector value at SRS time 1 and/or SRS time 0 to interpolate the channels or vectors of several downlink subframes between two SRS time 1 and 0 .
  • the channel of the terminal can be predicted, and the corresponding prediction result can be obtained.
  • a network device configures a base sequence identifier for multiple terminals through RRC signaling, and the base sequence indicated by the base sequence identifier of each terminal does not have orthogonal characteristics, and the terminal obtains
  • the network device transmits the sounding reference signal SRS based on the base sequence identifier configured by RRC signaling, and the network device performs SRS detection on the multiple terminals based on the quasi-orthogonal sequence, and obtains the channel through which each terminal sends the SRS Information, channel prediction is performed according to the channel information of each terminal to obtain the channel prediction result.
  • SRS configuration and detection based on quasi-orthogonal sequences, and channel prediction can effectively improve user mobility scenarios, the average throughput of the cell, and the average perception rate of users.
  • the network device may also obtain the motion state of the terminal according to the channel of each terminal at different SRS moments, and the motion state is used to indicate the terminal’s
  • the magnitude of the moving speed; the moving state includes: a quasi-stationary state, a low-speed moving state, or a medium-high-speed moving state; and then the weight of the terminal is calculated and obtained according to the channel prediction result and the moving state of each terminal.
  • the channel prediction result of the terminal predicted by the network device includes the SRS channel feature vector or weight.
  • the specific implementation is as follows.
  • the network device judges the user's movement state according to the correlation between the feature vectors or the weights calculated by the channels at different SRS moments.
  • the correlation calculation result can be further processed by means of time domain filtering or frequency domain filtering.
  • the weight correlation is calculated according to the following formula:
  • S is the number of sub-carriers the SRS frequency domain filtering (1) in the [rho] t after frequency domain filtering correlation values, ⁇ (2) of the correlation values t of the time domain filtering, a time domain [alpha] Filter coefficient.
  • the mobility state of the user is judged, and different weight calculation methods are selected according to the different mobility states of the user.
  • the weight calculation method 0 is used to calculate the weight
  • the weight calculation method 1 is used to calculate the weight
  • the weight is used to calculate the weight
  • Calculation method 2 calculates the weight.
  • Calculation method 2 calculates the weight. Among them, ⁇ 0 ⁇ 1 .
  • the terminal can be considered to be in a quasi-static state.
  • the weight optimization idea is to assume that the terminal weight is stable, and the terminal weight can be filtered in type-space and/or time domain. That is, the motion state of the terminal is a quasi-static state, and the SRS channel prediction vector or weight of the terminal is filtered in the type-space domain and/or time domain.
  • the least mean square filtering of the weights is implemented based on the subspace distance and gradient descent method.
  • w(t) be the predicted weight value at the time of SRS to be predicted
  • v(t) be the real weight value at the time of predicted SRS.
  • w(t) is the filtered weight at time t
  • w(t-1) is the filtered weight at time t-1
  • ⁇ is the gradient descent coefficient
  • v(t-1) is the loss weight at time t-1 value.
  • the terminal can be considered to be moving at a low speed.
  • the weight optimization idea is to assume that the weights of the terminal are related at different times, and the weights of the terminal can be used for type two airspace and/ Or time domain forecasting. That is, the motion state of the terminal is a low-speed motion state, and the SRS channel prediction vector or weight value of the terminal is filtered in the type 2 spatial domain and/or time domain.
  • a prediction algorithm based on Kalman filtering including its enhanced algorithm.
  • NLMS normalized least mean square adaptive filtering
  • RLS recursive least squares algorithm
  • the prediction algorithm based on autoregressive (AR) filtering includes its enhanced algorithm.
  • the prediction vector can be obtained by the following formula,
  • ⁇ can be obtained from the left pseudo-inverse matrix of X n ,
  • the dimension of X n is N t ⁇ N
  • the dimension of ⁇ is N ⁇ 1
  • the dimension of y n is N ⁇ 1.
  • H 1 [v(t-1),v(t-2)v H (t-2)v(t-1),...,v(tN)v H (tN)v(t-1)] ( 11)
  • the terminal can be considered to be moving at medium and high speeds.
  • the weight optimization idea is to assume that the correlation of the weights of the terminal at different times is low, and the second step distance calculation can be performed on the channel of the terminal to generate statistics The weight of the feature. That is, if the motion state of the terminal is a medium-to-high speed motion state, the second step distance calculation is performed on the channel of the terminal to generate the weight of the terminal.
  • the statistical weight calculation is performed in the time domain, and the SRS time-frequency resource includes the time domain and the frequency domain. Therefore, correspondingly, the statistical scheme here can be frequency-domain statistical weight, time-domain statistical weight, or two-dimensional time-frequency statistical weight.
  • N Sc is the frequency domain statistical SRS subcarrier
  • h(k) is the channel matrix or vector of the k-th subcarrier
  • R f represents the frequency domain statistical autocorrelation matrix
  • time domain alpha filtering For historical SRS moments (only consider this band), first averaging in the frequency domain, and then alpha filtering in the time domain.
  • the period of time domain alpha filtering is S times the configured SRS period, where ⁇ is the time domain filter coefficient, and R t is t Time correlation matrix.
  • Time-frequency domain statistical power is a combination of frequency-domain statistical power and time-domain statistical power. For historical SRS moments (considering the full bandwidth), first average in the frequency domain, and then filter in the time domain with alpha. The period of the time domain alpha filter is the configured SRS period. When ⁇ is 1, it degenerates into frequency domain statistical weight.
  • the calculation methods for the terminal in multiple users specifically include the following:
  • the baseline uses REZF for MU weight calculation
  • H is a column full-rank matrix composed of multiple user SU weights
  • ⁇ 2 is a regularization coefficient.
  • the statistical weight-based REZF algorithm is to replace the baseline SU weights with the weights (the eigenvectors corresponding to the largest eigenvalues) obtained by the statistical correlation matrix doing Singular Value Decomposition (SVD).
  • the SLNR algorithm criterion is to maximize the signal power relative to the signal power and noise leaked to other cells, so that the weight can be compatible with the leakage problem to the neighboring cell and the signal-to-noise ratio of the target user.
  • the weight generation constraint based on the SLNR criterion As follows,
  • H k is the frequency domain channel of the kth user
  • W k is the corresponding weight
  • K is the number of users (the number of streams), when H l is N T ⁇ 1, we can get:
  • V k is the SU weight of the k-th user
  • the above formula is the SLNR_W algorithm derived from the instantaneous channel coefficients, directly substituting the statistical correlation matrix and the principal eigenvectors obtained from the SVD decomposition of the statistical correlation matrix into the above formula to obtain
  • V k can be the main feature vector corresponding to the statistical covariance matrix R k , or the main feature vector of the instantaneous covariance matrix.
  • Using statistical SU weights can better filter and reduce the impact of mobility, while using instantaneous SU weights can overcome factors such as frequency selection, and can better match the actual channel in low-to-medium-speed scenarios. Which SU weight is used in different scenarios needs to be evaluated.
  • the SLNR algorithm can better avoid the interference caused by leakage to other users in principle, so the algorithm is expected to perform better.
  • the SLNR algorithm needs to solve the SVD decomposition or inversion operation of a large matrix (2*m*n) ⁇ (2*m*n), which is relatively complicated and difficult to implement in products. Therefore, the algorithm is simplified next.
  • R v1 , R v2 , R H1 , R H2 are n ⁇ n, n ⁇ n, m ⁇ m, m ⁇ m matrices, R v1 , R H1 correspond to a kind of polarization, R v2 , R H2 correspond to Another polarization.
  • R v1 is calculated as follows: first obtain the correlation matrix of a certain column according to the channel coefficients corresponding to n rows of the vertical dimension of a kind of polarization, and then average the correlation matrix of the columns corresponding to m columns.
  • R H1 is calculated as follows: first obtain the correlation matrix of a certain row according to the channel coefficients corresponding to m columns of the horizontal dimension of a polarization, and then average the row correlation matrix corresponding to n rows.
  • R v2 and R H2 have similar methods. If the statistical correlation matrix is written in such a block structure, it can be solved in blocks, thereby avoiding the SVD and inversion of the large matrix.
  • the current base station antenna structure is HmVnPt, similar to a statistical correlation matrix, it can be assumed that the SU weight of the channel (single stream per user) meets the following structure:
  • V v1 , V v2 , V H1 , and V H2 are matrices of n ⁇ 1, n ⁇ 1, m ⁇ 1, m ⁇ 1, respectively, V v1 , V H1 correspond to a kind of polarization, V v2 , V H2 correspond to Another polarization. V v1 , V v2 , V H1 , and V H2 are the left eigenvectors corresponding to the maximum eigenvalues of R v1 , R v2 , R H1 , and R H2 respectively.
  • the weights corresponding to the two polarizations can be averaged. You can average and then SVD or SVD and then average. The average and then SVD can reduce the SVD operation by half. The result of SVD first and then average is,
  • V v is the main feature vector corresponding to (R v1 +R v2 )/2
  • V H is the main feature vector corresponding to (R H1 +R H2 )/2.
  • the SLNR algorithm based on structured statistical weights can be derived.
  • R lvi statistics and statistical theory here satisfies R lHi Toeplitz structure, it is possible by the average diagonal manner and R lHi R lvi Toeplitz matrix, thereby achieving further compression matrix for easy storage.
  • TBT Toeplitz-Block-Toeplitz statistical weight mainly uses part of the properties of the statistical covariance matrix to simplify the complexity of a single calculation.
  • FIG 12 is a schematic diagram of the statistical covariance matrix structure of the TDD LTE field acquisition channel sample provided by this application, as shown in Figure 12.
  • the covariance matrix within the polarization Suppose that if the relative physical distance between any two co-polarized elements on the URA antenna array is equal, then the statistical correlation between the elements is also equal. Therefore, under this assumption, the single polarization internal
  • the channel statistical covariance matrix can be approximated by a Toeplitz-Block-Toeplitz matrix structure. This matrix has two characteristics:
  • Each block of the statistical covariance matrix (corresponding to the vertical ULA statistical covariance matrix), which is a Toeplitz matrix;
  • Each block of the statistical covariance matrix (corresponding to the level ULA statistical covariance matrix), the matrix is the Toeplitz matrix;
  • Figure 13 is a schematic diagram of the TBT structure of the statistical covariance matrix provided by this application. Based on the above two properties, the statistical covariance matrix of the end user can be approximately expressed as the structure shown in Figure 13.
  • the statistical covariance matrix between polarizations is zeroed out, and the planned statistical covariance matrix satisfies the Toeplitz structure.
  • the (m*n) ⁇ (m*n) matrix can be regarded as a vertical ULA statistical covariance matrix composed of n *n block matrix, each block matrix is an m ⁇ m Toeplitz matrix, corresponding to the horizontal ULA statistical covariance matrix.
  • FIG. 14 is a schematic diagram of the compression approximation method of the Toeplitz block matrix provided by this application. As shown in Fig. 14, for a polarized (m*n) ⁇ (m*n) matrix, first, each m ⁇ m matrix that is approximately equal Blocks are added and averaged, and because the matrix is conjugate and symmetric, only blocks above the diagonal need to be averaged, as shown in Figure 14 with blocks of the same color and letters.
  • the second step is to perform Toeplitz average within the block.
  • Figure 15 is a schematic diagram of the Toeplitz matrix compression approximation method provided by this application. For each block, we need to add the elements that should be equal on each diagonal line to average. As shown in Figure 15, each block passes through. Since each Toeplitz block is not necessarily a conjugate symmetric block (except the block on the diagonal), it is necessary to calculate the average of all X diagonals, and finally obtain a vector of X elements.
  • the third step is to merge into a compressed matrix.
  • Figure 16 is a schematic diagram of the comparison before and after compression of the single-polarization TBT matrix provided by this application.
  • each of the n m ⁇ m matrices is compressed into X element vectors to form an n ⁇ X compressed matrix, as shown in Figure 16.
  • the TBT approximation can greatly reduce the storage cost, and at the same time, it can simplify the matrix inversion by using the properties of the Toeplitz matrix.
  • the statistical covariance matrix in the form of TBT is:
  • the goal is to solve
  • network equipment based on the SRS configuration and detection of quasi-orthogonal sequences, SRS channel prediction based on frequency domain extrapolation, SRS channel prediction based on two-dimensional time-frequency-space mutual calibration, user movement status and prediction accuracy identification, adaptive user
  • the weight calculation can effectively improve the user's mobile scene, the average throughput of the cell, and the average perception rate of the user.
  • FIG. 17 is a schematic structural diagram of Embodiment 1 of an apparatus for improving multi-user multiplexing performance provided by this application.
  • the apparatus 10 for improving multi-user multiplexing performance includes:
  • the sending module 11 is configured to configure base sequence identifiers for multiple terminals through RRC signaling, and the base sequences indicated by the base sequence identifier of each terminal do not have orthogonal characteristics;
  • the processing module 12 is configured to perform SRS detection on the multiple terminals based on the quasi-orthogonal sequence, and obtain the channel information for each terminal to send SRS;
  • the processing module 12 is also used to perform channel prediction according to the channel information of each terminal to obtain a channel prediction result.
  • the apparatus for improving multi-user multiplexing performance provided in this embodiment is used to implement the technical solutions on the network device side in any of the foregoing method embodiments.
  • the implementation principles and technical effects are similar, and will not be repeated here.
  • the processing module 12 is further configured to:
  • the movement state of the terminal is acquired, and the movement state is used to indicate the movement speed of the terminal;
  • the movement state includes: a quasi-static state, a low-speed movement state, or a medium-high speed state Mobile state
  • the weight of the terminal is calculated and acquired.
  • the channel prediction result of the terminal includes the SRS channel feature vector or weight
  • the processing module 12 is specifically configured to:
  • the motion state of the terminal is a quasi-static state, perform type-spatial and/or time-domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the motion state of the terminal is a low-speed motion state, perform Type 2 spatial and/or time domain filtering on the SRS channel prediction vector or weight of the terminal;
  • the second step distance calculation is performed on the channel of the terminal to generate the weight of the terminal.
  • processing module 12 is specifically configured to:
  • the minimum mean square filtering is performed on the SRS prediction weight of the terminal based on the subspace distance and gradient descent method.
  • processing module 12 is specifically configured to:
  • the SRS channel prediction vector or weight value of the terminal is filtered based on the recursive least squares RLS prediction algorithm.
  • processing module 12 is specifically configured to:
  • the pretreatment includes at least one of the following treatments:
  • Equation (3) Perform time-domain interference cancellation on the frequency-domain signal to be processed at the received SRS position; where y is the frequency-domain signal to be processed at the reference signal position, and the dimension is N ⁇ 1; Is the time-domain cancellation transformation function, It is the frequency domain signal after time domain cancellation, the dimension dimension is N ⁇ 1, and N is the length of channel estimation.
  • the channel information of each terminal includes SRS frequency domain channel estimation results of some subbands, and the processing module 12 is further specifically configured to:
  • the full bandwidth of the SRS is predicted based on frequency domain extrapolation, and the channel prediction result of each terminal is obtained.
  • processing module 12 is also specifically configured to:
  • the channel prediction is performed on the SRS based on the two-dimensional time-frequency-space mutual calibration to obtain the channel prediction result of each terminal.
  • the sending module 11 is specifically configured to:
  • the second RRC signaling is sent to the terminal; the second RRC signaling carries an index identifier, and the index identifier is used to instruct the terminal to use the at least one base sequence identifier The first base sequence identifier for SRS transmission.
  • the sending module 11 is specifically configured to:
  • the first RRC signaling is sent to each terminal, and the first RRC signaling carries a base sequence identifier, and the base sequence identifier indicates a candidate collection of all SRS base sequence identifiers.
  • the apparatus for improving the multi-user multiplexing performance provided by any of the foregoing embodiments is used to implement the technical solutions on the network device side in the foregoing method embodiments.
  • the implementation principles and technical effects are similar, and details are not described herein again.
  • FIG. 18 is a schematic structural diagram of Embodiment 2 of an apparatus for improving multi-user multiplexing performance provided by this application.
  • the apparatus 20 for improving multi-user multiplexing performance includes:
  • the acquiring module 21 is configured to acquire a base sequence identifier configured by the network device through RRC signaling, and the base sequence indicated by the base sequence identifier does not have orthogonality with the base sequence used by other terminals;
  • the sending module 22 is configured to send a sounding reference signal SRS according to the base sequence identifier.
  • the obtaining module 21 is specifically configured to:
  • the apparatus for improving the multi-user multiplexing performance provided by any of the foregoing embodiments is used to implement the technical solutions on the terminal side in the foregoing method embodiments.
  • the implementation principles and technical effects are similar, and details are not described herein again.
  • This application also provides a network device, including: a processor, a memory, a receiver, and a transmitter; the memory is used to store programs and data, and the processor calls the programs stored in the memory to perform the enhancement provided by any of the foregoing embodiments.
  • the present application also provides a terminal, including: a processor, a memory, a receiver, and a transmitter; the memory is used to store programs and data, and the processor calls the programs stored in the memory to execute the terminal side in any of the foregoing embodiments.
  • the memory and the processor are directly or indirectly electrically connected to achieve data transmission or interaction.
  • these elements may be electrically connected to each other through one or more communication buses or signal lines, for example, they may be connected through a bus.
  • the memory stores computer execution instructions for implementing the data access control method, including at least one software function module that can be stored in the memory in the form of software or firmware.
  • the processor executes various software programs and modules by running the software programs and modules stored in the memory. Functional application and data processing.
  • the memory can be, but not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM) ), Erasable Programmable Read-Only Memory (EPROM), Electrical Erasable Programmable Read-Only Memory (EEPROM) and so on.
  • RAM Random Access Memory
  • ROM Read Only Memory
  • PROM Programmable Read-Only Memory
  • EPROM Erasable Programmable Read-Only Memory
  • EEPROM Electrical Erasable Programmable Read-Only Memory
  • the memory is used to store the program, and the processor executes the program after receiving the execution instruction.
  • the software programs and modules in the memory may also include an operating system, which may include various software components and/or drivers for managing system tasks (such as memory management, storage device control, power management, etc.), and Communicate with various hardware or software components to provide an operating environment for other software components.
  • the processor can be an integrated circuit chip with signal processing capabilities.
  • the foregoing processor may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), and so on.
  • the methods, steps, and logical block diagrams disclosed in the embodiments of the present application can be implemented or executed.
  • the general-purpose processor may be a microprocessor or the processor may also be any conventional processor or the like.
  • the present application also provides a computer-readable storage medium, the computer-readable storage medium includes a program, and when the program is executed by a processor, the program is used to perform the enhanced multi-user multiplexing on the network device side in any of the foregoing method embodiments Technical solutions for performance methods.
  • the present application also provides a computer-readable storage medium.
  • the computer-readable storage medium includes a program.
  • the program is executed by a processor, the program is used to execute any method embodiment in the method for improving the multi-user multiplexing performance. Side technical solutions.
  • the program product includes a computer program stored in a readable storage medium. At least one processor of the network device can read the computer program from the readable storage medium, and at least one processor executes the computer program.
  • the computer program enables the network device to implement the technical solution in any of the foregoing method embodiments.
  • the program product includes a computer program stored in a readable storage medium. At least one processor of the terminal can read the computer program from the readable storage medium, and at least one processor executes the computer program.
  • the computer program enables the terminal to implement the technical solution in any of the foregoing method embodiments.
  • the present application also provides a chip that can be applied to network equipment.
  • the chip includes: at least one communication interface, at least one processor, and at least one memory.
  • the communication interface, memory, and processor are interconnected by a bus.
  • the processor invokes the computer program stored in the memory to execute the technical solution on the network device side in any of the foregoing method embodiments.
  • the present application also provides a chip that can be applied to a terminal.
  • the chip includes: at least one communication interface, at least one processor, and at least one memory.
  • the communication interface, memory, and processor are interconnected by a bus.
  • the processor invokes the computer program stored in the memory to execute the technical solution on the terminal side in any of the foregoing method embodiments.

