Channel prediction for low-complexity channel aging countermeasures

Abstract

The communication device comprises a radio transceiver configured to communicate with other communication devices through a radio channel, and at least one processor configured to execute instructions, wherein the at least one processor a) estimates at least two channel response matrices of the radio channel based on predetermined signals received from the other communication device, b) obtains at least two first eigenvectors or singular vectors of the channel response matrices, and c) predicts at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, and the at least one second eigenvector or singular vector is used to transmit signals to the other communication device in time slots when channel estimation is not performed.

Description

[Technical Field]

[0001] This invention relates to a technique for counteracting the effects of channel aging in wireless communication channels. [Background technology]

[0002] A typical radio communication channel between a transmitter and receiver can be represented by a randomly time-varying impulse response, a detailed description of which is found in Non-Patent Literature (NPL) 1. Further, a single pulse transmitted over a multipath radio channel may be received as a sequence of pulses, each representing a single multipath component. Each multipath component undergoes a phase shift through events such as reflection, refraction, or scattering from surrounding scatterers in the transmission path, which can then be added together at the receiver to produce constructive or destructive results. This phenomenon may be called multipath fading.

[0003] Furthermore, each multipath component arrives at the receiver with a different time delay. Knowledge of instantaneous channel states can be effectively used to improve communication performance. More specifically, the capacity of a fading channel may depend on knowledge of time-varying radio channels in the transmitter and / or receiver. For example, channel information in the transmitter is extremely useful for employing performance enhancement techniques, including, but not limited to, power allocation, beamforming, or scheduling operations.

[0004] Knowledge about time-varying channels can be obtained through methods such as channel estimation. One such method involves the transmitter sending a known reference signal (e.g., a sounding reference signal (SRS)) to the receiver, allowing the receiver to calculate the channel's transfer function (more specifically, the channel's impulse response or frequency response). Furthermore, there are techniques that allow the transmitter to acquire knowledge about the channel response, such as employing time-division duplexing (TDD) or frequency-division duplexing (FDD) schemes.

[0005] As a typical form of communication, consider the transmission and reception of signals between a base station (BS) and a mobile user device (UE). In this scenario, signal transmission from the UE to the BS is called uplink (UL) communication, and signal transmission from the BS to the UE is called downlink (DL) communication. In the TDD method, the UL communication channel and the DL communication channel follow channel reciprocity, and it is known that the channel response estimated for the UL channel can be used for DL ​​transmission. Not limited to TDD, techniques to achieve channel reciprocity can also exist in the FDD method. Channel reciprocity in the FDD method can be achieved, for example, by employing frequency correction algorithms based on channel characteristics such as direction of arrival, channel covariance matrix, and channel space-time correlation, although these are not limited to TDD.

[0006] However, in many practical communication systems, a time gap can occur between the UL channel estimation time and the DL transmission time. During this time gap, the time-varying radio channel may change, and as a result, performance enhancement techniques for DL ​​transmission based on the estimated UL channel response are not immune to errors, as described in Non-Patent Literature (NPL) 2. [Prior art documents] [Non-patent literature]

[0007] [Non-Patent Document 1] A. Goldsmith, Wireless Communications. Cambridge, UK: Cambridge Univ. Press, 2005 [Non-Patent Document 2] A. Duel-Hallen, Shengquan Hu and H. Hallen, "Long-range prediction of fading signals," in IEEE Signal Processing Magazine, vol. 17, no. 3, pp. 62-75, May 2000, doi: 10.1109 / 79.841729 [Overview of the project] [Problems that the invention aims to solve]

[0008] In a typical multi-user, multi-input, multi-output (MU-MIMO) communication system, the base station (BS) may require DL channel information corresponding to each user (UE). For example, such channel information can be efficiently used to suppress interference between multiple users or multiple transmission streams. Accurately knowing the DL channel information during downlink transmission enables accurate interference cancellation operation through the calculation of correct beamforming weights. However, a time delay may occur between the time when channel estimation is performed at the BS and the time when the DL signal is transmitted. During this delay period, the time-varying radio channel response may change. Therefore, outdated channel information may be used in the calculation of beamforming weights. Using such inaccurate beamforming weights for downlink transmission can lead to a decrease in throughput for the reasons mentioned above.

[0009] In particular, the Third Generation Partnership Project (3GPP) has decided to advance research on MIMO advancements in 3GPP Release 18. In MU-MIMO system implementations, performance degradation can occur if older channel responses are used for downlink transmission from the base station (BS) to the UE when UE equipment is moving at medium to high speeds.

[0010] There is a method for predicting a channel response (e.g., a channel impulse response, a channel frequency response, or a variant including information about the channel) during signal transmission. Such prediction can be performed based on past values of the channel response measured during channel estimation. Based on the predicted channel response, either singular value decomposition (SVD) or eigenvalue decomposition (EVD) can be executed to calculate at least one of the singular vectors or eigenvectors. The singular vectors or eigenvectors obtained in this way can be used for beamforming. Therefore, when predicting the channel response in a time slot in which a beamformed signal is to be transmitted using channel prediction, it is necessary to execute SVD or EVD in those time slots. However, SVD or EVD may require heavy and complex calculations.

[0011] An exemplary object of the present disclosure is to predict at least one of the singular vectors or eigenvectors of a channel response (e.g., a channel impulse response, a channel frequency response, or a variant including channel-related information) during transmission of a beamformed signal. More specifically, during channel estimation, a channel response (impulse response or frequency response) matrix can be calculated. Such channel estimation calculations can be performed based on the received SRS or pilot signal. Thereafter, in the channel estimation time slot, either SVD or EVD can be executed to calculate the singular vectors or eigenvectors. Next, based on the singular vectors or eigenvectors calculated in the channel estimation time slot, predicted singular vectors or predicted eigenvectors in a time slot in which downlink transmission is scheduled but channel estimation is not performed can be obtained. An advantage of the present invention is that execution of the SVD or EVD operation becomes unnecessary in a time slot in which downlink transmission is scheduled but channel estimation is not performed.

Means for Solving the Problems

[0012] According to one aspect of the present invention, a communication device includes a wireless transceiver configured to communicate with other communication devices through a wireless channel, and at least one processor configured to execute instructions. The at least one processor is configured to: a) estimate at least two channel response matrices of the wireless channel based on a predetermined signal received from the other communication device; b) obtain at least two first eigenvectors or singular vectors of the channel response matrices; and c) predict at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors by a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to execute signal transmission to the other communication device in a time slot when channel estimation is not performed.

[0013] According to another aspect of the present invention, a channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel is as follows: a) estimate at least two channel response matrices of the wireless channel based on a predetermined signal received from the other communication device; b) obtain at least two first eigenvectors or singular vectors of the channel response matrices; and c) predict at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors by a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to execute signal transmission to the other communication device in a time slot when channel estimation is not performed.

[0014] As described above, according to the present invention, a second eigenvector or singular vector of a wireless channel can be predicted during signal transmission. Thereby, accurate signal transmission can be realized by using the predicted eigenvector or predicted singular vector during signal transmission.

[0015] This disclosure includes apparatus employing a plurality of steps and the interrelationships of these steps, as well as structural features, combinations of elements, and arrangements of components suitable for carrying out these steps. These details are illustrated in the following detailed description, and the scope of the disclosure is indicated in the claims. In addition to the purposes stated above, other obvious and apparent advantages of this disclosure will become apparent from the detailed description and drawings. [Brief explanation of the drawing]

[0016] [Figure 1] Figure 1 is a diagram illustrating the operation of uplink transmission from UE to BS in a wireless communication system to which an exemplary embodiment of the present invention can be applied. [Figure 2] Figure 2 is a diagram illustrating the operation of downlink transmission from BS to UE in a wireless communication system to which an exemplary embodiment of the present invention can be applied. [Figure 3] Figure 3 illustrates an uplink transmission from a UE to a BS under a typical multipath radio propagation environment in a wireless communication system to which exemplary embodiments of the present invention can be applied. [Figure 4] Figure 4 is a schematic diagram illustrating the challenges of calculating downlink beamforming weights that cause channel aging in wireless communication systems. [Figure 5] Figure 5 is a typical plot of average user throughput against uplink SRS transmission interval, illustrating the performance degradation due to wireless channel aging. [Figure 6] Figure 6 is a schematic diagram illustrating the challenges of downlink beamforming weight calculation, which result in increasing complexity. [Figure 7] Figure 7 is a schematic diagram illustrating an example of a proposed solution for eigenvector prediction. [Figure 8] Figure 8 is a graph illustrating the multipath resolution achieved by the proposed solution, showing examples of channel gain at different angles for each multipath component. [Figure 9]Figure 9 is a graph illustrating the multipath resolution achieved by the proposed solution, showing examples of channel gain at different delays for each multipath component. [Figure 10] Figure 10 is a graph illustrating an example of channel gain at different Doppler frequencies for each multipath component, illustrating the multipath resolution provided by the proposed solution. [Figure 11] Figure 11 is a schematic diagram illustrating the functional configuration of a BS device in an exemplary embodiment of the present invention. [Figure 12] Figure 12 is a schematic diagram illustrating the functional configuration of a UE terminal capable of communicating with a BS device according to an exemplary embodiment of the present invention. [Figure 13] Figure 13 is a sequence diagram illustrating a series of frame transmissions and operations of a wireless communication system according to an exemplary embodiment of the present invention. [Figure 14] Figure 14 is a flowchart illustrating the prediction operation of eigenvectors or singular vectors in the subcarrier domain and antenna domain according to an exemplary embodiment of the present invention. [Modes for carrying out the invention]

[0017] In the following, the term “exemplary” means “an example, case, or illustration.” Examples described as “exemplary” in this specification are not construed as being preferable or advantageous to other examples.