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Power Engineering (AREA)
  • Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Quality & Reliability (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请提供一种提升多用户复用性能的方法、装置、设备和存储介质，该方法中，网络设备通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性，终端获取网络设备通过RRC信令配置的基序列标识，根据所述基序列标识，发送探测参考信号SRS，网络设备基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息，根据每个终端的信道信息进行信道预测，得到信道预测结果。基于准正交序列的SRS配置和检测、进行信道预测，计算终端的权值，可以有效提用户移动场景，小区的平均吞吐量以及用户的平均感知速率。

Description

提升多用户复用性能的方法、装置、设备和存储介质
本申请要求于2019年04月29日提交中国专利局、申请号为201910355875.6、申请名称为“提升多用户复用性能的方法、装置、设备和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信技术，尤其涉及一种提升多用户复用性能的方法、装置、设备和存储介质。
背景技术
在多天线时分双工复用(Time Division Duplexing，TDD)Massive多输入多输出(Multiple Input Multiple Output，MIMO)的实际商用中，由于用户移动导致的性能恶化是多用户(Multiple User，MU)MIMO中一个极具挑战的难题。这种恶化首先是移动用户本身的性能恶化，然后是整个网络的性能恶化。
在当前使用的MU MIMO算法中，在进行多用户配对时，首先估计单用户权值，然后使用迫零，或者使用以迫零为基础的方法进行正交化处理，以获得多用户权值，即预编码权值。上述方法对信道的准确度非常敏感，首先，单用户权值计算基于最近一次的探测参考信号(Sounding Reference Signal，SRS)信息以获得信道，而SRS信息发送的时间间隔都比较长，一般为几十ms的量级，而用户调度的周期一般是ms量级。当用户移动时，信道变化比较快，这就导致了用户使用的信道信息与当前的真实信道不匹配，从而导致了移动用户性能下降，这种性能下降首先表现为这种不匹配导致的目标用户的信号功率下降。然后，经过迫零步骤后，这种错误会进一步扩散，除去目标用户的信号功率下降外，还带来了其他用户对目标用户的干扰增加，在相同SRS周期下，移动速度越快，不准确的程度越高，在相同终端移动速度，SRS周期越大，不准确的程度越高，从而进一步恶化了移动用户的性能。
发明内容
本申请实施例提供一种提升多用户复用性能的方法、装置、设备和存储介质，用于解决现在有的MU MIMO算法导致移动用户性能下降的问题。
本申请第一方面提供一种提升多用户复用性能的方法，应用于网络设备，所述方法包括：
通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性；
基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息；
根据每个终端的信道信息进行信道预测，得到信道预测结果。
在一种具体实现方式中，所述方法还包括：
根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；
根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。
进一步的，在一种具体实现方式中，终端的信道预测结果包括SRS信道特征矢量或者权值，所述根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值，包括：
若终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波；
若终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波；
若终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
在一种具体实现方式中，所述对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波，包括：
基于子空间距离及梯度下降法对所述终端的SRS预测权值进行最小均方滤波。
在一种具体实现方式中，所述对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波，包括：
基于卡尔曼滤波类预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于归一化最小均方自适应滤波预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于递推最小二乘法RLS预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于自回归AR滤波的预测算法对所述终端的SRS信道预测矢量或者权值进行滤波。
在一种具体实现方式中，所述基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息，包括：
对接收到的SRS位置处待处理的频域信号进行预处理，并根据预处理后的频域信号获取每个终端发送SRS的信道信息；
其中，所述预处理包括以下至少一种处理：
根据公式(1)：
Figure PCTCN2020086711-appb-000001
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y(n)为SRS位置处的频域信号，w(n)为频域窗系数，
Figure PCTCN2020086711-appb-000002
为频域滤波后信号，n＝0,…,N为信道估计位置索引，N为信道估计的长度；
根据公式(2)：
Figure PCTCN2020086711-appb-000003
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y为SRS位置处的频域信号，维度为N×1；w为频域窗系数，维度为N×N；
Figure PCTCN2020086711-appb-000004
为频域滤波后信号，维度为N×1，N为信道估计的长度；
根据公式(3)：
Figure PCTCN2020086711-appb-000005
对接收到的SRS位置处待处理的频域信号进行时域干扰消除；其中，y为SRS位置处的待处理的频域信号，维度为N×1；
Figure PCTCN2020086711-appb-000006
为时域对消变换函数，
Figure PCTCN2020086711-appb-000007
为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
在一种具体实现方式中，每个终端的信道信息包括部分子带的SRS频域信道估计结果，所述根据每个终端的信道信息进行信道预测，得到信道预测结果，包括：
根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
在一种具体实现方式中，所述根据每个终端的信道信息进行信道预测，得到信道预测结 果，包括：
根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
在一种具体实现方式中，所述通过RRC信令为多个终端配置基序列标识，包括：
向每个终端发送第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
在终端需要改变SRS基序列时，向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
在一种具体实现方式中，所述通过RRC信令为多个终端配置基序列标识，包括：
向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。
本申请第二方面提供一种提升多用户复用性能的方法，应用于终端，所述方法包括：
获取网络设备通过RRC信令配置的基序列标识，所述基序列标识指示的基序列与其他终端采用的基序列之间不具有正交性；
根据所述基序列标识，发送探测参考信号SRS。
在一种具体实现方式中，所述获取网络设备通过RRC信令配置的基序列标识，包括：
接收所述网络设备发送的第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；
根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送；
或者，
接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
本申请第三方面提供一种提升多用户复用性能的装置，包括：
发送模块，用于通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性；
处理模块，用于基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息；
所述处理模块还用于根据每个终端的信道信息进行信道预测，得到信道预测结果。
可选的，所述处理模块还用于：
根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；
根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。
可选的，终端的信道预测结果包括SRS信道特征矢量或者权值，所述处理模块具体用于：
若终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波；
若终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波；
若终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
可选的，所述处理模块具体用于：
基于子空间距离及梯度下降法对所述终端的SRS预测权值进行最小均方滤波。
可选的，所述处理模块具体用于：
基于卡尔曼滤波类预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于归一化最小均方自适应滤波预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于递推最小二乘法RLS预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于自回归AR滤波的预测算法对所述终端的SRS信道预测矢量或者权值进行滤波。
可选的，所述处理模块具体用于：
对接收到的SRS位置处待处理的频域信号进行预处理，并根据预处理后的频域信号获取每个终端发送SRS的信道信息；
其中，所述预处理包括以下至少一种处理：
根据公式(1)：
Figure PCTCN2020086711-appb-000008
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y(n)为SRS位置处的频域信号，w(n)为频域窗系数，
Figure PCTCN2020086711-appb-000009
为频域滤波后信号，n＝0,…,N为信道估计位置索引，N为信道估计的长度；
根据公式(2)：
Figure PCTCN2020086711-appb-000010
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y为SRS位置处的频域信号，维度为N×1；w为频域窗系数，维度为N×N；
Figure PCTCN2020086711-appb-000011
为频域滤波后信号，维度为N×1，N为信道估计的长度；
根据公式(3)：
Figure PCTCN2020086711-appb-000012
对接收到的SRS位置处待处理的频域信号进行时域干扰消除；其中，y为SRS位置处的待处理的频域信号，维度为N×1；
Figure PCTCN2020086711-appb-000013
为时域对消变换函数，
Figure PCTCN2020086711-appb-000014
为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
可选的，每个终端的信道信息包括部分子带的SRS频域信道估计结果，所述处理模块还具体用于：
根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
可选的，所述处理模块还具体用于：
根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
可选的，所述发送模块具体用于：
向每个终端发送第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
在终端需要改变SRS基序列时，向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
可选的，所述发送模块具体用于：
向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。
本申请第四方面提供一种提升多用户复用性能的装置，包括：
获取模块，用于获取网络设备通过RRC信令配置的基序列标识，所述基序列标识指示的基序列与其他终端采用的基序列之间不具有正交性；
发送模块，用于根据所述基序列标识，发送探测参考信号SRS。
可选的，所述获取模块具体用于：