[0018] 1. Overview of Exemplary Embodiments

[0019] The technical challenges of the background technology described above can be solved by predicting at least one of the singular vectors or eigenvectors of the channel response matrix of a time-varying radio channel at the time of signal transmission, based on singular vectors or eigenvectors calculated during one or more past channel estimations. The prediction of singular vectors or eigenvectors can be performed by extrapolation. More specifically, it is possible to obtain singular vectors or eigenvectors in time slots where channel estimation has not been performed but beamformed signal transmission is desired.

[0020] Therefore, using the predicted eigenvectors of the wireless channel, it is possible to perform signal transmission based on accurate beamforming at a time when channel estimation is not performed. This makes it possible to compensate for the performance degradation (particularly with respect to metrics such as user throughput) that occurs when calculating beamforming weights using old or obsolete channel information.

[0021] In the extrapolation operations used in some embodiments of the proposed method, techniques including, but not limited to, linear extrapolation, linear regression, least squares estimation, nonlinear regression, polynomial regression, spline regression, and curve fitting may be employed. Where the term “extrapolation” appears in this disclosure, it is to be interpreted as including methods other than those listed above. Below, an overview of a channel prediction method according to exemplary embodiments will be described with reference to Figures 1 to 14.

[0022] 1.1) System Configuration

[0023] As illustrated in Figures 1 and 2, a wireless communication system consists of multiple communication devices, where specific communication devices, such as a base station (BS) or access point, can communicate with other communication devices. For simplicity, let us assume that BS device 100 can communicate with multiple UE terminals 200, including two UE terminals UE1 and UE2. UE terminals UE1 and UE2 are located within a wireless coverage area (cell) 100A formed by BS 100, each UE terminal can perform uplink (UL) transmission, and BS device 100 can perform downlink (DL) transmission using beamforming.

[0024] Figure 3 illustrates a typical multipath propagation environment in wireless communication. The uplink signal transmitted from UE terminal 200 reaches BS device 100 through multiple paths (five paths are shown in Figure 3), each path representing a copy of the signal with some attenuation and delay. BS device 100 receives the sum of the signal copies from all paths.

[0025] Beamforming technology is employed by multi-antenna transmitters to impart directionality to transmissions, thereby enabling spatial multiplexing of multiple signals. Such a mechanism can be used in multi-user MIMO systems to allow a multi-antenna BS (BS device 100) to simultaneously transmit multiple signals destined for different users (UE terminals 200). Beamforming can be implemented using analog or digital methods. In analog beamforming, the magnitude and phase corresponding to each transmitting antenna can be varied by using different amplifiers and phase shifters for the same analog signal at radio frequencies. This enables power variation and beam steering.

[0026] On the other hand, in digital beamforming, different digital baseband signals can be constructed for each transmitting antenna by multiplying them with different weighting coefficients. In digital beamforming, the transmitter of the BS device 100 requires channel information between itself and the receivers of each UE terminal 200 in order to design effective beamforming weights. Techniques such as channel estimation may be employed to obtain channel information. In a typical channel estimation technique, a known reference signal is transmitted from the UE terminal 200 to the BS device 100. The BS device 100 can use this known reference signal to calculate a channel impulse response or frequency response.

[0027] 1.2) Performance degradation

[0028] Transmitting a known reference signal excessively frequently (for example, more than a predetermined number of times within a certain time interval) can increase the overhead associated with channel estimation, potentially reducing the opportunity for actual data communication. However, if the time interval for channel estimation becomes too long, the state of the time-varying radio channel may change during that time, which could result in outdated channel estimation information at the time of the next transmission. This situation is illustrated in Figure 4.

[0029] In Figure 4, the UE terminal 200 transmits uplink SRS (UL-SRS) to the BS device 100 at 40ms (millisecond) intervals. Each time the BS device 100 receives a UL-SRS, it calculates digital beamforming (BF) weights based on channel estimation. The radio channel between the BS device 100 and the UE terminal 200 may change over time due to the movement of the UE terminal 200 or other factors contributing to radio channel fluctuations. If the state of the radio channel between the BS device 100 and the UE terminal 200 changes within a 40ms period from t=0 (when the UE transmits UL-SRS1 and the BS receives it) to t=40ms (when the UE transmits UL-SRS2 and the BS receives it), the BS device 100 may not have knowledge of the latest channel response. In such a situation, the BS device 100 may need to reuse the beamforming weights calculated in the previous channel estimation event until the next channel estimation event. For example, BS device 100 might calculate channel estimation and beamforming weights at t=0 and continue using the same beamforming weights for downlink transmission until the next channel estimation event occurs at t=40ms. Therefore, BS device 100 using outdated channel estimation results and digital beamforming weights can lead to performance degradation due to inefficient interference cancellation between spatially multiplexed streams and interference between signals destined for different UE terminals.

[0030] Figure 5 illustrates an exemplary scenario of performance degradation due to channel aging. Figure 5 plots the average user throughput (bits / sec / Hz) against the time interval (milliseconds) between two consecutive uplink SRS transmissions. Assume that UE terminal 200 is moving at 3 km / h. If the channel is fully tracked by BS device 100 (as shown in reference no. 301), no degradation in average user throughput is observed. This means that BS device 100 knows the channel response perfectly accurately for all time slots. This occurs because BS device 100 can use the exact values ​​of the channel impulse response or channel frequency response to calculate the precise beamforming weights for downlink transmissions, and as a result, can efficiently cancel interference between spatially multiplexed streams and users.

[0031] However, if perfect channel tracking is impossible, the time-varying radio channel changes between the time the beamforming weights are calculated and the time the downlink transmission takes place. Therefore, due to channel aging and inaccurate beamforming weights, the average user throughput can be found to be reduced, as shown in reference number 302.

[0032] Figure 6 illustrates the challenges of channel prediction. According to this channel prediction, the channel response matrix is ​​predicted in time slots where channel estimation is not performed. Subsequently, SVD or EVD calculations are performed to obtain eigenvectors, which are then used for beamforming. For example, BS device 100 receives the first uplink SRS (SRS 1) in time slot = 0 and the second SRS (SRS 2) in time slot = 40. Therefore, channel estimation can be performed in time slots = 0 and 40, and the corresponding estimated channel response matrices can be obtained. Now, consider the case where the next channel estimation is performed when the SRS is received in time slot = 80 (not shown in Figure 6). In this case, if the time slot is greater than 40 and less than 80, predicting the channel response matrix can mitigate performance degradation due to channel aging. More specifically, based on the two past channel response matrices obtained from the channel estimation operations in time slots = 0 and 40, the channel response matrices in time slots = 41, 42, ..., 79 can be predicted. For each time slot = 41, 42, ..., 79, at least one of the SVD or EVD operations is performed on the predicted channel matrix. This enables pre-coded and beamformed transmission in time slots = 41, 42, ..., 79. Therefore, the channel prediction method shown in the example in Figure 6 may require SVD or EVD operations in all time slots where channel prediction is performed (specifically, time slots = 41, 42, ..., 79). However, SVD or EVD operations are highly complex. Therefore, it is desirable to reduce the number of SVD or EVD operations.

[0033] 1.3) Proposed solution

[0034] The above problem can be solved by an exemplary embodiment of the present invention. Specifically, the above problem can be solved by predicting at least one eigenvector of the channel response matrix at the time when channel prediction is to be performed. This is illustrated by an example in Figure 7. More specifically, referring to Figure 7, it is proposed to predict an eigenvector instead of predicting the channel response matrix in all time slots of downlink transmission. This proposal eliminates the need to perform complex SVD or EVD calculations in all time slots of downlink transmission where channel estimation is not performed. In the proposed method, SVD or EVD calculations are performed only in time slots where channel estimation is performed. This significantly reduces the total number of SVD or EVD calculations between two channel estimation time points, thereby reducing the complexity of the system.

[0035] As illustrated in Figure 7, the UE terminal 200 transmits an uplink SRS as a reference signal at predetermined intervals. The reference signal may be periodic (e.g., 40 ms) or aperiodic. Each time the BS device 100 receives an uplink SRS, it calculates the channel response matrix between the UE terminal 200 and the BS device 100 by channel estimation. For example, the BS device 100 receives uplink SRS in time slot 0 and time slot 40. The BS device 100 calculates the estimated channel matrix for time slot 0 and time slot 40, respectively. Then, it performs at least one of SVD or EVD operations on the estimated channel response matrix to obtain the eigenvectors for time slot 0 and time slot 40. Next, based on the eigenvectors calculated for time slot 0 and time slot 40, the eigenvectors for time slots 41, 42, ..., 79 are predicted. This enables precoded and beamformed transmission in time slots 41, 42, ..., 79. Therefore, in the channel prediction method shown in Figure 7, SVD or EVD calculations are only required in time slots where uplink SRS is received and channel estimation is performed (time slots 0 and 40 in this example). In time slots where channel estimation is not available (time slots 41 to 79 in this example), SVD or EVD calculations are not performed. This reduces the overall complexity of the system.

[0036] Eigenvector prediction is performed by at least one extrapolation method, including linear extrapolation, linear regression, least squares estimation, nonlinear regression, polynomial regression, spline regression, or curve fitting. In some embodiments, simple linear extrapolation is employed, and future values ​​may be predicted based on two or more past instances of the eigenvector elements. In some embodiments, eigenvector prediction may be performed by employing machine learning techniques. Machine learning techniques include, but are not limited to, long-short-term memory (LSTM) models, Transformer models, or reinforcement learning (online learning). When the time variation of eigenvectors or singular vectors is treated as a time series, LSTM models can be used to predict eigenvectors. LSTM is a recurrent neural network (RNN) that can learn long-short-term dependencies in a time series and can be used to predict eigenvectors by treating the eigenvectors obtained at the channel estimation point as a time series. Furthermore, Transformer models can also be used for time series prediction of eigenvectors or singular vectors. Time series forecasting of eigenvectors or singular vectors can be performed not by batch learning, but by online reinforcement learning, which continuously updates the best predictive model based on available data.

[0037] Hereinafter, without loss of generality, the channel response obtained by channel estimation based on a reference signal may be referred to as the estimated channel response matrix. The eigenvectors or singular vectors obtained from the estimated channel response matrix may be referred to as the first eigenvectors or singular vectors. On the other hand, eigenvectors or singular vectors obtained by prediction based on previously obtained eigenvectors or singular vectors may be referred to as predicted eigenvectors or singular vectors. In general, the input to the prediction process may be referred to as the first eigenvector or singular vector, and the output to the prediction process may be referred to as the second eigenvector or singular vector. Prediction of eigenvectors or singular vectors may include predicting the amplitude and phase of the predicted eigenvectors or singular vectors, or predicting the real and imaginary parts of the predicted eigenvectors or singular vectors.

[0038] <Multipath propagation environment>

[0039] As illustrated in Figure 3, UE terminal 200 transmits UL-SRS in a multipath propagation environment. Without loss of generality, we consider UL-SRS as a reference signal. UL-SRS propagates along separate paths, indicated by reference numbers 511-515, reflecting off objects such as buildings 501-504. Therefore, the UL-SRS reaching BS device 100 may follow different paths and thus arrive at BS device 100 at different times and angles. Furthermore, UE terminal 200 may be moving towards or away from BS device 100, in which case the Doppler effect must be considered.

[0040] To predict time-varying radio channels in a multipath propagation environment, it is useful to first decompose the received superimposed signal into its multipath components. Following the multipath decomposition, the amplitude and phase of each multipath component are calculated. Such calculations are performed during the channel estimation event, based on a known reference signal received.

[0041] Efficient separation of multipath components is crucial for accurate channel prediction. For example, multipath components can be separated based on angular profiles, delay profiles, or Doppler profiles.

[0042] Refer to Figure 8 to illustrate the separation of multipath components based on angle. In some examples, the angle domain can be called the beamspace domain. The five multipath components 550, 551, 552, 553, and 554 have different angles from each other. Therefore, in the angle domain, the channel gains of these components can be obtained separately. Similarly, the phases of each of the five multipath components can also be obtained individually.

[0043] Furthermore, referring to Figure 9, the separation of multipath components based on delay is described. The five multipath components 570, 571, 572, 573, and 574 have different delays from each other. Therefore, the channel gains of these components can be obtained separately in the delay domain. Similarly, the phase of each of the five multipath components can also be obtained individually.

[0044] According to embodiments of this disclosure, it is possible to separate multipath components based on Doppler frequencies. Such multipath separation in the Doppler domain is illustrated using the example in Figure 10. The five multipath components 650, 651, 652, 653, and 654 have different Doppler frequencies. Therefore, their channel gains can be obtained separately in the Doppler domain. Similarly, the phases of each of the five multipath components can also be obtained individually.

[0045] 2. Exemplary Embodiments

[0046] The implementation of the above prediction method will be explained in detail below.

[0047] 2.1) System Configuration As shown in Figure 11, the BS device 100 includes an array antenna consisting of M antennas ANT.(1)-ANT.(M), where M is an integer greater than 1. Each antenna ANT.(1)-ANT.(M) is connected to a radio transceiver TR(1)-TR(M). Each radio transceiver includes an RF (radio frequency) front end 101, a Fast Fourier Transform (FFT) unit 102, and an Inverse Fast Fourier Transform (IFFT) unit 103. The RF front end 101 receives the RF signal received from the corresponding antenna and outputs the received data sequence to the FFT unit 102. The RF front end 101 receives the transmitted data sequence from the IFFT unit 103 and outputs an RF transmit signal to the corresponding antenna. The FFT unit 102 decomposes the received data sequence into frequency components. The IFFT unit 103 constructs the transmitted data sequence from the frequency components.

[0048] The BS device 100 further includes a channel estimator 104, an eigenvector predictor 105, and a precoder 106. The channel estimator 104 receives the frequency components of the UL-SRS from the FFT section 102 of each radio transceiver and outputs the channel estimation signal to the eigenvector predictor 105. The eigenvector predictor 105 predicts the eigenvectors of the channel response matrix at a time when channel estimation has not been performed. This prediction can be made by the eigenvector predictor 105 based on two or more past eigenvectors, as already mentioned (see the example in Figure 7). The eigenvector predictor 105 outputs the predicted eigenvectors to the precoder 106.

[0049] The BS device 100 further includes a scheduler 107 and multiple data processing units. The multiple data processing units implement the functions of a data generator 108, a forward error correction (FEC) unit 109, a modulator 110, and a resource mapper 111, respectively. The scheduler 107 determines which users are scheduled to perform DL transmission in a given time slot. The data generator 108 generates the transmission data, which is then processed by the FEC unit 109. The modulator 110 modulates the output of the FEC unit 109 and outputs the modulated transmission data to the resource mapper 111. The resource mapper 111 performs resource mapping of the transmission data and outputs the frequency components to the IFFT 103 of each transceiver through the precoder 106. The precoder 106 performs precoding according to the predicted eigenvectors received from the eigenvector predictor 105. As already mentioned, eigenvector prediction is performed for future time slots, and the predicted eigenvectors are stored in memory. Alternatively, eigenvector prediction can be performed in real time for each time slot.

[0050] In Figure 11, the functions indicated by reference numerals 104-111 can be implemented by a processor or central processing unit (CPU) that executes programs stored in program memory. These programs include eigenvector prediction programs, which can implement the functions of the eigenvector predictor 105.

[0051] As illustrated in Figure 12, the UE terminal 200 includes a processor 201, program memory 202, a communication interface 203, an input / output device 204, and a battery 205. The processor 201 controls the operation of the UE, including UL-SRS transmission, by executing a program stored in the program memory 202. UL-SRS transmission is performed in response to signaling from the BS device 100. When the UE terminal 200 is within the radio coverage of a network device such as the BS device 100, the communication interface 203 can connect to the network device via a radio channel. Furthermore, the power required for all operations in the UE terminal (such as the execution of the processor 201, transmission / reception of signals using the communication interface 203, and other power-driven operations) can be supplied from the battery 205.

[0052] Referring to Figure 13, an exemplary frame sequence diagram is shown between a network device (e.g., BS device 100) and a UE terminal 200. In a particular example, BS device 100 transmits a Radio Resource Control (RRC) signal to UE terminal 200, as shown in reference number 701. UE terminal 200 transmits a UL-SRS to BS device 100, as shown in reference number 702. Upon detection of the UL-SRS, BS device 100 can obtain a channel response between BS device 100 and UE terminal 200 (e.g., channel impulse response, channel frequency response, or other information related to channel state) based on the UL-SRS described in step 703. The channel response is represented in matrix form. At least one SVD or EVD operation can be performed on the channel response matrix to obtain at least one singular vector or eigenvector. Based on one or more eigenvectors obtained from the UL-SRS, at point T where channel estimation has not been performed... d The eigenvectors at time T can be predicted using the extrapolation operation described in step 704. Furthermore, based on the predicted eigenvectors, time T dThe downlink beamforming weights can be calculated as shown in step 705. Finally, using the downlink beamforming weights, the downlink signal can be transmitted from the BS device 100 to the UE terminal 200, as shown in reference number 706.

[0053] 2.2) Description of intrinsic mode transmission with full channel tracking using SVD or EVD

[0054] The following shows examples of SVD and EVD calculations and describes a typical eigenmode transmission with complete channel tracking. Here, complete channel tracking means that the transmitter has complete knowledge of the time-varying radio channel (or equivalently, the channel response matrix between the transmitter and receiver) with respect to the receiver in each time slot that the transmitter intends to transmit to the receiver. The estimated channel response matrix in BS device 200 is N R ×N T Assume a matrix H of size. The EVD operation is possible for square matrices. On the other hand, the SVD operation is a more general operation and is possible for general matrices, including non-square matrices. The SVD of matrix H can be expressed as follows:

[0055]

number

[0056]

number

[0057] Here,

number

[0058]

Number

[0059]

Number

[0060] Here, an example of the SVD operation using a non-square matrix is shown. A (2×3) channel response matrix is defined as follows.

[0061]

Number

[0062]

number

[0063]

number

[0064]

number

[0065] The right singular vectors that are columns of matrix V are H H It can be obtained from the orthonormal set of eigenvectors of H. First, H H The eigenvalues ​​of H are found to be 25, 9, and 0. The H corresponding to eigenvalue e=25 is... H The eigenvectors of H are the matrix H H This can be obtained by finding the unit norm vector in the H-25I kernel. Matrix H H H-25I can be expressed as follows:

[0066]

number

[0067]

number

[0068]

number

[0069]

number

[0070]

number

[0071]

number

[0072]

number

[0073]

number

[0074]

number

[0075] Here, we show an example of EVD operation using a square matrix. Consider the H matrix defined as follows.