接收所述网络设备发送的第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；
根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送；
或者，
接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
本申请第五方面提供一种网络设备，包括：
处理器、存储器、接收器和发送器；
存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行第一方面任一实现方式提供的提升多用户复用性能的方法。
本申请第六方面提供一种终端，包括：
处理器、存储器、接收器和发送器；
存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行第二方面任一项提供的提升多用户复用性能的方法。
本申请第七方面提供一种计算机可读存储介质，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行第一方面任一实现方式提供的提升多用户复用性能的方法。
本申请第八方面提供一种计算机可读存储介质，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行第二方面任一实现方式提供的提升多用户复用性能的方法。
本申请提供的提升多用户复用性能的方法、装置、设备和存储介质，网络设备通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性，终端获取网络设备通过RRC信令配置的基序列标识，根据基序列标识，发送探测参考信号SRS，网络设备基于准正交序列对多个终端进行SRS检测，获取每个终端发送SRS的信道信息，根据每个终端的信道信息进行信道预测，得到信道预测结果。基于准正交序列的SRS配置和检测、进行信道预测，计算终端的权值，可以有效提用户移动场景，小区的平均吞吐量以及用户的平均感知速率。
附图说明
图1为本申请提供的提升多用户复用性能的方法实施例一的流程图；
图2为本申请提供的SRS基序列示意图；
图3为本申请提供的重构示意图；
图4为本申请提供的SRS IC过程示意图；
图5为本申请提供的滤波干扰消除示意图；
图6为本申请提供的多用户滤波干扰消除模块A的示意图；
图7为本申请提供的多用户滤波干扰消除模块B0的示意图；
图8为本申请提供的多用户滤波干扰消除模块B1的示意图；
图9为本申请提供的SRS时频资源示意图；
图10为本申请提供的基于时频空二维互校准的SRS信道预测方案架构图；
图11为本申请提供的基站侧天线位置和极化方式示意图；
图12为本申请提供的TDD LTE外场采集信道样本统计协方差阵结构示意图；
图13为本申请提供的统计协方差阵的TBT结构示意图；
图14为本申请提供的Toeplitz块阵压缩近似方式示意图；
图15为本申请提供的Toeplitz矩阵压缩近似方式示意图；
图16为本申请提供的单极化TBT矩阵压缩前后对比示意图；
图17为本申请提供的提升多用户复用性能的装置实施例一的结构示意图；
图18为本申请提供的提升多用户复用性能的装置实施例二的结构示意图。
具体实施方式
常见的在多用户场景中的信道预测方案中，不区分移动用户和静止用户，这样不利于针对移动用户进行针对性的算法设计和优化。在计算单用户(Single User，SU)波束成形(Beam Forming，BF)权值中使用的信道信息来自于最近一次的探测参考信号(Sounding Reference Signal，SRS)，而SRS信息发送的时间间隔都比较长，一般为几十ms的量级，而用户调度的周期一般是ms量级，这中间存在着明显的使用信息与当前真实的信道信息不匹配的问题。或者在进行多用户配对时，使用迫零的方法获得预编码权值，该权值对信道的准确度十分敏感，计算该权值使用的信道信息来自最近一次的SRS信息，当用户移动时，该信息与当前的真实信道不匹配，成为移动用户性能下降的主要原因。
还有一种常见的方案，在移动用户识别时用的是用户不同时刻的信道。然而，因为信道存在衰落和很多其他非理想因素，索引用户信道来判别用户的移动状态不一定准确，在进行信道预测时，采用简单的信道加权和对信道的非理想因素鲁棒性很差，例如：不同时刻的信道存在定时偏差和终端侧相位跳变的问题，导致不同时刻信道存在较大相差，简单的加权和信道来进行预测会导致错误的预测估计结果。
针对上述存在的一些问题，本申请提供一种能够提升多用户复用性能的方法，具体具体改进点包括以下几个：基于准正交序列的SRS配置和检测、基于频域外推的SRS信道预测、基于时频空二维互校准的SRS信道预测，用户移动状态和预测精度判别、自适应用户权值计算，包括根据用户状态判别结果，在三种权值中进行选择：1、基于最小子空间距离的权值预测，2、基于自回归理论的权值预测，3、基于统计协方差矩阵配对后，进行信号与干扰加噪声比(Signal to Interference plus Noise Ratio，SINR)修正。本专利中的数据元素除特别说明外，均定义为复数。
下面通过几个具体实施方式，对本申请提供的提升多用户复用性能的方法进行详细描述。
图1为本申请提供的提升多用户复用性能的方法实施例一的流程图，如图1所示，该提升多用户复用性能的方法主要应用在网络设备中，为终端配置基序列标识，终端根据该基序列标识发送SRS，网络设备进行SRS检测，进行信道预测，具体包括以下步骤：
S101：通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性。
在本步骤中，网络侧，也就是网络设备可以选择通过终端RRC信令配置给终端SRS基序列标识，终端通过接收基站侧发送的RRC信令来获得SRS基序列标识。该SRS基序列标识用于生成SRS基序列，同时，该方案中的每个终端的基序列之间不具有正交性。在该方案的具体实现中，至少包括以下两种情况：
第一种情况，网络设备向每个终端发送第一RRC信令，所述第一RRC信令携带至少一 个基序列标识，然后在终端需要改变SRS基序列时，网络设备再向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。对于终端来说，则接收所述网络设备发送的第一RRC信令，在需要改变SRS基序列时候，接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
在具体实现中，以网络设备是基站为例，基站可以预先通过RRC信令0配置给终端一套SRS基序列标识，其中包括至少一个SRS基序列标识，然后每次需要终端改变SRS基序列时，基站通过RRC信令1配置终端一个索引标识，该标识用于指示终端采用上述一套SRS基序列标识中多个SRS基序列标识中的一个来进行SRS的发送。
第二种情况，网络设备向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。对于终端来说，则接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
在该方案的具体实现中，以网络设备是基站为例，基站可以通过RRC信令0配置给终端一个SRS基序列标识，该序列标识指示的范围包括所有SRS基序列标识的候选集合，然后每次需要终端改变SRS基序列时，基站通过RRC信令0配置终端另外一个SRS基序列标识，该序列标识指示的范围包括所有SRS基序列标识的候选集合。
图2为本申请提供的SRS基序列示意图，如图2所示，本方案中，基序列名词仅仅为了区别不同终端SRS的循环移位导致的SRS序列不同。如果不同用户通过配置循环移位生成SRS序列，序列变换到时域是加窗可分的，即不同用户占用不用的时域段，如用户0～7采用不同的循环移位SRS序列，在时域是可分的。如果不同用户通过配置不同的基序列生成SRS序列，序列变换到时域是加窗不可分的，如用户0和用户8之间。
S102：根据基序列标识，发送SRS。
在本步骤中，而对于终端侧来说，在网络设备配置了基序列标识之后，对应的终端获取网络设备通过RRC信令配置的基序列标识，然后根据上述描述获取对应的基序列，然后发送SRS。
S103：基于准正交序列对多个终端进行SRS检测，获取每个终端发送SRS的信道信息。
在本步骤中，在终端根据网络设备配置的基序列标识进行SRS发送之后，则网络设备可以进行SRS检测，在本申请中则采用基于准正交序列对多个终端的SRS进行检测，以获取到每个终端发送SRS的信道信息。传统信道估计算法基站采用把接收的参考信号序列进行去参考信号序列后，进行不同用户的时域加窗，从而把复用的多个不同SRS循环移位序列的用户分离开来。但是，当网络设备给复用的多个终端配置多个基序列时，传统信道估计方法无法分离不同基序列的终端(也可以称为用户)，从而造成不同基序列终端间的序列干扰，导致信道估计性能下降。下述PDP代表功率时延分布的简称。
本申请提供的基于准正交序列的SRS检测方法，可以对接收到的SRS位置处待处理的频域信息进行预处理，然后根据预处理后的频域信号货到每个终端发送SRS的信道信息，该预处理过程至少包括以下操作过程中的一个过程，也可以包括多个操作过程的组合：
1、网络设备对接收到的SRS位置处待处理的频域信号进行频域滤波：
Figure PCTCN2020086711-appb-000015
其中，y(n)为SRS位置处的频域信号，w(n)为频域窗系数，
Figure PCTCN2020086711-appb-000016
为频域滤波后信号，n＝ 0,…,N为信道估计位置索引，N为信道估计的长度。w(n)可以预存预先计算好进行存储或者通过一些参数在线计算获得。优选的，在进行多天线或者多波束维度上，多个天线和多个波束可以应用相同的w(n)。
2、网络设备对接收到的SRS位置处待处理的频域信号进行频域滤波：
Figure PCTCN2020086711-appb-000017
其中，y为SRS位置处的频域信号，维度为N×1。w为频域窗系数，维度为N×N。
Figure PCTCN2020086711-appb-000018
为频域滤波后信号，维度为N×1，N为信道估计的长度。w可以预先计算好进行存储或者通过一些参数在线计算获得，所述参数包括但不限于：SRS序列，信噪比信息，时域窗相关系数信息(包括子径的数目，子径的量化时延，子径的幅度和相位等等)等。优选的，在进行多天线或者多波束维度上，多个天线和多个波束可以应用相同的w(n)。
3、网络设备对接收到的SRS位置处的频域信号进行时域干扰消除：
Figure PCTCN2020086711-appb-000019
其中，y为SRS位置处的待处理的频域信号，维度为N×1。
Figure PCTCN2020086711-appb-000020
为时域对消变换函数(也可以表示为
Figure PCTCN2020086711-appb-000021
为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
Figure PCTCN2020086711-appb-000022
的作用在于把y在时域消除不同基序列信道之间互干扰，得到最小化基序列间干扰的各个独立基序列信道，再在每个基序列上进行时域加窗降噪后变换的频域，获得不同SRS序列对应的信道值。
其中，
Figure PCTCN2020086711-appb-000023
函数可以等效为执行以下操作至少之一：
1)对待处理频域信号进行时频变换，变换到时域，在时域按照顺序或者并行消除不同基序列间干扰，然后每个基序列独立按照时域加窗的方式进行信道估计。
2)对不同基序列和/或不同序列的SRS分组，即相同基序列的SRS分为一组，这样可以分为若干组SRS序列。
3)按照一定准则给不同SRS分组(和/或者序列和/或子径)进行排序，对应准则包括但不限于不同分组的平均信干噪比，SRS序列的数量，SRS序列的平均接收功率等等。
4)按照排序结果给第p组SRS序列进行最小二乘信道估计：
Figure PCTCN2020086711-appb-000024
对
Figure PCTCN2020086711-appb-000025
进行频域到时域的变换，把频域信号
Figure PCTCN2020086711-appb-000026
变换到时域，在时域选择满足预设条件的时域点(或者子径)置零，和/或，记录对应置零前时域点的值L p，L p为第p次执行记录的时域点值的集合，然后进行时域到频域的变化，把处理后的时域信号变换到频域
Figure PCTCN2020086711-appb-000027
然后进行加参考信号操作：
Figure PCTCN2020086711-appb-000028
如上操作过程可以执行K次，K可选的可以等于SRS的组数，或者满足预设条件的已经执行的次数。
5)对第4)步存储L p集合进行加回操作：
Figure PCTCN2020086711-appb-000029
Figure PCTCN2020086711-appb-000030
代表在L p时域对应的点上加回对应L p的值。
6)对上一次操作获得的每个组的时域值y p进行加窗降噪，即利用预先计算好或者存储的窗W p,j对y p进行加窗处理：
Figure PCTCN2020086711-appb-000031
其中，
Figure PCTCN2020086711-appb-000032
为第p组第j个用户时域信道系数，W p,j为第p组第j个用户时域窗系数。然后对得到第p组第j个用户时域窗系数进行时域到频域的变换获得第p组第j个用户频域信道系数。
7)计算各个SRS基序列的互相关序列
Figure PCTCN2020086711-appb-000033
Figure PCTCN2020086711-appb-000034
N base为基序列数目，和/或补零至N FFT点后的离散傅里叶变换(Inverse Discrete Fourier Transform，IDFT)或者快速傅里叶逆变换(Inverse Fast Fourier Transform，IFFT)操作或者等效操作，N FFT为离散傅里叶变换(Discrete Fourier Transform，DFT)或者IFFT操作点数，结果记为