[0076]

number

[0077]

number

[0078]

number

[0079]

number

[0080]

number

[0081] 2.3) Precoding on the transmitter Here is an example of a precoding operation performed by BS device 100: x=[x1, x2, ..., x N ] is used as the transmission symbol vector. In the precoding of the transmitter of BS device 100, vector x is multiplied by matrix V to form another vector

[0082]

number

[0083]

number

[0084]

number

[0085]

number

[0086]

number

[0087] 2.4) Separation or parallelization of operations in the receiver Here, we illustrate the advantages of precoding. By performing precoding in the transmitter as described above, N TN arriving at the receiver from different transmitting antennas T It becomes possible to separate individual different transmission streams. For example, if the received signal is

[0088]

number

[0089]

number

[0090]

number

[0091]

number

[0092]

number

[0093] In some examples,

[0094]

number

[0095]

number

[0096]

number

[0097] 2.5) Performance degradation of eigenmode transmission due to old values ​​of the V matrix Here, we will explain the performance degradation that occurs in eigenmode transmission when complete channel tracking is not possible. Complete channel tracking means that the transmitter knows the exact value of the channel matrix H for each transmission time slot, which allows for the accurate calculation of the V matrix for precoding. However, due to the overhead associated with SRS transmission, channel estimation may not be possible for all time slots. Therefore, the H matrix estimated at a previous channel estimation time may be used until the next channel estimation time. More specifically, the precoding matrix V calculated from the old H matrix may be used until the next channel estimation time. old We will represent it as follows. Also, if complete channel tracking is possible, the exact precoding matrix in a given time slot is V perf This is how it will be expressed.

[0098] As follows,

[0099]

number

[0100]

number

[0101] Alternatively, the calculation can be performed as follows:

[0102]

number

[0103]

number

[0104]

number

[0105] An effective solution to the above problem is matrix V in each transmission time slot. perf The goal is to know the current channel response matrix. Therefore, a channel prediction method can be employed that uses past channel response matrix values ​​to predict the latest channel response matrix. However, deriving the precoding matrix from the predicted channel response matrix requires SVD or EVD operations, which are highly complex. Therefore, it is necessary to obtain the latest and updated precoding matrix for each transmission time slot with low complexity.

[0106] 2.6) Prediction of eigenvectors An exemplary solution to the aforementioned problem is to predict the precoding matrix for each transmission time slot. This method may eliminate the need to perform SVD or EVD operations for each time slot. This method is illustrated with reference to Figure 14.

[0107] In Figure 14, at time T1 (reference number 3100), the uplink SRS is received by BS device 100, and subsequently, BS device 100 estimates the channel response matrix in the frequency (subcarrier) and antenna domain (reference number 3101). As shown in reference number 3102, an EVD (or SVD) operation is performed on the estimated channel response matrix. A similar process is performed at time T2 (shown in reference number 3110), where T2 > T1. More specifically, the channel response matrix is ​​estimated in the frequency and antenna domain (reference number 3111), and EVD or SVD is performed on the channel response matrix estimated in reference number 3111 (reference number 3112). Based on the eigenvectors or singular vectors or precoding matrices obtained in reference numbers 3102 and 3112, predictions are made to obtain eigenvectors or singular vectors or precoding matrices at time points T2 and beyond (reference number 3120).

[0108] In some exemplary embodiments of the present invention, the prediction may be performed based on two or more past eigenvectors (or singular vectors or precoding matrices). More specifically, the channel estimation may be performed in P (P≧2) different time slots, i.e., T1···TP.

[0109] In some embodiments, eigenvector prediction may be performed in a transformed domain (a domain different from the antenna-subcarrier domain). For example, at least one of the antenna, subcarrier, delay, beamspace, time, and Doppler domains, or a combination of these domains, may be used. For example, eigenvector prediction may be performed in a delay-beamspace domain or a delay-Doppler-beamspace domain. Once eigenvector prediction is performed in the transformed domain, the predicted eigenvector matrix is ​​transformed into an antenna-subcarrier domain so that it can be applied to beamforming transmissions.

[0110] In some exemplary embodiments, domain transformation can be achieved by employing at least one of the Discrete Fourier Transform (DFT) or the Inverse Discrete Fourier Transform (IDFT), as shown below: a) From antenna domain to beamspace domain: DFT along the antenna index; b) From subcarrier domain to delayed domain: IDFT along the subcarrier index; c) From the time domain to the Doppler domain: DFT along the time index; d) From beamspace domain to antenna domain: IDFT along the beam index; e) From delayed domain to subcarrier domain: DFT along delayed tap index; and f) From Doppler domain to time domain: IDFT along the Doppler tap index. Furthermore, in some embodiments, the Fast Fourier Transform (FFT) or Inverse Fast Fourier Transform (IFFT) algorithm can be used to implement the DFT and IDFT operations, respectively.

[0111] In some exemplary embodiments, the prediction calculation (reference number 3120) may be performed by linear extrapolation as described herein. Assume that the elements of the eigenvector obtained at time T1 in step 3102 of Figure 14 are a1 + jb1, where a1 is the real part, b1 is the imaginary part, and j is the square root of -1. Furthermore, assume that the elements of the eigenvector obtained at time T2 in step 3112 of Figure 14 are a2 + jb2. Then, the predicted value of the real part at time T3 (T3 > T2) can be calculated by the following equation.

[0112]

number

[0113]

number

[0114] In some exemplary embodiments, for example, in embodiments where domain transformation is not used, prediction or extrapolation in complex number form may be preferred as described above. More specifically, if the elements of the eigenvectors are complex numbers, the real and imaginary parts of the complex numbers can be extrapolated separately. Here, "domain transformation" means a transformation from a first domain to a second domain, such as from an antenna domain to a beamspace domain, from a frequency (subcarrier) domain to a delay domain, or from a time domain to a Doppler domain. It has been confirmed that extrapolation in complex number form can yield better performance (e.g., higher throughput) when domain transformation is not used.

[0115] Furthermore, in some embodiments, the values ​​of a1+jb1 and a2+jb2 are first converted into magnitude and phase forms. More specifically, the magnitude of a1+jb1 is,

[0116]

number

[0117]

number

[0118]

number

[0119]

number

[0120] As described above, in some exemplary embodiments in which eigenvector or singular vector prediction is performed in the form of magnitude and phase, it is possible to impose a constraint so that the predicted magnitude is always non-negative.

[0121] In some exemplary embodiments, for example, in embodiments employing domain transformation, it is desirable to perform prediction or extrapolation in the form of magnitude and phase as described above. More specifically, if the elements of the eigenvectors are complex numbers, the real and imaginary parts of the complex numbers can be used to calculate the magnitude and phase values. Furthermore, instead of the real and imaginary parts of the complex numbers, magnitude and phase can be used for extrapolation. Here, “domain transformation” means a transformation from a first domain to a second domain, including, for example, a transformation from an antenna domain to a beamspace domain, a frequency (subcarrier) domain to a delay domain, or a time domain to a Doppler domain. When domain transformation is used, better performance (e.g., higher throughput) can be obtained by performing extrapolation in magnitude-phase form.

[0122] In some embodiments, the predicted eigenvectors can be optimized to reduce prediction errors or improve performance. For example, the predicted eigenvectors can be optimized to have a unit norm. By performing a normalization operation on the predicted eigenvectors, their norm can be made equal to 1.

[0123] In one exemplary embodiment, a method can be employed to continuously track the time evolution of the same element of the same eigenvector across consecutive channel estimation time points. More specifically, in order to accurately predict the elements of the eigenvector corresponding to the same eigenvalue, the sort order (from the largest eigenvalue to the smallest eigenvalue in the S matrix obtained from the EVD operation) must not affect the elements being tracked. For example, consider the case of tracking and predicting the time evolution of the elements of the eigenvector corresponding to the largest eigenvalue. A problem may arise if the eigenvalue being tracked is no longer the largest eigenvalue at a certain channel estimation time point. A change in the order of the eigenvalues ​​may result in tracking different eigenvalues. To solve this problem, it is necessary to employ a method that consistently tracks the same eigenvalue even if the order of the eigenvalues ​​changes between time points before and after channel estimation.

[0124] In some exemplary embodiments, a method can be employed to compute eigenvectors in a consistent manner at all points in the EVD (or SVD) operation. Since eigenvectors are generally not unique, any vector satisfying the properties of an eigenvector may be eligible to be a valid eigenvector. For example, a scalar multiplier of an eigenvector may also be recognized as a valid eigenvector. Therefore, for accurate prediction, it is important that eigenvectors are computed in a consistent manner and by consistent formulas. Specifically, if a scalar multiplier occurs in an eigenvector at some stage of the EVD operation, such an effect must be reversed before using it in the prediction operation.

[0125] In some exemplary embodiments, prediction calculations can be omitted for eigenvectors (or singular vectors) corresponding to very small eigenvalues ​​(or singular values). More specifically, prediction calculations can be omitted if the eigenvalue (or singular value) is below a pre-defined threshold. 2.7) Prediction in the antenna-subcarrier domain In a multi-carrier communication system such as orthogonal frequency division multiplexing, in which a high-speed broadband channel is divided into multiple low-speed subchannels (or subcarriers), the BS device 100 can calculate the channel's frequency response from the SRS received by each antenna.

[0126] More specifically, in the channel estimation method, the amplitude and phase of the channel corresponding to each subcarrier can be calculated for each antenna of the BS device 100. The estimated channel response can be used to obtain at least one eigenvector or singular vector. EVD or SVD calculations are used for this purpose, but other approaches are not excluded. By obtaining eigenvectors or singular vectors at two or more time points, it becomes possible to predict the values ​​of eigenvectors or singular vectors at times when channel estimation is not performed. More specifically, methods including but not limited to linear extrapolation, nonlinear extrapolation, curve fitting, linear regression, nonlinear regression, and machine learning-based methods can be employed in the prediction calculation. The predicted values ​​of eigenvectors or singular vectors can be used for downlink beamforming.

[0127] 2.8) Prediction in the transformation domain

[0128] In a multipath propagation environment, multipath components can be separated based on differences in delay (as shown in Figure 9), differences in angle (as shown in Figure 8), or differences in Doppler frequency (as shown in Figure 10). More specifically, Figure 9 shows multipaths 570–574 with different delays, and similarly, Figure 8 shows multipaths 550–554 with different angles.

[0129] In some exemplary embodiments, the BS device 100 first performs channel estimation to obtain an estimated channel response matrix in the antenna-frequency domain for at least two different time slots (e.g., time slots 0 and 40 shown in Figure 7). Next, the estimated channel response matrix is ​​transformed to other domains by applying one or more first transformation operations. As an example of a transformation operation, an inverse discrete Fourier transform (IDFT) along the frequency subcarrier index is performed to transform the estimated channel response matrix to the antenna-delay domain. Another first transformation operation, namely applying a discrete Fourier transform (DFT) along the antenna index to the transformed channel response matrix in the antenna-delay domain, yields an estimated channel response matrix in the angle-delay domain (also called the beamspace-delay domain). Next, from the transformed channel response matrix in the beamspace-delay domain, an eigenvector matrix or singular vector matrix can be obtained using, for example, EVD, SVD, or an equivalent operation. Thus, two or more eigenvector matrices or singular vector matrices obtained from two or more channel response matrices of the transformation domain at two different time points (for example, two different time points, time slots 0 and 40, as shown in Figure 7) can be used to predict another eigenvector or singular vector in the transformation domain (in this example, the beamspace-delay domain). The predicted values ​​of the eigenvector or singular vector are obtained by applying one or more second transformation operations to obtain the values ​​in the antenna-frequency domain. In this particular example, if the second transformation is an IDFT along the angular index, then a predicted eigenvector or singular vector matrix in the antenna-delay domain can be obtained. If another second transformation is a DFT along the delay tap index, then a predicted eigenvector or singular vector in the antenna-frequency domain can be obtained.

[0130] The prediction of eigenvectors or singular vectors is not limited to the beamspace-delay domain; the transformation domain in which the prediction is performed may be a combination of the delay domain, beamspace domain, and Doppler domain. For example, the prediction of an eigenvector matrix or singular vector matrix is ​​performed in the delay-beamspace-Doppler domain. The predicted values ​​are then transformed into the antenna-frequency domain. The transformation from the time domain to the Doppler domain can be achieved using a DFT over time slot indices. Similarly, the transformation from the Doppler domain to the time domain can be achieved using an IDFT over Doppler tap indices.

[0131] In some embodiments, the BS device 100 first performs channel estimation and obtains estimated channel response matrices in the antenna-frequency domain for at least two different time slots, e.g., time slots 0 and 40 shown in Figure 7. Next, a first eigenvector matrix or singular vector matrix is ​​obtained from the estimated channel response matrices in at least two different time slots in the antenna-frequency domain, e.g., time slots 0 and 40. The first eigenvector matrix or singular vector matrix in the two different time slots can be obtained using operations such as EVD or SVD, or equivalent methods. In some embodiments, the eigenvectors or singular vectors may be obtained in other ways that reduce complexity. The at least two first eigenvector matrices or singular vector matrices obtained in the antenna-frequency domain are applied to one or more first transformation operations to transform them into another domain. As an example of a transformation operation, an inverse discrete Fourier transform (IDFT) is performed along the frequency-subcarrier index to transform the eigenvector matrix or singular vector matrix into the antenna-delay domain. Another first transformation operation, using a Discrete Fourier Transform (DFT) along the antenna index on the transformed eigenvector matrix or singular vector matrix in the antenna-delay domain, yields the eigenvector matrix or singular vector matrix in the angle-delay domain (equivalently also called the beamspace-delay domain). Thus, using two or more eigenvector matrices or singular vector matrices obtained at two different time points (e.g., time slots 0 and 40 shown in Figure 7) in the transformed domain (the beamspace-delay domain in this example), it is possible to predict other eigenvector matrices or singular vector matrices in the transformed domain (the beamspace-delay domain in this example). Therefore, the predicted values ​​of the eigenvector matrices or singular vector matrices can be applied to one or more second transformation operations to obtain values ​​in the antenna-frequency domain. In this particular example, if the second transformation is an IDFT along the angle index, it is possible to obtain the predicted eigenvector matrix or singular vector matrix in the antenna-delay domain.If the other second transformation is a DFT along the delay tap index, a predicted eigenvector matrix or singular vector matrix can be obtained from the antenna-delay domain to the antenna-frequency domain. The predicted eigenvector matrix or singular vector matrix obtained in the antenna-frequency domain is used for beamforming signal transmission.

[0132] In another exemplary embodiment of the present invention, a controller can be used to determine the optimal prediction method. For example, a metric can be used to calculate the difference between the predicted eigenvectors and the actual eigenvectors in the time slot in which channel estimation is performed. Such metrics include, but are not limited to, the mean squared error. The actual eigenvectors can be obtained from the estimated channel response matrix in the time slot of channel estimation. Based on the value of the metric, the controller selects a prediction method such that the metric is optimized (e.g., the mean squared error is minimized). Furthermore, if the value of the metric exceeds a predetermined threshold, the controller may decide not to perform a prediction.

[0133] The application software described herein is a computer program executed by a device and may be stored on one or more computer-readable media. The steps identified herein may also be performed using one or more general-purpose computers or dedicated computers and / or computer systems, including but not limited to networked systems. The order of the steps described herein may be changed, combined into composite steps, or divided into substeps as appropriate to implement the functions described herein.

[0134] 3.Additional information In this disclosure, user equipment (or "UE," "mobile station," "mobile device," or "wireless device") is an entity connected to a network via a wireless interface.

[0135] This disclosure is not limited to dedicated communication devices, but can be applied to any device having communication functions as described in the following paragraphs.

[0136] The terms "User Equipment" or "UE," "Mobile Station," "Mobile Device," and "Wireless Device" (as used in 3GPP) are generally intended to be synonymous with each other and include standalone mobile stations such as terminals, mobile phones, smartphones, tablets, cellular IoT devices, and IoT devices. It should be understood that the terms "Mobile Station" and "Mobile Device" also encompass equipment that has been installed for extended periods.

[0137] Furthermore, UE may also include items of production equipment, manufacturing equipment and / or energy-related machinery (for example, boilers, engines, turbines, solar panels, wind turbines, hydroelectric generators, thermal power generators, nuclear power generators, batteries, nuclear systems, nuclear-related equipment, heavy electrical equipment, pumps including vacuum pumps, compressors, fans, blowers, hydraulic equipment, pneumatic equipment, metalworking machinery, manipulators, robots, robot application systems, tools, molds, rolls, conveying equipment, lifting equipment, cargo handling equipment, textile machinery, sewing machinery, printing presses, printing-related machinery, paper processing machinery, chemical machinery, mining machinery, mining-related machinery, construction machinery, construction-related machinery, agricultural machinery and / or equipment, forestry machinery and / or equipment, fishing machinery and / or equipment, safety and / or environmental protection equipment, tractors, bearings, precision bearings, chains, gears, power transmissions, lubrication systems, valves, pipe fittings and / or application systems of any of the above-mentioned equipment or machinery).

[0138] Furthermore, UE may also be items related to transportation equipment (for example, vehicles, automobiles, motorcycles, bicycles, trains, buses, handcarts, rickshaws, ships and other watercraft, airplanes, rockets, satellites, drones, balloons, etc.).

[0139] Furthermore, the UE may also be, for example, an item for information and communication equipment (such as a computer and related equipment, communication equipment and related equipment, or electronic components).

[0140] Furthermore, UE may also include, for example, refrigerators, refrigerator applications and equipment, commercial and service equipment, vending machines, automated service machines, office machinery and equipment, and consumer electrical and electronic machinery and appliances (such as audio equipment, speakers, radios, video equipment, televisions, microwave ovens, rice cookers, coffee makers, dishwashers, washing machines, dryers, fans, ventilation fans and related products, vacuum cleaners, etc.).

[0141] Furthermore, the UE may also be, for example, an electronic application system or electronic application device (such as an X-ray device, particle accelerator, radioactive material application device, sound wave application device, electromagnetic application device, power application device, etc.).

[0142] Furthermore, UE may also include, for example, light bulbs, lighting, weighing machines, analytical instruments, testing machines and measuring instruments (such as smoke detectors, human alarm sensors, motion sensors, and wireless tags), watches or clocks, scientific and chemical instruments, optical instruments, medical equipment and / or medical systems, weapons, tools and implements, or hand tools.

[0143] Furthermore, the UE may be, for example, a personal digital assistant or device equipped with wireless communication capabilities (for example, an electronic device configured to have a wireless card or wireless module attached or inserted into it, such as a personal computer or electronic measuring instrument).

[0144] Furthermore, UE may be a device or part of a device that provides the following applications, services, and solutions in the "Internet of Things (IoT)" using wired or wireless communication technologies, for example:

[0145] An IoT device (or thing) comprises appropriate electronics, software, sensors, network connectivity, etc., that enable the device to collect and exchange data with each other and with other communication devices. An IoT device may also be an automated device that follows software instructions stored in its internal memory. An IoT device may operate without requiring human supervision or response. An IoT device may be a permanently installed device and / or remain inactive for extended periods. An IoT device may be implemented as part of a stationary device. An IoT device may be embedded in a non-stationary device (e.g., a vehicle) or attached to animals or people being monitored / tracked.