Figure PCTCN2020086711-appb-000035
8)按p＝1,2,…,N base的顺序进行以下计算：
a)按照一定准则搜索第p小区大时域径点。
b)在其他小区的PDP内消除第p小区大时域径点。
可选的，对搜索结果中满足一定准则的所有l以及
Figure PCTCN2020086711-appb-000036
的所有
Figure PCTCN2020086711-appb-000037
计算
Figure PCTCN2020086711-appb-000038
Figure PCTCN2020086711-appb-000039
(a th,p，p＝1,2,…,N base是可调参数)或者理论等效操作。其中，y PDPp为y PDPp补零或者加虚拟子载波外推至N FFT点并IDFT得到PDP结果。加虚拟子载波外推的方法包括，利用已有的全部和部分子载波的信号通过乘以预存或者在线计算的权值，获得子载波外的信号值。所述权值计算可以利用最小均方误差准则和/或信噪比参数和/或SRS序列相关参数获得。％为取模符号。
9)对干扰消除后的y PDPp进行DFT操作并取子载波对应位置的值，结果记为
Figure PCTCN2020086711-appb-000040
对
Figure PCTCN2020086711-appb-000041
继续进行基线SRS信道估计得到目标小区所有UE的频域信道估计结果。
10)迭代干扰消除，利用各小区干扰消除后的PDP和原始PDP重复进行干扰消除X次，其中X为预定义或者满足一定条件的迭代次数。
11)重构出更精确的子载波用户干扰消除，指的是用部分大径时域点重构出更加完整的PDP，从而提高干扰消除的质量。
12)重构操作可以表示为：
Figure PCTCN2020086711-appb-000042
或者等效操作，其中1 T代表T×1维全1列矢量。假设
Figure PCTCN2020086711-appb-000043
代表第p小区选出的Y个大径时域点(Y的维数为通过预定义，或者预定义时域点的最大选择数目，或者满足一定条件的径时域点数)，则重构的第一步为求解方程
Figure PCTCN2020086711-appb-000044
Figure PCTCN2020086711-appb-000045
(类似反卷积或者等效操作)或者等效操作，其中A是根据大径数目和位置从
Figure PCTCN2020086711-appb-000046
中选取相应元素构成的矩阵，其中，A的维度为Y×Y,x代表用这些大径拟合的时域径，x的维度为Y×1。求得x后再与
Figure PCTCN2020086711-appb-000047
进行圆周卷积或者等效操作，就得到了长度为N FFT的重构PDP，用该重构PDP进行干扰消除就可以消除掉与大径强相关的弱径分量。图3为本申请提供的重构示意图，如图3所示，该图只包含一条真径且没有干扰、噪声。图3中圆点代表没有重构的干扰消除方案筛选出的一组大径(即第一组大径)，交叉代表用更高阈值筛选出的3条重构用大径(即第二组大径)，较细的线条为求逆、圆周卷积后得到的重构PDP(Power Delay Profile，功率时延分布)(即重构路径)。得到重构PDP后即可把与第一组大径位置不重叠的其他弱径一起消除掉(重叠位置还是以第一次筛选出的大径为准)。
根据上述的描述，下面提供几种具体实现过程中获取终端的信道信息的实施例：
实施例1：
图4为本申请提供的SRS干扰消除(Interference Cancellation，IC)过程示意图，如图4所示，横坐标从左到右分别是三个基序列的频域接收信号经过各自序列最小二乘(Least Square，LS)操作，再变换到时域的PDP图谱记为PDP 0，PDP 1，PDP 2，然后计算PDP图谱中信号功率对信号时域窗外干扰噪声平均功率归一化结果，记为时域点选择值PDP 0,NI。第一步，对于第一个基序列，时域点选择值PDP 0,NI中大于预设门限的若干个时域点值T 0，然后利用第一个基序列和其他基序列的相关性，在其他两个PDP谱中消除第一基序列时域点值T 0的干扰。更新第二基序列和第三基序列时域的PDP图谱。第二步，对于第二个基序列，时域点选择值PDP 1,NI中大于预设门限的若干个时域点值T 1，然后利用第一个基序列和其他基序列的相关性，在其他两个PDP谱中消除第一基序列时域点值T 1的干扰。更新第一基序列和第三基序列时域的PDP图谱。第三步，对于第三个基序列，时域点选择值PDP 2,NI中大于预设门限的若干个时域点值T 2，然后利用第一个基序列和其他基序列的相关性，在其他两个PDP谱中消除第一基序列时域点值T 2的干扰。更新第一基序列和第三基序列时域的PDP图谱。然后分别输出三个基序列对应的时域PDP，后续可以进行加窗降噪和时域到频域的变换操作。
可选的，可以引入迭代机制，把上一段输出的三个基序列时域PDP作为下一次迭代的输入，按照上一段的步骤，再进行一轮干扰消除，迭代次数可以预设，或者满足某一条件停止迭代。
实施例2：
把图4的操作作为模块A，在模块A之前可以进行频域滤波，图5为本申请提供的滤波干扰消除示意图，如图5所示。例如：SRS接收频域接收信号为：y,y的维度为S×1，ω为频域滤波系数，ω的维度为S×1，S为子载波的数目，对y进行频域滤波：
Figure PCTCN2020086711-appb-000048
其中，“*”表示元素乘，
Figure PCTCN2020086711-appb-000049
为频域滤波后频域值。
经过滤波后的频域值
Figure PCTCN2020086711-appb-000050
作为模块A的输入信号xn，经过处理(不迭代或者迭代多次)后，输出yn。
实施例3：
把图4的操作作为模块A，在模块A中的每一个基序列大径时域点选择后，干扰消除前进行时域径重构。重构操作可以表示为：
Figure PCTCN2020086711-appb-000051
或者等效操作，其中1 T代表T×1维全1列矢量。假设
Figure PCTCN2020086711-appb-000052
代表第p小区选出的Y个大径时域点(Y的维数为通过预定义，或者预定义时域点的最大选择数目，或者满足一定条件的径时域点数)，则重构的第一步为求解方程
Figure PCTCN2020086711-appb-000053
(类似反卷积或者等效操作)或者等效操作，其中A是根据大径数目和位置从
Figure PCTCN2020086711-appb-000054
中选取相应元素构成的矩阵，其中，A的维度为Y×Y,x代表用这些大径拟合的时域径，x的维度为Y×1。求得x后再与
Figure PCTCN2020086711-appb-000055
进行圆周卷积或者等效操作，就得到了长度为N FFT的重构PDP，用该重构PDP进行干扰消除就可以消除掉与大径强相关的弱径分量。图3为上述重构过程的示意图，该图只包含一条真径且没有干扰、噪声。图中圆点代表没有重构的干扰消除方案筛选出的一组大径(第一组大径)，交叉代表用更高阈值筛选出的3条重构用大径(第二组大径)，细线为求逆、圆周卷积后得到的重构PDP(重构路径)。得到重构PDP后即可把与第一组大径位置不重叠的其他弱径一起消除掉(重叠位置还是以第一次筛选出的大径为准)。
实施例4：
把图4的操作作为模块A，在模块A之前可以进行频域滤波步骤如公式(8)，在模块A中的每一个基序列大径时域点选择后，干扰消除前进行时域径重构。重构操作可以表示为：
Figure PCTCN2020086711-appb-000056
Figure PCTCN2020086711-appb-000057
或者等效操作，其中1 T代表T×1维全1列矢量。假设
Figure PCTCN2020086711-appb-000058
代表第p小区选出的Y个大径时域点(Y的维数为通过预定义，或者预定义时域点的最大选择数目，或者满足一定条件的径时域点数)，则重构的第一步为求解方程
Figure PCTCN2020086711-appb-000059
(类似反卷积或者等效操作)或者等效操作，其中A是根据大径数目和位置从
Figure PCTCN2020086711-appb-000060
中选取相应元素构成的矩阵，其中，A的维度为Y×Y,x代表用这些大径拟合的时域径，x的维度为Y×1。求得x后再与
Figure PCTCN2020086711-appb-000061
进行圆周卷积或者等效操作，就得到了长度为N FFT的重构PDP，用该重构PDP进行干扰消除就可以消除掉与大径强相关的弱径分量。图3为上述重构过程的示意图，该图只包含一条真径且没有干扰、噪声。图中圆点代表没有重构的干扰消除方案筛选出的一组大径(第一组大径)，交叉代表用更高阈值筛选出的3条重构用大径(第二组大径)，细线为求逆、圆周卷积后得到的重构PDP(重构路径)。得到重构PDP后即可把与第一组大径位置不重叠的其他弱径一起消除掉(重叠位置还是以第一次筛选出的大径为准)。
实施例5：
图6为本申请提供的多用户滤波干扰消除模块A的示意图，如图6所示，输入为x0，首先对输入的x0进行矩阵频域滤波，其中滤波矩阵A -1维度为T T，滤波后输出为y0，y0对第p n个基序列进行LS操作：y f,pn＝y0*p n *，然后对y pn进行时频变换为时域序列y t,pn，利用信噪比的参数计算出或者预存软窗系数
Figure PCTCN2020086711-appb-000062
对y t,pn在时域执行加软窗操作：
Figure PCTCN2020086711-appb-000063
对
Figure PCTCN2020086711-appb-000064
时频变换变换到频域
Figure PCTCN2020086711-appb-000065
然后乘以对应导频序列p n，获得
Figure PCTCN2020086711-appb-000066
上述所有操作记为模块A。
图7为本申请提供的多用户滤波干扰消除模块B0的示意图，模块B0集成了模块A的操作，利用模块A输出的
Figure PCTCN2020086711-appb-000067
来进行进一步信号处理，如图7所示，更新
Figure PCTCN2020086711-appb-000068
然后对
Figure PCTCN2020086711-appb-000069
进行降噪处理(例如，变换到时域进行加窗降噪或者时域点选择的降噪算法)，输出
Figure PCTCN2020086711-appb-000070
然后第n+1个基序列利用前n个更新过的
Figure PCTCN2020086711-appb-000071
更新
Figure PCTCN2020086711-appb-000072
输出的序列
Figure PCTCN2020086711-appb-000073
为经过时域窗用户分离过程的同一基序列不同循环移位每个用户的频域数据输出。
实施例6：
图8为本申请提供的多用户滤波干扰消除模块B1的示意图，如图8所示，模块B1集成了模块A的操作，利用模块A输出的
Figure PCTCN2020086711-appb-000074
来进行进一步信号处理，更新
Figure PCTCN2020086711-appb-000075
然后对
Figure PCTCN2020086711-appb-000076
进行降噪处理和时域窗用户分离过程的同一基序列不同循环移位每个用户的频域数据输出输出
Figure PCTCN2020086711-appb-000077
根据上述任一方案进行SRS检测处理，可以得到每个终端的信道信息。
S104：根据每个终端的信道信息进行信道预测，得到信道预测结果。
在本步骤中，网络设备获取到每个终端的信道信息之后，可以进行信道预测，本申请提供了以下几种预测方式：
第一种预测方式，基于频域外推的SRS信道预测。
该方案中，每个终端的信道信息包括部分子带的SRS频域信道估计结果，网络设备根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
在实际系统中，考虑到终端发送功率的约束，SRS的发送采用部分带宽的方式进行发送，图9为本申请提供的SRS时频资源示意图，如图9所示，SRS时频资源包括时域和频域，图中横向为时间，纵向为频域，SRS四次跳频跳满SRS全带宽，每用户每次使用1/4SRS全带宽进行承载SRS信号。
这里希望能够使用部分子带的SRS频域信道估计结果外推至全带宽，进而得到了全带宽的信道估计结果。通过这样的方式，每个用户仅需要使用1/4的SRS带宽即可实现信道信息的反馈，此时SRS容量是基线的4倍，在SRS总量不变的情况下，此时每个用户的反馈周期缩减至原来的1/4，进而实现了频域的扩容，达到提升移动场景性能的目的。
这里主要用的是时域稀疏性，也可以表示为频域信号由少量时域稀疏径基张成的空间构 成。利用稀疏性时域信号可以表示成：
Figure PCTCN2020086711-appb-000078
因此，利用部分带宽得到多径时延与对应的幅值，即可得到全带宽的频域信号，其中N paths为时域稀疏径的数目，a p为时域径的幅度和相位参数，W p(t-τ p)为多普勒频移，多普勒扩展以及多径时延随时间带来的相位和/或幅度变化参数。利用获得的部分SRS带宽频域数据估计各个时域点的N paths、A p以及W p和τ p参数，然后通过各个参数恢复出h(t)，再利用时域到频域的变换，恢复出其他频带的SRS信道信息。优选的，可以通过计算频域滤波系数来外推出其他频带的信道信息：
H extended＝WH exist      (10)
其中，W的维度为N×S列，H exist的维度为S×1，H extended的维度为N×1。
Figure PCTCN2020086711-appb-000079
其中，N为全带宽的子载波数目，S为部分带宽的子载波数目，f 0,0为映射函数，把部分带宽各个子载波的频域信道映射到全部带宽的各个子载波上。
优选的，可以通过以下方式寻找时域相关参数：     (11)
把频域信号通过稀疏基变换，变换到稀疏基域，通过1范数或者0范数稀疏化方法，迭代逐个(组)选择满足一定约束条件的基来作为候选基，当满足一定条件时，停止迭代，把候选基集合通过稀疏域到频域信号的变换，恢复出全带宽的频域信道。
第二种预测方式，基于时频空二维互校准的SRS信道预测。
该方案中，网络设备根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
在该方案的具体实现中，在时域基于开尔曼滤波类预测算法，或者自回归预测算法通过历史的SRS的信道预测未来SRS信道的值。图10为本申请提供的基于时频空二维互校准的SRS信道预测方案架构图，如图10所示：虚线部分表示可选部分，具体操作包括以下步骤至少之一：
1、利用已经估计得出的N个历史SRS信道值，通过时域信道预测算法方案估计出0时刻的信道估计值，其中1时刻为当前时刻，N-1～2时刻为历史时刻，0时刻为未来时刻，优选的，对应时域信道预测算法方案可以采用卡尔曼滤波类算法，或者自回归算法。
2、利用已经估计得出的N个历史SRS信道特征矢量或者权值，通过时域矢量预测算法方案估计出0时刻的信道特征矢量或者权值，其中1时刻为当前时刻，N-1～2时刻为历史时刻，0时刻为未来时刻，优选的，对应时域信道预测算法方案可以采用卡尔曼滤波类算法，或者自回归算法。
3、利用历史的S，(S≥1)个频域信道系数，通过频域外推算法来校正1、2步骤后获得结果，作为0时刻的信道或者矢量值。
4、采用线性插值算法，计算两个SRS时刻中间的若干个下行子帧的信道或者矢量。
5、利用时域预测中间参数和频域外推参数，以及SRS时刻1和/或SRS时刻0的信道值或者矢量值插值出两个SRS时刻1和0中间的若干个下行子帧的信道或者矢量。
通过上述的两种方法的多种实现方式，可以对终端的信道进行预测，得到相应的预测结果。
上述实施例提供的提升多用户复用性能的方法，网络设备通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性，终端获取网络设备通过RRC信令配置的基序列标识，根据所述基序列标识，发送探测参考信号SRS，网络设备基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息，根据每个终端的信道信息进行信道预测，得到信道预测结果。基于准正交序列的SRS配置和检测、进行信道预测，可以有效提用户移动场景，小区的平均吞吐量以及用户的平均感知速率。
在上述实施例的基础上，在该方案的具体实现中，网络设备还可以根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；然后根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。网络设备预测的终端的信道预测结果包括SRS信道特征矢量或者权值。
具体的实现方式如下，网络设备根据不同SRS时刻的信道计算出的特征矢量或者权值间的相关性来判断用户的移动状态。优选的，相关性计算结果可以进一步采用时域滤波或者频域滤波的方式进行处理。
优选的，根据下面的公式计算的权值相关性：
Figure PCTCN2020086711-appb-000080
ρ t＝ρ t-1+αρ t     (2)
其中，S为频域滤波的SRS的子载波数目，(1)中的ρ t为频域滤波后的相关值，(2)中的ρ t为时域滤波后的相关值，α为时域滤波系数。
根据计算出的ρ t值，判断用户的移动性状态，根据用户不同移动状态，选择不同权值计算方式。优选的，当ρ t≤γ 0，采用权值计算方式0计算权值，当γ 0≤ρ t≤γ 1，采用权值计算方式1计算权值，当ρ t＞γ 1，采用权值计算方式2计算权值。其中，γ 0＜γ 1。
1、当ρ t≤γ 0，可以考虑终端为准静止状态，这时的权值优化思路为假定终端的权值稳定，可以对终端的权值进行类型一空域和/或者时域滤波。即终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波。
一种可选的方式中，基于子空间距离及梯度下降法实现对权值的最小均方滤波。设w(t) 为要预测SRS时刻的预测权值，v(t)为要预测SRS时刻的真实权值。用二者的子空间距离作为目标函数：
Figure PCTCN2020086711-appb-000081
进一步，可选的，
w(t)＝w(t-1)+μv(t-1)v H(t-1)w(t-1)     (4)
其中，w(t)为t时刻滤波后权值，w(t-1)为t-1时刻滤波后权值，μ为梯度下降系数，v(t-1)为t-1时刻的损失权值。
2、当γ 0≤ρ t≤γ 1，可以考虑终端为低速移动状态，这时的权值优化思路为假定终端不同时刻权值具有相关性，可以对终端的权值进行类型二空域和/或者时域预测。即终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波。
可选的，基于卡尔曼滤波类预测算法，包括其增强算法类。
可选的，基于归一化最小均方自适应滤波(NLMS)预测算法，包括其增强算法类。
可选的，基于递推最小二乘算法(RLS)预测算法，包括其增强算法类。
可选的，基于自回归(AR)滤波的预测算法，包括其增强类算法。
进一步地，
将N个历史SRS信道矢量或者权值拼在一起写成矩阵形式，
X n＝[x n-N,x n-N+2,…,x n-1]     (5)
根据自回归模型，预测向量可以由下式得到，
y n＝X nα     (6)
α＝[α 1,α 2,…,α N] T     (7)
由于X n是列满秩矩阵，可以由X n的左伪逆矩阵求出α，
Figure PCTCN2020086711-appb-000082
由于预测值无法提前得到，也就无法求出精确的α值。如果往前推一个时刻，是可以得到α n-1的，即：
Figure PCTCN2020086711-appb-000083
其中，X n的维度为N t×N，α的维度为N×1，y n的维度为N×1。
进一步地，传统AR无法解决不同SRS时刻的相位跳变问题，对其进行改进，将预测目标 从权值变成了不同时刻组成的基的投影权值，通过这种投影的方式能够去除相位跳变的影响，此时
H＝[v(t-1),v(t-2)v H(t-2)v(t-1)，…,v(t-N)v H(t-N)v(t-1)]   (10)
H 1＝[v(t-1),v(t-2)v H(t-2)v(t-1)，…,v(t-N)v H(t-N)v(t-1)]   (11)
因此，预测时刻v′(t)可以表示为：
Figure PCTCN2020086711-appb-000084
3、当ρ t＞γ 1，可以考虑终端为中高速移动状态，这时的权值优化思路为假定终端不同时刻权值相关性较低，可以对终端的信道进行二阶距计算，生成统计特性的权值。即终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
可选的，在时域进行统计权计算，SRS时频资源包括时域和频域。因此相应的，这里的统计方案可以是频域统计权、时域统计权，还可以是时频二维统计权。
(1)频域统计权
根据最近SRS时刻(考虑全带宽)的1/S SRS全带宽对应的信道系数求统计相关矩阵，这样得到的统计权能保证在时间上距离当前时刻最近，其中N Sc为频域统计的SRS子载波数目，h(k)为第k子载波的信道矩阵或者矢量，R f代表频域统计自相关矩阵。
Figure PCTCN2020086711-appb-000085
(2)时域统计权
对历史SRS时刻(只考虑本带)先在频域求平均，再在时域alpha滤波，时域alpha滤波的周期为S倍配置的SRS周期，其中α为时域滤波系数，R t为t时刻相关矩阵。
R t＝(1-α)R t-1+αR t    (14)
(3)时频域统计权
时频域统计权是频域统计权和时域统计权的结合。对历史SRS时刻(考虑全带宽)先在频域求平均，再在时域alpha滤波，时域alpha滤波的周期为配置的SRS周期。α为1时，退化成频域统计权。
可选的，终端在多用户(Multiple User，MU)计算方法具体包括以下几种：
(1)正则化增强迫零(REZF)算法
基线采用REZF进行MU权值计算
W＝H(H HH+δ 2I) -1    (15)
其中，H为多个用户SU权值拼成的列满秩矩阵，δ 2为正则化系数。基于统计权的REZF算 法即将基线SU权值换成由统计相关阵做奇异值分解(Singular Value Decomposition，SVD)得到的权值(最大特征值对应的特征向量)。
(2)信漏噪比(SLNR)最大化算法
SLNR算法准则是实现信号功率相对于泄露到其他小区的信号功率及噪声最大化，从而使权值能够兼容对邻小区的泄露问题及目标用户的信噪比问题，基于SLNR准则的权值生成约束如下所示，
Figure PCTCN2020086711-appb-000086
这里H k是第k个用户的频域信道，W k是对应的权值，
Figure PCTCN2020086711-appb-000087
是高斯白噪
Figure PCTCN2020086711-appb-000088
表示功率。
因此，其基于SLNR准则的权值最优值为：
Figure PCTCN2020086711-appb-000089
K为用户数(流数)，当H l为N T×1时，可以得出：
Figure PCTCN2020086711-appb-000090
其中V k是第k个用户的SU权值，上式是根据瞬时信道系数推导出的SLNR_W算法，直接将统计相关阵及由统计相关阵SVD分解得到的主特征向量代入上式得
Figure PCTCN2020086711-appb-000091
这里的V k可以是统计协方差阵R k对应的主特征向量，也可以用瞬时协方差阵的主特征向量。使用统计SU权值能够较好地滤波，降低移动性带来的影响，而使用瞬时SU权值能够克服频选等因素，在中低速场景能够与实际信道更加匹配。在不同场景具体使用哪一种SU权值是需要进行评估的。
这里可以看到SLNR算法在原理上能够更好地避免泄漏造成对其他用户造成的干扰，因此算法预期性能较优。但可以看到SLNR算法需要求解大矩阵(2*m*n)×(2*m*n)的SVD分解或求逆运算，复杂度较高，难以在产品中实现。因此，接下来对算法进行简化。
(3)结构化统计权算法
因为目前基站侧天线的结构为HmVnPt，其中，m为天线阵同一行水平同极化阵子数目，n为天线阵同一列垂直同极化阵子数目，t为极化数目，图11为本申请提供的基站侧天线位置和极化方式示意图，如图11所示，图中M表示行数，N表示列数，P表示极化数，可以假设信道的统计相关阵满足如下结构：
Figure PCTCN2020086711-appb-000092
其中，R v1，R v2，R H1，R H2分别为n×n，n×n，m×m，m×m的矩阵，R v1，R H1对应一种极化，R v2，R H2对应另一种极化。R v1的求法为：先根据一种极化的垂直维n行对应的信道系数求出某一列的相关阵，再将m列对应的列相关阵平均。R H1的求法为：先根据一种极化的水平维m列对应的信道系数求出某一行的相关阵，再将n行对应的行相关阵平均。R v2，R H2的求法类似。将统计相关阵写成如此分块结构，就可以分块求解，从而避免大矩阵的SVD和求逆。
(4)EZF结构化统计权算法
因为目前基站侧天线的结构为HmVnPt，类似统计相关阵，可以假设信道的SU权值(每用户单流)满足如下结构：
Figure PCTCN2020086711-appb-000093
其中，V v1，V v2，V H1，V H2分别为n×1，n×1，m×1，m×1的矩阵，V v1，V H1对应一种极化，V v2，V H2对应另一种极化。V v1，V v2，V H1，V H2分别为R v1，R v2，R H1，R H2最大特征值对应的左特征向量。
为了提升性能，可以将两种极化对应的权值进行平均，可以先平均再SVD也可以先SVD再平均，其中先平均再SVD可以将SVD操作减少一半。先SVD再平均的结果为，
Figure PCTCN2020086711-appb-000094
先平均再SVD的结果为
Figure PCTCN2020086711-appb-000095
其中V v为(R v1+R v2)/2对应的主特征向量，V H为(R H1+R H2)/2对应的主特征向量。
(5)SLNR算法
使用结构化的统计相关阵和结构化的SU权值，可以推导出基于结构化统计权的SLNR算法。
利用克罗内克积的相关性质，
Figure PCTCN2020086711-appb-000096
将式(19)重写如下
Figure PCTCN2020086711-appb-000097
而后为了简化求逆，将上式近似改写成
Figure PCTCN2020086711-appb-000098
化简得到
Figure PCTCN2020086711-appb-000099
因此，最终得到
Figure PCTCN2020086711-appb-000100
因此，通过结构化的方式，可以将(2*m*n)×(2*m*n)大矩阵的求逆降维成m×m和n×n的求逆，同时可以将不同极化相关阵进行平均，即R′ v1＝R′ v2＝(R v1+R v2)/2，R′ H1＝R′ H2＝(R H1+R H2)/2，所以R vH1,k＝R vH2,k，减少求逆的次数，大大降低了复杂度，达到了产品的要求。
这里的统计R lvi和统计R lHi理论上满足Toeplitz结构，因此可以通过对角平均的方式将R lvi和R lHi矩阵进行Toeplitz化，从而实现进一步的矩阵压缩，便于存储。
(6)TBT统计权算法
TBT(Toeplitz-Block-Toeplitz)统计权主要利用了统计协方差矩阵的部分性质对单次计算的复杂度进行简化。
首先假设极化间协方差阵包含能量较少，因此将极化间协方差阵置零，图12为本申请提供的TDD LTE外场采集信道样本统计协方差阵结构示意图，如图12所示，其次考虑极化内协方差阵，假设URA天线阵面上若任意两个同极化振子间相对物理位置距离相等，那么阵子间的统计相关性也相等，因此在该假设下，单极化内信道统计协方差阵可以近似写成形如Toeplitz-Block-Toeplitz矩阵的结构。该矩阵具有两个特性：
1、统计协方差阵的每一块(对应垂直ULA统计协方差阵)，该矩阵为Toeplitz矩阵；