[0146] It will be understood that IoT technology can be implemented on any communication device that can connect to a communication network that sends and receives data, regardless of whether it is controlled by human input or by software instructions stored in memory.

[0147] It will be understood that IoT devices are sometimes called Machine Type Communication (MTC) devices or Machine to Machine (M2M) communication devices. It will also be understood that a UE can support one or more IoT or MTC applications. Some examples of MTC applications are listed in the table below (Source: 3GPP TS22.368 V13.2.0 (2017-01-13) Annex B, whose contents are incorporated herein by reference). This list is not exhaustive and presents only examples of MTC applications.

[0148] [Table 1]

[0149] Applications, services, and solutions include, for example, MVNO (Mobile Virtual Network Operator) services / systems, disaster prevention wireless services / systems, in-building wireless telephone (PBX (Private Branch eXchange)) services / systems, PHS / digital cordless telephone services / systems, POS (Point of sale) systems, advertising transmission services / systems, multicast (MBMS (Multimedia Broadcast and Multicast Service)) services / systems, V2X (Vehicle to Everything: vehicle-to-vehicle communication and roadside-to-vehicle / pedestrian-to-vehicle communication) services / systems, in-train mobile wireless services / systems, location information-related services / systems, disaster / emergency wireless communication services / systems, IoT (Internet of Things) services / systems, community services / systems, video distribution services / systems, Femto cell application services / systems, VoLTE (Voice over LTE) services / systems, wireless TAG services / systems, billing services / systems, radio on-demand services / systems, roaming services / systems, user behavior monitoring services / systems, communication carrier / communication NW selection services / systems, function-limiting services / systems, PoC (Proof of Concept) services / systems, personal information management services / systems for terminals, display / video services / systems for terminals, non-communication services / systems for terminals, ad hoc NW / DTN (Delay Tolerant Networking) services / systems, and the like.

[0150] Note that the categories of UEs described above are merely application examples of the technical ideas and embodiments described in this specification. They are not limited to these examples, and it is of course possible for those skilled in the art to make various changes.

[0151] Furthermore, it should be understood that the embodiments of this disclosure are not limited to these embodiments, and that a number of modifications and variations that a person skilled in the art could make in accordance with the principles of this disclosure are included in the spirit and scope of this disclosure.

[0152] 4. Addendum

[0153] Some or all of the embodiments and examples described above may also be described as follows, but are not limited to these.

[0154] (Note 1) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. Communication device.

[0155] (Note 2) In a) above, the at least two channel response matrices are estimated based on reference signals received from the other communication device at predetermined intervals. In the above c), the second eigenvector or singular vector is used until the next channel response matrix is ​​estimated. The communication device described in Appendix 1.

[0156] (Note 3) In the above c), If the prediction is performed in the same domain as the estimation in a), the prediction is performed using the real and imaginary parts of the at least two first eigenvectors or singular vectors. The prediction is performed using the magnitude and phase of the at least two first eigenvectors or singular vectors if the prediction is performed in a different domain than the estimation in a). A communication device as described in Appendix 1 or 2.

[0157] (Note 4) The communication device described in Appendix 1 or 2, which is performed when the eigenvalue or singular value corresponding to the at least one first eigenvector or singular vector is greater than a predetermined threshold.

[0158] (Note 5) The communication device described in any one of the appendices 1-4, wherein the prediction is performed for the same elements of the at least two first eigenvectors or singular vectors in all time slots in which the prediction is performed.

[0159] (Note 6) The communication device according to any one of the appendices 1-5, wherein, in (b) above, the at least two first eigenvectors or singular vectors are obtained in a consistent manner.

[0160] (Note 7) The communication device described in any one of Appendix 1-6, wherein the predetermined prediction method is at least one of linear extrapolation, nonlinear extrapolation, or machine learning-based time series forecasting.

[0161] (Note 8) The communication device according to any one of the appendices 1-7, wherein the at least two channel response matrices are estimates of the channel impulse response or channel frequency response.

[0162] (Note 9) The communication device according to any one of the appendices 1-8, wherein, in b) and c), at least one of the precoding matrix and beamforming weights of the wireless transceiver is calculated based on the at least two first eigenvectors or singular vectors and the at least one second eigenvector or singular vector for signal transmission.

[0163] (Note 10) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) The at least two channel response matrices are transformed from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation to generate at least two intermediate channel response matrices, b.2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication device.

[0164] (Note 11) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b.2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication device.

[0165] (Note 12) The communication device according to appendix 10 or 11, wherein the at least one first transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform, and the at least one second transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform.

[0166] (Note 13) The communication device according to any one of the appendices 10-12, wherein b.1) and b.2) are performed only in the time slot in which a) is performed, and c.1) and c.2) are performed in all time slots in which the prediction of at least one second eigenvector or singular vector is performed.

[0167] (Note 14) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) The at least two channel response matrices are transformed by at least one first transformation to generate at least two first intermediate channel response matrices, b.2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c.2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b.1) above, the at least one first transformation is, b.1.1) From the frequency domain to the delay domain, b.1.2) From the antenna domain to the angle or beamspace domain, and b.1.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In c.2) above, the at least one second transformation is c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain or beamspace domain to the antenna domain, c.2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Communication device.

[0168] (Note 15) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b.2) Transform the at least two eigenvectors or singular vectors by at least one first transformation to generate at least two first intermediate eigenvectors or singular vectors, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c.2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b.2) above, the at least one first transformation is, b.2.1) From the frequency domain to the delay domain, b.2.2) From the antenna domain to the angle or beamspace domain, and b.2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In c.2) above, the at least one second transformation is c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain or beamspace domain to the antenna domain, c.2.3) From the delay domain to the frequency domain, perform at least one of the conversions, the at least one second conversion performs an inverse conversion of the at least one first conversion, a communication device.

[0169] (Appendix 16) The communication device according to Appendix 14 or 15, wherein at least one of c.1) and c.2) is performed until a) becomes possible.

[0170] (Appendix 17) In c.1), 1) When neither the first conversion nor the second conversion is used in the prediction, the prediction is performed using the real and imaginary parts of the at least two first intermediate eigenvectors or eigenvectors, 2) When either the first conversion or the second conversion is used in the prediction, the prediction is performed using the magnitudes and phases of the at least two first intermediate eigenvectors or eigenvectors, The communication device according to any one of Appendices 10 - 16.

[0171] (Appendix 18) The communication device according to any one of Appendices 1 - 9, wherein the second eigenvector or eigenvector is optimized to reduce errors in prediction.

[0172] (Appendix 19) The communication device according to Appendix 18, wherein the second eigenvector or eigenvector is normalized to have a unit norm.

[0173] (Appendix 20) The communication device according to any one of Appendices 10 - 16, wherein the second intermediate eigenvector or eigenvector is optimized to reduce errors in prediction.

[0174] (Appendix 21) The communication device as described in Appendix 20, wherein the second intermediate eigenvector or singular vector is normalized to have a unit norm.

[0175] (Note 22) A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. Channel prediction methods.

[0176] (Note 23) In a) above, the at least two channel response matrices are estimated based on reference signals received from the other communication device at predetermined intervals. In the above c), the second eigenvector or singular vector is used until the next channel response matrix is ​​estimated. The channel prediction method described in Appendix 22.

[0177] (Note 24) The channel forecasting method according to Appendix 22 or 23, wherein the predetermined forecasting method is at least one of linear extrapolation, nonlinear extrapolation, or machine learning-based time series forecasting.

[0178] (Note 25) The channel prediction method according to any one of the appendices 22-24, wherein the at least two channel response matrices are estimates of the channel impulse response or channel frequency response.

[0179] (Note 26) The channel prediction method according to any one of the appendices 22-25, wherein, in b) and c), at least one of the precoding matrix and beamforming weights of the radio transceiver is calculated based on the at least two first eigenvectors or singular vectors and the at least one second eigenvector or singular vector for signal transmission.

[0180] (Note 27) A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) The at least two channel response matrices are transformed from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation to generate at least two intermediate channel response matrices, b.2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Channel prediction methods.

[0181] (Note 28) A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b.2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Channel prediction methods.

[0182] (Note 29) The channel prediction method according to Appendix 27 or 28, wherein the at least one first transformation is at least one of a discrete Fourier transform and an inverse discrete Fourier transform, and the at least one second transformation is at least one of a discrete Fourier transform and an inverse discrete Fourier transform.

[0183] (Note 30) The channel prediction method according to Appendix 27 or 28, wherein b.1) and b.2) are performed only in the time slot in which a) is performed, and c.1) and c.2) are performed in all time slots in which the prediction of at least one second eigenvector or singular vector is performed.

[0184] (Appendix 31) A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, comprising: a) Estimating at least two channel response matrices of the wireless channel based on a predetermined signal received from the other communication device; b.1) Converting the at least two channel response matrices by at least one first transformation to generate at least two first intermediate channel response matrices; b.2) Obtaining at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices; c.1) Predicting at least one second intermediate eigenvector or singular vector by the predetermined prediction method using the at least two first intermediate eigenvectors or singular vectors; c.2) Converting the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector; In b.1), the at least one first transformation is: b.1.1) From the frequency domain to the delay domain; b.1.2) From the antenna domain to the angle or beamspace domain, and b.1.3) From the time domain to the Doppler domain; Performing at least one of the transformations; In c.2), the at least one second transformation is: c.2.1) From the Doppler domain to the time domain; c.2.2) From the angle domain to the antenna domain, and c.2.3) From the delay domain to the frequency domain; Performing at least one of the transformations; The at least one second transformation performs an inverse transformation of the at least one first transformation. Channel prediction method.