2、统计协方差阵的各块(对应水平ULA统计协方差阵)，该矩阵为Toeplitz矩阵；
图13为本申请提供的统计协方差阵的TBT结构示意图，基于上述两个性质，则最终用户的 统计协方差阵可近似表示为图13所示的结构。
将极化间统计协方差阵进行置零处理，而计划内统计协方差阵满足Toeplitz结构，(m*n)×(m*n)的矩阵可以看成有垂直ULA统计协方差阵构成的n*n块阵，每个块阵是一个m×m的Toeplitz矩阵，对应着水平ULA统计协方差阵。
基于上述的假设，对实际的统计协方差进行平均将协方差阵近似为TBT结构，具体思路如下：
第一步，进行托普利兹(Toeplitz)块平均。图14为本申请提供的Toeplitz块阵压缩近似方式示意图，如图14所示，对于一个极化的(m*n)×(m*n)矩阵，首先将近似相等的每个m×m矩阵块相加平均，而由于矩阵共轭对称，仅需平均对角线以上的块，如图14同色同字母块所示。
第二步，进行块内Toeplitz平均。图15为本申请提供的Toeplitz矩阵压缩近似方式示意图，对于每一块，我们需要将各对角线上理应相等的元素相加进行平均，如图15所示，每个线经过的块。由于每一个Toeplitz块并不一定是共轭对称块(除了对角线上的块)，因此需要计算全部X个对角线上的平均，最终得到一个X元素的向量。
第三步，合并成压缩矩阵。图16为本申请提供的单极化TBT矩阵压缩前后对比示意图，当完成前两部后，将n个m×m矩阵各自压缩成X元素向量最终组成n×X的压缩矩阵，如图16所示。TBT近似能够大幅降低了存储开销，同时还可以利用Toeplitz矩阵的性质简化矩阵求逆。
(7)TBT矩阵求逆简化
TBT形式的统计协方差阵和为：
Figure PCTCN2020086711-appb-000101
目标是求解
Figure PCTCN2020086711-appb-000102
1、计算
Figure PCTCN2020086711-appb-000103
(维度p×p)；
2、计算
Figure PCTCN2020086711-appb-000104
(维度p×p)；
3、计算
Figure PCTCN2020086711-appb-000105
(维度p×p)；
4、For i＝1:m-2
5、计算
Figure PCTCN2020086711-appb-000106
(维度p×p)，其中J i为i×i的反对角单位阵，
Figure PCTCN2020086711-appb-000107
Figure PCTCN2020086711-appb-000108
6、计算
Figure PCTCN2020086711-appb-000109
(维度p×p)；
7、计算
Figure PCTCN2020086711-appb-000110
(维度(i+1)p×p)；
8、End for
9、令
Figure PCTCN2020086711-appb-000111
代表
Figure PCTCN2020086711-appb-000112
的第(i,j)块子矩阵，计算
Figure PCTCN2020086711-appb-000113
10、对于
Figure PCTCN2020086711-appb-000114
j∈[i,m-i-1]区间的任意(i,j)组合
Figure PCTCN2020086711-appb-000115
11、利用
Figure PCTCN2020086711-appb-000116
反对角线对称的特性，利用已有块补全
Figure PCTCN2020086711-appb-000117
的右侧三角部分；
12、利用
Figure PCTCN2020086711-appb-000118
主对角线共轭对称的特性，利用已有块补全
Figure PCTCN2020086711-appb-000119
的左下部分。
通过上述方式，网络设备基于准正交序列的SRS配置和检测、基于频域外推的SRS信道预测、基于时频空二维互校准的SRS信道预测，用户移动状态和预测精度判别、自适应用户权值计算，可以有效提用户移动场景，小区的平均吞吐量以及用户的平均感知速率。
图17为本申请提供的提升多用户复用性能的装置实施例一的结构示意图，如图17所示，该提升多用户复用性能的装置10包括：
发送模块11，用于通过RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性；
处理模块12，用于基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息；
所述处理模块12还用于根据每个终端的信道信息进行信道预测，得到信道预测结果。
本实施例提供的提升多用户复用性能的装置，用于执行前述任一方法实施例中，网络设备侧的技术方案，其实现原理和技术效果类似，在此不再赘述。
在该提升多用户复用性能的装置10的具体实现方式中，所述处理模块12还用于：
根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；
根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。
进一步地，终端的信道预测结果包括SRS信道特征矢量或者权值，所述处理模块12具体用于：
若终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波；
若终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波；
若终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
可选的，所述处理模块12具体用于：
基于子空间距离及梯度下降法对所述终端的SRS预测权值进行最小均方滤波。
可选的，所述处理模块12具体用于：
基于卡尔曼滤波类预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于归一化最小均方自适应滤波预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
或者，
基于递推最小二乘法RLS预测算法对所述终端的SRS信道预测矢量或者权值进行滤波。
在一种具体实现方式中，所述处理模块12具体用于：
对接收到的SRS位置处待处理的频域信号进行预处理，并根据预处理后的频域信号获取每个终端发送SRS的信道信息；
其中，所述预处理包括以下至少一种处理：
根据公式(1)：
Figure PCTCN2020086711-appb-000120
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y(n)为参考信号位置处的频域信号，w(n)为频域窗系数，
Figure PCTCN2020086711-appb-000121
为频域滤波后信号，n＝0,…,N为信道估计位置索引，N为信道估计的长度；
根据公式(2)：
Figure PCTCN2020086711-appb-000122
对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y为参考信号位置处的频域信号，维度为N×1；w为频域窗系数，维度为N×N；
Figure PCTCN2020086711-appb-000123
为频域滤波后信号，维度为N×1，N为信道估计的长度；
根据公式(3)：
Figure PCTCN2020086711-appb-000124
对接收到的SRS位置处待处理的频域信号进行时域干扰消除；其中，y为参考信号位置处的待处理的频域信号，维度为N×1；
Figure PCTCN2020086711-appb-000125
为时域对消变换函数，
Figure PCTCN2020086711-appb-000126
为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
可选的，每个终端的信道信息包括部分子带的SRS频域信道估计结果，所述处理模块12还具体用于：
根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
可选的，所述处理模块12还具体用于：
根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
可选的，所述发送模块11具体用于：
向每个终端发送第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
在终端需要改变SRS基序列时，向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
可选的，所述发送模块11具体用于：
向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。
上述任一实施例提供的提升多用户复用性能的装置，用于执行前述方法实施例中的网络设备侧的技术方案，其实现原理和技术效果类似，在此不再赘述。
图18为本申请提供的提升多用户复用性能的装置实施例二的结构示意图，如图18所示，所述提升多用户复用性能的装置20包括：
获取模块21，用于获取网络设备通过RRC信令配置的基序列标识，所述基序列标识指示的基序列与其他终端采用的基序列之间不具有正交性；
发送模块22，用于根据所述基序列标识，发送探测参考信号SRS。
可选的，所述获取模块21具体用于：
接收所述网络设备发送的第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；
根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送；
或者，
接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
上述任一实施例提供的提升多用户复用性能的装置，用于执行前述方法实施例中的终端侧的技术方案，其实现原理和技术效果类似，在此不再赘述。
本申请还提供一种网络设备，包括：处理器、存储器、接收器和发送器；存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行前述任一实施例提供的提升多用户复用性能的方法的技术方案。
本申请还提供一种终端，包括：处理器、存储器、接收器和发送器；存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行前述任一实施例中终端侧的提升多用户复用性能的方法的技术方案。
在上述在网络设备、终端的实现中，存储器和处理器之间直接或间接地电性连接，以实现数据的传输或交互。例如，这些元件相互之间可以通过一条或者多条通信总线或信号线实现电性连接，如可以通过总线连接。存储器中存储有实现数据访问控制方法的计算机执行指令，包括至少一个可以软件或固件的形式存储于存储器中的软件功能模块，处理器通过运行存储在存储器内的软件程序以及模块，从而执行各种功能应用以及数据处理。
存储器可以是，但不限于，随机存取存储器(Random Access Memory，简称：RAM)，只读存储器(Read Only Memory，简称：ROM)，可编程只读存储器(Programmable Read-Only Memory，简称：PROM)，可擦除只读存储器(Erasable Programmable Read-Only Memory，简称：EPROM)，电可擦除只读存储器(Electric Erasable Programmable Read-Only Memory，简称：EEPROM)等。其中，存储器用于存储程序，处理器在接收到执行指令后，执行程序。进一步地，上述存储器内的软件程序以及模块还可包括操作系统，其可包括各种用于管理系统任务(例如内存管理、存储设备控制、电源管理等)的软件组件和/或驱动，并可与各种硬件或软件组件相互通信，从而提供其他软件组件的运行环境。
处理器可以是一种集成电路芯片，具有信号的处理能力。上述的处理器可以是通用处理器，包括中央处理器(Central Processing Unit，简称：CPU)、网络处理器(Network Processor，简称：NP)等。可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行前述任一方法实施例中网络设备侧的提升多用户复用性能的方法的技术方案。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行任一方法实施例中提升多用户复用性能的方法中终端侧的技术方案。
本申请还提供一种程序产品该程序产品包括计算机程序，该计算机程序存储在可读存储介质中，网络设备的至少一个处理器可以从可读存储介质读取该计算机程序，至少一个处理器执行该计算机程序使得网络设备实施前述任一方法实施例中的技术方案。
本申请还提供一种程序产品，该程序产品包括计算机程序，该计算机程序存储在可读存储介质中，终端的至少一个处理器可以从可读存储介质读取该计算机程序，至少一个处理器执行该计算机程序使得终端实施前述任一方法实施例中的技术方案。
本申请还提供一种芯片，所述芯片可应用于网络设备，所述芯片包括：至少一个通信接口，至少一个处理器，至少一个存储器，所述通信接口、存储器和处理器通过总线互联，所述处理器调用所述存储器中存储的计算机程序，以执行前述任一方法实施例中网络设备侧的技术方案。
本申请还提供一种芯片，所述芯片可应用于终端，所述芯片包括：至少一个通信接口，至少一个处理器，至少一个存储器，所述通信接口、存储器和处理器通过总线互联，所述处理器调用所述存储器中存储的计算机程序，以执行前述任一方法实施例中终端侧的技术方案。
本领域普通技术人员应理解：实现上述各方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成。前述的程序可以存储于一计算机可读取存储介质中。该程序在执行时，执行包括上述各方法实施例的步骤；而前述的存储介质包括：ROM、RAM、磁碟或者光盘等各种可以存储程序代码的介质，具体的介质类型本申请不做限制。

Claims (32)

  1. 一种提升多用户复用性能的方法，其特征在于，应用于网络设备，所述方法包括：
    通过无线资源控制RRC信令为多个终端配置基序列标识，每个终端的基序列标识指示的基序列之间不具有正交特性；
    基于准正交序列对所述多个终端进行探测参考信号SRS检测，获取每个终端发送SRS的信道信息；
    根据每个终端的信道信息进行信道预测，得到信道预测结果。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；
    根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。
  3. 根据权利要求2所述的方法，其特征在于，终端的信道预测结果包括SRS信道特征矢量或者权值，所述根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值，包括：
    若终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波；
    若终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波；
    若终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
  4. 根据权利要求3所述的方法，其特征在于，所述对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波，包括：
    基于子空间距离及梯度下降法对所述终端的SRS预测权值进行最小均方滤波。
  5. 根据权利要求3所述的方法，其特征在于，所述对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波，包括：
    基于卡尔曼滤波类预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于归一化最小均方自适应滤波预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于递推最小二乘法RLS预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于自回归AR滤波的预测算法对所述终端的SRS信道预测矢量或者权值进行滤波。
  6. 根据权利要求1至5任一项所述的方法，其特征在于，所述基于准正交序列对所述多个终端进行SRS检测，获取每个终端发送SRS的信道信息，包括：
    对接收到的SRS位置处待处理的频域信号进行预处理，并根据预处理后的频域信号获取每个终端发送SRS的信道信息；
    其中，所述预处理包括以下至少一种处理：
    根据公式(1)：
    Figure PCTCN2020086711-appb-100001
    对接收到的SRS位置处待处理的频域信号进行频域滤 波；其中，y(n)为SRS位置处的频域信号，w(n)为频域窗系数，
    Figure PCTCN2020086711-appb-100002
    为频域滤波后信号，n＝0,…,N为信道估计位置索引，N为信道估计的长度；
    根据公式(2)：
    Figure PCTCN2020086711-appb-100003
    对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，为SRS位置处的频域信号，维度为N×1；w为频域窗系数，维度为N×N；
    Figure PCTCN2020086711-appb-100004
    为频域滤波后信号，维度为N×1，N为信道估计的长度；
    根据公式(3)：
    Figure PCTCN2020086711-appb-100005
    对接收到的SRS位置处待处理的频域信号进行时域干扰消除；其中，y为SRS位置处的待处理的频域信号，维度为N×1；
    Figure PCTCN2020086711-appb-100006
    为时域对消变换函数，
    Figure PCTCN2020086711-appb-100007
    为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
  7. 根据权利要求1至6任一项所述的方法，其特征在于，每个终端的信道信息包括部分子带的SRS频域信道估计结果，所述根据每个终端的信道信息进行信道预测，得到信道预测结果，包括：
    根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
  8. 根据权利要求1至6任一项所述的方法，其特征在于，所述根据每个终端的信道信息进行信道预测，得到信道预测结果，包括：
    根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
  9. 根据权利要求1至8任一项所述的方法，其特征在于，所述通过RRC信令为多个终端配置基序列标识，包括：
    向每个终端发送第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
    在终端需要改变SRS基序列时，向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
  10. 根据权利要求1至8任一项所述的方法，其特征在于，所述通过RRC信令为多个终端配置基序列标识，包括：
    向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。
  11. 一种提升多用户复用性能的方法，其特征在于，应用于终端，所述方法包括：
    获取网络设备通过无线资源控制RRC信令配置的基序列标识，所述基序列标识指示的基序列与其他终端采用的基序列之间不具有正交性；
    根据所述基序列标识，发送探测参考信号SRS。
  12. 根据权利要求11所述的方法，其特征在于，所述获取网络设备通过RRC信令配置的基序列标识，包括：
    接收所述网络设备发送的第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
    接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；
    根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送；
    或者，
    接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
  13. 一种提升多用户复用性能的装置，其特征在于，包括：
    发送模块，用于通过无线资源控制RRC信令为多个终端配置基序列标识，每个终端的基 序列标识指示的基序列之间不具有正交特性；
    处理模块，用于基于准正交序列对所述多个终端进行探测参考信号SRS检测，获取每个终端发送SRS的信道信息；
    所述处理模块还用于根据每个终端的信道信息进行信道预测，得到信道预测结果。
  14. 根据权利要求13所述的装置，其特征在于，所述处理模块还用于：
    根据每个终端在不同SRS时刻的信道，获取所述终端的运动状态，所述运动状态用于指示所述终端的移动速度大小；所述运动状态包括：准静止状态、低速移动状态或者中高速移动状态；
    根据每个终端的信道预测结果以及运动状态，计算获取所述终端的权值。
  15. 根据权利要求14所述的装置，其特征在于，终端的信道预测结果包括SRS信道特征矢量或者权值，所述处理模块具体用于：
    若终端的运动状态为准静止状态，则对所述终端的SRS信道预测矢量或者权值进行类型一空域和/或时域滤波；
    若终端的运动状态为低速移动状态，则对所述终端的SRS信道预测矢量或者权值进行类型二空域和/或时域滤波；
    若终端的运动状态为中高速移动状态，则对所述终端的信道进行二阶距计算，生成所述终端的权值。
  16. 根据权利要求15所述的装置，其特征在于，所述处理模块具体用于：
    基于子空间距离及梯度下降法对所述终端的SRS预测权值进行最小均方滤波。
  17. 根据权利要求15所述的装置，其特征在于，所述处理模块具体用于：
    基于卡尔曼滤波类预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于归一化最小均方自适应滤波预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于递推最小二乘法RLS预测算法对所述终端的SRS信道预测矢量或者权值进行滤波；
    或者，
    基于自回归AR滤波的预测算法对所述终端的SRS信道预测矢量或者权值进行滤波。
  18. 根据权利要求13至17任一项所述的装置，其特征在于，所述处理模块具体用于：
    对接收到的SRS位置处待处理的频域信号进行预处理，并根据预处理后的频域信号获取每个终端发送SRS的信道信息；
    其中，所述预处理包括以下至少一种处理：
    根据公式(1)：
    Figure PCTCN2020086711-appb-100008
    对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y(n)为SRS位置处的频域信号，w(n)为频域窗系数，
    Figure PCTCN2020086711-appb-100009
    为频域滤波后信号，n＝0,…,N为信道估计位置索引，N为信道估计的长度；
    根据公式(2)：
    Figure PCTCN2020086711-appb-100010
    对接收到的SRS位置处待处理的频域信号进行频域滤波；其中，y为SRS位置处的频域信号，维度为N×1；w为频域窗系数，维度为N×N；
    Figure PCTCN2020086711-appb-100011
    为频域滤波后信号，维度为N×1，N为信道估计的长度；
    根据公式(3)：
    Figure PCTCN2020086711-appb-100012
    对接收到的SRS位置处待处理的频域信号进行时域干扰消除；其中，y为SRS位置处的待处理的频域信号，维度为N×1；
    Figure PCTCN2020086711-appb-100013
    为时域对消变换函数，
    Figure PCTCN2020086711-appb-100014
    为时域对消后的频域信号，维度维度为N×1，N为信道估计的长度。
  19. 根据权利要求13至18任一项所述的装置，其特征在于，每个终端的信道信息包括部分子带的SRS频域信道估计结果，所述处理模块还具体用于：
    根据每个终端的部分子带的SRS频域信道估计结果，基于频域外推的方式对SRS的全带宽进行预测，得到每个终端的信道预测结果。
  20. 根据权利要求13至18任一项所述的装置，其特征在于，所述处理模块还具体用于：
    根据每个终端的信道信息，基于时频空二维互校准对SRS进行信道预测，得到每个终端的信道预测结果。
  21. 根据权利要求13至20任一项所述的装置，其特征在于，所述发送模块具体用于：
    向每个终端发送第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
    在终端需要改变SRS基序列时，向所述终端发送第二RRC信令；所述第二RRC信令携带索引标识，所述索引标识用于指示所述终端采用所述至少一个基序列标识中的第一基序列标识进行SRS发送。
  22. 根据权利要求13至20任一项所述的装置，其特征在于，所述发送模块具体用于：
    向每个终端发送第一RRC信令，所述第一RRC信令携带一个基序列标识，所述基序列标识指示所有SRS基序列标识的候选合集。
  23. 一种提升多用户复用性能的装置，其特征在于，包括：
    获取模块，用于获取网络设备通过无线资源控制RRC信令配置的基序列标识，所述基序列标识指示的基序列与其他终端采用的基序列之间不具有正交性；
    发送模块，用于根据所述基序列标识，发送探测参考信号SRS。
  24. 根据权利要求23所述的装置，其特征在于，所述获取模块具体用于：
    接收所述网络设备发送的第一RRC信令，所述第一RRC信令携带至少一个基序列标识；
    接收所述网络设备发送的第二RRC信令，所述第二RRC信令携带索引标识；
    根据所述索引标识确定采用所述至少一个基序列标识中的第一基序列标识进行SRS发送；
    或者，
    接收所述网络设备发送的第一RRC信令，所述第一RRC信令一个基序列标识。
  25. 一种网络设备，其特征在于，包括：
    处理器、存储器、接收器和发送器；
    存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行权利要求1至10任一项所述的提升多用户复用性能的方法。
  26. 一种终端，其特征在于，包括：
    处理器、存储器、接收器和发送器；
    存储器用于存储程序和数据，所述处理器调用存储器存储的程序，以执行权利要求11或12所述的提升多用户复用性能的方法。
  27. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行权利要求1至10任一项所述的提升多用户复用性能的方法。
  28. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质包括程序，所述程序在被处理器执行时用于执行权利要求11或12所述的提升多用户复用性能的方法。
  29. 一种程序产品，其特征在于，该程序产品包括计算机程序，该计算机程序存储在可读存储介质中，网络设备的至少一个处理器可以从可读存储介质读取该计算机程序，至少一个处理器执行该计算机程序使得网络设备实施权利要求1至10任一项所述的提升多用户复 用性能的方法。
  30. 一种程序产品，其特征在于，该程序产品包括计算机程序，该计算机程序存储在可读存储介质中，终端的至少一个处理器可以从可读存储介质读取该计算机程序，至少一个处理器执行该计算机程序使得终端实施权利要求11或12所述的提升多用户复用性能的方法。
  31. 一种芯片，其特征在于，所述芯片可应用于网络设备，所述芯片包括：至少一个通信接口，至少一个处理器，至少一个存储器，所述通信接口、存储器和处理器通过总线互联，所述处理器调用所述存储器中存储的计算机程序，以执行权利要求1至10任一项所述的提升多用户复用性能的方法。
  32. 一种芯片，其特征在于，所述芯片可应用于终端，所述芯片包括：至少一个通信接口，至少一个处理器，至少一个存储器，所述通信接口、存储器和处理器通过总线互联，所述处理器调用所述存储器中存储的计算机程序，以执行权利要求11或12所述的提升多用户复用性能的方法。
PCT/CN2020/086711 2019-04-29 2020-04-24 提升多用户复用性能的方法、装置、设备和存储介质 WO2020221126A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP20799301.5A EP3952544A4 (en) 2019-04-29 2020-04-24 METHOD, APPARATUS AND DEVICE FOR IMPROVING THE IMPLEMENTATION OF MULTIPLEXING BETWEEN MULTIPLE USERS, AND STORAGE MEDIUM
US17/513,666 US12101205B2 (en) 2019-04-29 2021-10-28 Method and apparatus for improving multi-user multiplexing performance, device, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910355875.6 2019-04-29
CN201910355875.6A CN111865450B (zh) 2019-04-29 2019-04-29 提升多用户复用性能的方法、装置、设备和存储介质