[0185] (Note 32) A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b.1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b.2) Transform the at least two eigenvectors or singular vectors by at least one first transformation to generate at least two first intermediate eigenvectors or singular vectors, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c.2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b.2) above, the at least one first transformation is, b.2.1) From the frequency domain to the delay domain, b.2.2) From the antenna domain to the angle or beamspace domain, and b.2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In c.2) above, the at least one second transformation is c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain or beamspace domain to the antenna domain, c.2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Channel prediction methods.

[0186] (Note 33) The channel prediction method according to Appendix 31 or 32, wherein at least one of c.1) and c.2) is performed until a) becomes possible.

[0187] (Note 34) A non-temporary recording medium for storing a computer-readable program for channel prediction in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, wherein the computer-readable program is a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. A non-temporary recording medium consisting of instructions.

[0188] (Note 35) A computer-readable program for channel prediction that runs on at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, wherein the computer-readable program is a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. A computer-readable program consisting of instructions.

[0189] (Note 36) At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b) Obtain at least two first eigenvectors or singular vectors of the at least two channel response matrices, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the first wireless device during time slots in which channel estimation is not performed. Communication system.

[0190] (Note 37) At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b.1) The at least two channel response matrices are transformed from the antenna-frequency domain to the other domain using at least one first transformation to generate at least two intermediate channel response matrices, b.2) In the other domain, obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c.1) Predict the at least two second intermediate eigenvectors or singular vectors in the other domains, c.2) Convert the at least one second intermediate eigenvector or singular vector to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication system.

[0191] (Note 38) At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b.1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b.2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c.1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication system.

[0192] (Note 39) The aforementioned at least one first transformation includes at least one discrete Fourier transform and an inverse discrete Fourier transform. The communication system according to Appendix 37 or 38, wherein the at least one second transform includes at least one inverse discrete Fourier transform and a discrete Fourier transform.

[0193] (Note 40) In b.1) above, the at least two channel response matrices are, b.1.1) From the frequency domain to the delay domain, b.1.2) From the antenna domain to the angle domain, and b.1.3) From the time domain to the Doppler domain, Converted by at least one of the following: In c.2) above, the at least two second intermediate eigenvectors or singular vectors are c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain to the antenna domain, and c.2.3) From the delay domain to the frequency domain, Converted by at least one of the following: The communication system described in Appendix 37.

[0194] (Note 41) In b.2) above, the at least one first transformation is, b.2.1) From the frequency domain to the delay domain, b.2.2) From the antenna domain to the angle or beamspace domain, and b.2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In c.2) above, the at least one second transformation is c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain or beamspace domain to the antenna domain, c.2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. The communication system described in Appendix 38.

[0195] (Note 42) In the above c), the second eigenvector is sequentially predicted based on a predetermined number of first eigenvectors most recently calculated based on the predetermined signal received from the other communication device, as described in any one of the items in Appendix 1-21.

[0196] 5. Additional notes The above exemplary embodiments, in whole or in part, may be described as follows (abbreviated as FSN), but are not limited to these.

[0197] (FSN1) Memory for storing a program containing instructions for predicting eigenvectors or singular vectors, A controller configured to execute the aforementioned instructions, The controller has, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors of the at least two channel response matrices, c) Predicting at least one second eigenvector based on the at least two first eigenvectors using the prediction method, d) Transmitting a signal to the other communication device using at least one second eigenvector at a time, Device.

[0198] (FSN2) The above b) is, The at least two channel response matrices are transformed from the antenna-frequency domain to the other domain using at least one first transformation, Obtain the at least two first eigenvectors in the other domains, The above c) is, In the other domains mentioned above, predict the at least one second eigenvector, The at least one second eigenvector is converted to the antenna-frequency domain using at least one second transformation, The device described in FSN1.

[0199] (FSN3) The above b) is, b.1) Obtain at least two eigenvectors of the at least two channel response matrices, b.2) Convert the at least two eigenvectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors in the beamspace-delay domain, The above c) is, c.1) Using the at least two first intermediate eigenvectors, predict at least one second intermediate eigenvector using a predetermined prediction method, c.2) Convert the at least one second intermediate eigenvector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector. The device described in FSN1.

[0200] (FSN4) (a) The first transformation is at least one of the discrete Fourier transform or the inverse discrete Fourier transform or a variant thereof, (b) The apparatus according to FSN2 or 3, wherein the second transformation is at least one inverse discrete Fourier transform or discrete Fourier transform or a variant thereof.

[0201] (FSN5) The at least one first transformation is, b.1) From the frequency domain to the delay domain, b.2) From the antenna domain to the angle domain, b.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: The aforementioned at least one second transformation is c.1) From the Doppler domain to the time domain, c.2) From the angle domain to the antenna domain, c.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The device described in FSN2.

[0202] (FSN6) In b.2) above, the at least one first transformation is, b.2.1) From the frequency domain to the delay domain, b.2.2) From the antenna domain to the angle or beamspace domain, and b.2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In c.2) above, the at least one second transformation is c.2.1) From the Doppler domain to the time domain, c.2.2) From the angle domain or beamspace domain to the antenna domain, c.2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. The device described in FSN3.

[0203] (FSN7) A communication system comprising at least one first wireless device and at least one second wireless device, The first wireless device transmits at least one reference signal to the second wireless device, The second wireless device includes a controller, and the controller is a) Using the reference signal, estimate at least two channel response matrices between the second wireless device and the second wireless device, b) Obtain at least two first eigenvectors of the estimated at least two channel response matrices, c) Predicting at least one second eigenvector based on the at least two first eigenvectors using the prediction method, d) Transmitting a signal to the other communication device using at least one second eigenvector at a time, Communication system.

[0204] (FSN8) The above a) transforms the at least two estimated channel response matrices from the antenna-frequency domain to the other domain using at least one first transformation, The above b) obtains the at least two first eigenvectors or singular vectors in the other domain, c) predicts the at least one second eigenvector in the other domain and transforms the at least one second eigenvector to the antenna-frequency domain using at least one second transformation. The communication system described in FSN7.

[0205] (FSN9) (a) The first transformation is at least one of the discrete Fourier transform or the inverse discrete Fourier transform or a variant thereof, (b) The second transformation is at least one of the inverse discrete Fourier transform or the discrete Fourier transform or a variant thereof. The communication system described in FSN8.

[0206] (FSN10) (a) The at least two estimated channel response matrices are, a.1) From the frequency domain to the delay domain, a.2) From the antenna domain to the angle domain, and a.3) From the time domain to the Doppler domain, It is converted to another domain by at least one of the following: (b) The output of the prediction calculation is b.1) From the Doppler domain to the time domain, b.2) From the angle domain to the antenna domain, b.3) From the delay domain to the frequency domain, At least one of the above is converted to the other domains, The communication system described in FSN8.

[0207] 6. Further additions All or part of the exemplary embodiments described above may be described as further additional notes (abbreviated as SFSN) below, but are not limited thereto.

[0208] (SFSN1) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors of the estimated at least two channel response matrices, c) The prediction method predicts, based on the at least two first eigenvectors, at least one second eigenvector for performing signal transmission to the other communication device in a time slot where channel estimation is not performed. Communication device.

[0209] (SFSN2) A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) In the first domain, estimate at least two channel response matrices of the wireless channel based on predetermined signals received from the other communication device, b) In the first domain, obtain at least two first eigenvectors of the estimated at least two channel response matrices, c) The at least two first eigenvectors of the estimated at least two channel response matrices are transformed from the first domain to the second domain, d) In the second domain, predict at least one second eigenvector based on the at least two first eigenvectors using the prediction method, e) The predicted at least one second eigenvector is transformed from the second domain to the first domain, f) In order to transmit a signal to the other communication device during a time slot in which channel estimation is not performed, use the predicted at least one second eigenvector in the first domain, Communication device.

[0210] (SFSN3) It has a wireless transceiver with a controller, The aforementioned controller a) Calculate a metric to compare the predicted eigenvectors with the actual eigenvectors, b) Select a prediction method based on the metric. Communication device.

[0211] (SFSN4) The communication device according to SFSN3, wherein, in (b) above, the second eigenvector is sequentially predicted based on a predetermined number of first eigenvectors most recently acquired based on the predetermined signal received from the other communication device. [Industrial applicability]

[0212] The above exemplary embodiments are applicable to wireless communication systems employing beamforming transmission. [Explanation of symbols]

[0213] 100 Base Station (BS) Devices 101 RF (Radio Frequency) Front End 102 Fast Fourier Transform (FFT) section 103 Inverse FFT (IFFT) section TR(1)-TR(M) Wireless Transceiver 104 Channel Estimator 105 Eigenvector Predictor 106 Precoder

Claims

1. A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. Communication device.

2. In a) above, the at least two channel response matrices are estimated based on reference signals received from the other communication device at predetermined intervals. In the above c), the second eigenvector or singular vector is used until the next channel response matrix is ​​estimated. The communication device according to claim 1.

3. In the above c), If the prediction is performed in the same domain as the estimation in a), the prediction is performed using the real and imaginary parts of the at least two first eigenvectors or singular vectors. The prediction is performed using the magnitude and phase of the at least two first eigenvectors or singular vectors if the prediction is performed in a domain different from the estimation in a). A communication device according to claim 1 or 2.

4. The communication device according to claim 1 or 2, wherein the above c) is performed when the eigenvalue or singular value corresponding to the at least one first eigenvector or singular vector is greater than a predetermined threshold.