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/513,666 Continuation US12101205B2 (en) 2019-04-29 2021-10-28 Method and apparatus for improving multi-user multiplexing performance, device, and storage medium

Publications (1)

Publication Number Publication Date
WO2020221126A1 true WO2020221126A1 (zh) 2020-11-05

Family

ID=72964940

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/086711 WO2020221126A1 (zh) 2019-04-29 2020-04-24 提升多用户复用性能的方法、装置、设备和存储介质

Country Status (4)

Country Link
US (1) US12101205B2 (zh)
EP (1) EP3952544A4 (zh)
CN (1) CN111865450B (zh)
WO (1) WO2020221126A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022186473A1 (en) 2021-03-01 2022-09-09 Samsung Electronics Co., Ltd. Method and apparatus for bandwidth-adaptive and model-order-adaptive channel prediction
EP4073991A4 (en) * 2020-01-13 2023-06-14 Samsung Electronics Co., Ltd. METHOD AND APPARATUS FOR UPLINK AND FREQUENCY OFFSET ESTIMATION AND COMPENSATION IN A WIRELESS COMMUNICATION SYSTEM

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2022226685A1 (zh) * 2021-04-25 2022-11-03 华为技术有限公司 一种信号处理方法和通信装置
CN115941394B (zh) * 2021-08-10 2024-09-17 大唐移动通信设备有限公司 一种信道预测方法、设备、装置及存储介质

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150085787A1 (en) * 2012-05-10 2015-03-26 Sharp Kabushiki Kaisha Terminal, base station, communication method, and integrated circuit
CN109150431A (zh) * 2017-06-15 2019-01-04 株式会社Ntt都科摩 一种探测参考信号传输方法、基站和用户设备

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101360084A (zh) * 2008-09-24 2009-02-04 北京创毅视讯科技有限公司 子载波间干扰消除方法、装置和信号接收方法及接收机
CN102006258A (zh) * 2009-09-02 2011-04-06 联芯科技有限公司 用于频偏校正的方法和系统
CN102771062B (zh) * 2012-04-11 2014-12-03 华为技术有限公司 一种传输模式配置方法和装置
US9300451B2 (en) * 2013-03-13 2016-03-29 Samsung Electronics Co., Ltd. Transmission of sounding reference signals for adaptively configured TDD communication systems
KR20160041043A (ko) * 2013-08-09 2016-04-15 엘지전자 주식회사 대규모 mimo 방식을 위한 안테나 결합
CN105472745B (zh) * 2014-09-04 2020-09-08 北京三星通信技术研究有限公司 一种探测导频信号资源的分配方法及设备
CN107210793B (zh) * 2015-01-16 2020-07-24 梁平 用于大规模mimo信道估计的增强型srs
JP6607101B2 (ja) * 2016-03-15 2019-11-20 富士通株式会社 無線装置用干渉抑制パラメータ演算装置、無線通信システム、無線通信システムの干渉抑制方法、基地局装置および無線端末装置
US10680866B2 (en) * 2017-03-24 2020-06-09 Huawei Technologies Co., Ltd. Sounding reference signal design
CN108111270B (zh) * 2017-06-16 2022-12-30 中兴通讯股份有限公司 导频信号发送、接收方法及装置

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150085787A1 (en) * 2012-05-10 2015-03-26 Sharp Kabushiki Kaisha Terminal, base station, communication method, and integrated circuit
CN109150431A (zh) * 2017-06-15 2019-01-04 株式会社Ntt都科摩 一种探测参考信号传输方法、基站和用户设备

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
LG ELECTRONICS: "On SRS design and related operations", 3GPP TSG RAN WG1 MEETING #88BIS R1-1704892, 7 April 2017 (2017-04-07), XP051243027, DOI: 20200708164449A *
NTT DOCOMO, INC.: "Discussions on NR SRS", 3GPP TSG RAN WG1 NR AD-HOC#2 R1-1711087, 30 June 2017 (2017-06-30), XP051300287, DOI: 20200708164306X *
NTT DOCOMO, INC.: "Discussions on NR SRS", 3GPP TSG RAN WG1 NR AD-HOC#2 R1-1711087, 30 June 2017 (2017-06-30), XP051300287, DOI: 20200708164325Y *
See also references of EP3952544A4

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4073991A4 (en) * 2020-01-13 2023-06-14 Samsung Electronics Co., Ltd. METHOD AND APPARATUS FOR UPLINK AND FREQUENCY OFFSET ESTIMATION AND COMPENSATION IN A WIRELESS COMMUNICATION SYSTEM
WO2022186473A1 (en) 2021-03-01 2022-09-09 Samsung Electronics Co., Ltd. Method and apparatus for bandwidth-adaptive and model-order-adaptive channel prediction
EP4205312A4 (en) * 2021-03-01 2024-03-06 Samsung Electronics Co., Ltd. METHOD AND DEVICE FOR BANDWIDTH ADAPTIVE AND MODEL ORDER ADAPTIVE CHANNEL PREDICTION
US12068799B2 (en) 2021-03-01 2024-08-20 Samsung Electronics Co., Ltd. Method and apparatus for bandwidth-adaptive and model-order-adaptive channel prediction

Also Published As

Publication number Publication date
US12101205B2 (en) 2024-09-24
CN111865450B (zh) 2022-09-02
US20220052881A1 (en) 2022-02-17
EP3952544A4 (en) 2022-06-08
EP3952544A1 (en) 2022-02-09
CN111865450A (zh) 2020-10-30

Similar Documents

Publication Publication Date Title
WO2020221126A1 (zh) 提升多用户复用性能的方法、装置、设备和存储介质
Decurninge et al. CSI-based outdoor localization for massive MIMO: Experiments with a learning approach
CN107483088B (zh) 大规模mimo鲁棒预编码传输方法
US10212723B2 (en) User pairing method, device and system for realizing user scheduling
CN103475397B (zh) 一种三维波束赋形的方法、通信站及移动站
CN105577587B (zh) 一种60GHz室内场景下MIMO系统信道估计方法及装置
US10243634B2 (en) Method and device for dual layer beamforming
KR101440202B1 (ko) 무선 네트워크 내의 다운링크 멀티유저 ｍｉｍｏ 송신을 위한 방법 및 장치
WO2017193907A1 (zh) 用于信道状态信息反馈的方法、基站、终端设备及系统
US20180343044A1 (en) Estimation of channel conditions
CN111512571B (zh) 用于mimo通信系统中的信号检测的方法和装置
US20160127922A1 (en) Channel estimation for a subset of resource elements of a resource block
CN102546088A (zh) 一种块对角化预编码方法及装置
CN106685501B (zh) 一种波束赋形的方法及装置
CN104022979B (zh) 一种联合稀疏信道估计方法、装置及系统
KR101418766B1 (ko) 다중―이용자 무선 시스템들에서 셀간 간섭을 감소시키기 위한 다중 안테나 방법
CN114553640B (zh) 多频段大规模mimo系统中的跨频段统计信道状态信息估计方法
CN107248878A (zh) 一种预编码的方法及装置
CN105429925B (zh) 基于快速独立分量分析的多天线ofdma信号解码方法
CN102891816B (zh) 一种基于mimo-ofdm相关信道的去耦合预测方法
CN109120321B (zh) 一种波束赋形的方法、装置、基站及计算机可读存储介质
CN107888261B (zh) 一种信道矩阵确定方法及相关设备
KR101234245B1 (ko) 다중 안테나를 사용하는 직교 주파수 분할 시스템에서의 블라인드 채널 추정을 위한 선형 프리코딩과 통신 용량을 최대화시키는 프리코딩과의 호환 방법
WO2024175168A1 (en) Method, network node, and computer program for downlink interference suppression in an access network
JP2023173064A (ja) 算出装置、方法、及びプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20799301

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020799301

Country of ref document: EP

Effective date: 20211104