5. The communication device according to any one of claims 1 to 4, wherein the c) is performed for the same elements of the at least two first eigenvectors or singular vectors in all time slots in which the prediction is performed.

6. The communication device according to any one of claims 1 to 5, wherein, in (b) above, the at least two first eigenvectors or singular vectors are obtained in a consistent manner.

7. The communication device according to any one of claims 1 to 6, wherein the predetermined prediction method is at least one of linear extrapolation, nonlinear extrapolation, or machine learning-based time series prediction.

8. The communication device according to any one of claims 1 to 7, wherein the at least two channel response matrices are estimates of the channel impulse response or channel frequency response.

9. The communication device according to any one of claims 1-8, wherein in b) and c), at least one of the precoding matrix and beamforming weights of the wireless transceiver is calculated based on the at least two first eigenvectors or singular vectors and the at least one second eigenvector or singular vector for signal transmission.

10. A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Convert the at least two channel response matrices from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation to generate at least two intermediate channel response matrices, b. 2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c. 2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication device.

11. A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b. 2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c. 2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication device.

12. The communication device according to claim 10 or 11, wherein the at least one first transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform, and the at least one second transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform.

13. The communication device according to any one of claims 10-12, wherein b. 1) and b. 2) are performed only in the time slot in which a) is performed, and c. 1) and c. 2) are performed in all time slots in which the prediction of at least one second eigenvector or singular vector is performed.

14. A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) At least two first intermediate channel response matrices are generated by transforming the at least two channel response matrices by at least one first transformation, b. 2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c. 2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b. 1) above, the at least one first transformation is, b. 1.1) From the frequency domain to the delay domain, b. 1.2) From the antenna domain to the angle or beamspace domain, and b. 1.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In the above c. 2), the at least one second transformation is, c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain or beamspace domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Communication device.

15. A wireless transceiver configured to communicate with other communication devices via a wireless channel, At least one processor configured to execute instructions, The processor has, and the at least one processor a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b. 2) Transform the at least two eigenvectors or singular vectors by at least one first transformation to generate at least two first intermediate eigenvectors or singular vectors, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c. 2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b. 2) above, the at least one first transformation is, b. 2.1) From the frequency domain to the delay domain, b. 2.2) From the antenna domain to the angle or beamspace domain, and b. 2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In the above c. 2), the at least one second transformation is, c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain or beamspace domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Communication device.

16. The communication device according to claim 14 or 15, wherein at least one of c. 1) and c. 2) is performed until a) becomes possible.

17. In the above c. 1), 1) If neither the first transformation nor the second transformation is used, the prediction is performed using the real and imaginary parts of the at least two first intermediate eigenvectors or singular vectors. 2) The prediction is performed using the magnitude and phase of the at least two first intermediate eigenvectors or singular vectors, if either the first transformation or the second transformation is used. A communication device according to any one of claims 10-16.

18. The communication device according to any one of claims 1 to 9, wherein the second eigenvector or singular vector is optimized to reduce errors in prediction.

19. The communication device according to claim 18, wherein the second eigenvector or singular vector is normalized to have a unit norm.

20. The communication device according to any one of claims 10-16, wherein the second intermediate eigenvector or singular vector is optimized to reduce errors in prediction.

21. The communication device according to claim 20, wherein the second intermediate eigenvector or singular vector is normalized to have a unit norm.

22. A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. Channel prediction methods.

23. In a) above, the at least two channel response matrices are estimated based on reference signals received from the other communication device at predetermined intervals. In the above c), the second eigenvector or singular vector is used until the next channel response matrix is ​​estimated. The channel prediction method according to claim 22.

24. The channel prediction method according to claim 22 or 23, wherein the predetermined prediction method is at least one of linear extrapolation, nonlinear extrapolation, or machine learning-based time series prediction.

25. The channel prediction method according to any one of claims 22-24, wherein the at least two channel response matrices are estimates of the channel impulse response or channel frequency response.

26. The channel prediction method according to any one of claims 22-25, wherein in b) and c), at least one of the precoding matrix and beamforming weights of the wireless transceiver is calculated based on the at least two first eigenvectors or singular vectors and the at least one second eigenvector or singular vector for signal transmission.

27. A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Convert the at least two channel response matrices from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation to generate at least two intermediate channel response matrices, b. 2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c. 2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Channel prediction methods.

28. A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b. 2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c. 2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Channel prediction methods.

29. The channel prediction method according to claim 27 or 28, wherein the at least one first transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform, and the at least one second transform is at least one of a discrete Fourier transform and an inverse discrete Fourier transform.

30. The channel prediction method according to claim 27 or 28, wherein b. 1) and b. 2) are performed only in the time slot in which a) is performed, and c. 1) and c. 2) are performed in all time slots in which the prediction of at least one second eigenvector or singular vector is performed.

31. A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) At least two first intermediate channel response matrices are generated by transforming the at least two channel response matrices by at least one first transformation, b. 2) Obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using the predetermined prediction method, c. 2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b. 1) above, the at least one first transformation is, b. 1.1) From the frequency domain to the delay domain, b. 1.2) From the antenna domain to the angle or beamspace domain, and b. 1.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In the above c. 2), the at least one second transformation is, c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Channel prediction methods.

32. A channel prediction method by at least one processor in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b. 1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b. 2) Transform the at least two eigenvectors or singular vectors by at least one first transformation to generate at least two first intermediate eigenvectors or singular vectors, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c. 2) Transform the at least one second intermediate eigenvector or singular vector by at least one second transformation to generate the at least one second eigenvector or singular vector, In b. 2) above, the at least one first transformation is, b. 2.1) From the frequency domain to the delay domain, b. 2.2) From the antenna domain to the angle or beamspace domain, and b. 2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In the above c. 2), the at least one second transformation is, c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain or beamspace domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. Channel prediction methods.

33. The channel prediction method according to claim 31 or 32, wherein at least one of c. 1) and c. 2) is performed until a) becomes possible.

34. A non-temporary recording medium for storing a computer-readable program for channel prediction in a communication device including a wireless transceiver configured to communicate with other communication devices through a wireless channel, wherein the computer-readable program is a) Based on predetermined signals received from the other communication device, estimate at least two channel response matrices of the wireless channel, b) Obtain at least two first eigenvectors or singular vectors of the channel response matrix, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the other communication device during time slots in which channel estimation is not performed. A non-temporary recording medium consisting of instructions.

35. At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b) Obtain at least two first eigenvectors or singular vectors of the at least two channel response matrices, c) Predicting at least one second eigenvector or singular vector based on the at least two first eigenvectors or singular vectors using a predetermined prediction method, wherein the at least one second eigenvector or singular vector is used to transmit a signal to the first wireless device during time slots in which channel estimation is not performed. Communication system.

36. At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b. 1) The at least two channel response matrices are transformed from the antenna-frequency domain to other domains using at least one first transformation to generate at least two intermediate channel response matrices, b. 2) In the other domain, obtain at least two first intermediate eigenvectors or singular vectors of the at least two intermediate channel response matrices, c. 1) Predict the at least two second intermediate eigenvectors or singular vectors in the other domains, c. 2) Convert the at least one second intermediate eigenvector or singular vector to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication system.

37. At least one first wireless device, A second wireless device including a wireless transceiver and a controller, The wireless transceiver of a second wireless device is configured to communicate with the first wireless device through a wireless channel. The aforementioned controller a) Using at least one reference signal received from the first wireless device, estimate at least two channel response matrices between the second wireless device and the first wireless device. b. 1) Obtain at least two eigenvectors or singular vectors of the at least two channel response matrices, b. 2) Convert the at least two eigenvectors or singular vectors from the antenna-frequency domain to the beamspace-delay domain by at least one first transformation, and generate at least two first intermediate eigenvectors or singular vectors in the beamspace-delay domain, c. 1) Using the at least two first intermediate eigenvectors or singular vectors, predict at least one second intermediate eigenvector or singular vector using a predetermined prediction method, c. 2) Convert the at least one second intermediate eigenvector or singular vector from the beamspace-delay domain to the antenna-frequency domain by at least one second transformation to generate the at least one second eigenvector or singular vector. Communication system.

38. The at least one first transformation includes at least one of the discrete Fourier transform and the inverse discrete Fourier transform. The communication system according to claim 36 or 37, wherein the at least one second transform includes at least one inverse discrete Fourier transform and discrete Fourier transform.

39. In b. 1) above, the at least two channel response matrices are, b. 1.1) From the frequency domain to the delay domain, b. 1.2) From the antenna domain to the angle domain, and b. 1.3) From the time domain to the Doppler domain, Converted by at least one of the following: In the above c. 2), the at least two second intermediate eigenvectors or singular vectors are c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Converted by at least one of the following: The communication system according to claim 36.

40. In b. 2) above, the at least one first transformation is, b. 2.1) From the frequency domain to the delay domain, b. 2.2) From the antenna domain to the angle or beamspace domain, and b. 2.3) From the time domain to the Doppler domain, Perform at least one of the following transformations: In the above c. 2), the at least one second transformation is, c. 2.1) From the Doppler domain to the time domain, c. 2.2) From the angle domain or beamspace domain to the antenna domain, and c. 2.3) From the delay domain to the frequency domain, Perform at least one of the following transformations: The at least one second transformation performs the inverse transformation of the at least one first transformation. The communication system according to claim 37.

41. The communication device according to any one of claims 1 to 21, wherein, in c) above, the second eigenvector is sequentially predicted based on a predetermined number of first eigenvectors most recently calculated based on the predetermined signal received from the other communication device.

Images