Method and appartus for communications

Abstract

Embodiments of the present application provide a method and an apparatus for communication. Channel estimation is implemented on the UE side, and limited prior knowledge about channel sub-space is known in advance. A base station shares a sensing matrix and indexes of columns to be used in a dictionary with a UE. The BS transmits CSI-RS on a few symbols, which are far less than the number of symbols for transmitting CSI-RS in the conventional channel estimation procedure using orthogonal beam training, to a UE to obtain a sparse matrix based on the sensing matrix and the column indexes of the dictionary by solving an inverse regularized problem. The UE provides a compressed measurement report to the BS to obtain an optimized analog precoder. The UE further transmits SRS to the BS, and the BS receives the SRS using the optimized analog precoder and obtains effective channel estimation which is further used to determine a baseband precoder. The BS determines a final precoder related to the optimized analog precoder and the baseband precoder, which is used for data transmission. The proposed solution reduces search space and beam training overhead in time domain.

Description

METHOD AND APPARATUS FOR COMMUNICATIONSTECHNICAL FIELD

[0001] Embodiments of the present invention relate to the field of wireless technologies, and more specifically, to a method and an apparatus for communications.BACKGROUND

[0002] A hybrid beamforming architecture (HBF) provides reasonable trade-off between hardware complexity and performance for a large MIMO system as a classical full digital architecture in most of cases is not feasible due to cost, energy efficiency and hardware complexity. There are plenty of difference types of HBF architectures such as a partially connected (PC) architecture, a fully connected (FC) architecture, an overlapped sub-arrays architecture, and an adaptive port mapping (APM) architecture. The PC architecture is the most practical as it provides significant hardware simplification by the expense of performance and beamforming flexibility degradation. Only a non-intersected sub-set (i.e., partial number) of antennas is connected to each RF chain, i.e., any antenna is connected to only one RF chain simultaneously and such connection cannot be dynamically changed due to passive nature of an analog power divider which maps physical antennas to logical ports (RF chains), and this way of antenna mapping is “fixed mapping”.

[0003] Although the PC-HBF is the most suitable from hardware complexity point view, channel estimation for the PC-HBF is done via a beam training procedure consuming a number of symbols in time domain in a conventional solution. Therefore, overhead reduction in time domain is an urgent problem that needs to be solved.SUMMARY

[0004] Embodiments of the present application provide a method and an apparatus forcommunications, which can reduce overhead in time domain in a channel estimation procedure in a PC-HBF.

[0005] According to a first aspect, there is provided a method for communication. The method can be implemented by a first device. The first device may be a terminal device or a device installed in the terminal device such as a chip, a chip system or a circuit, etc. The method includes: receiving a configuration of a sensing matrix and a set of indexes of columns to be used in a dictionary, where the sensing matrix is determined according to a set of T0analog precoders and a channel structure descried with the dictionary, a first analog precoder is determined according to a first mapping matrix and a phase of every column of the first analog precoder, the first analog precoder is any one of the T0analog precoders, the first mapping matrix is used to indicate a connection relationship between physical antennas and NRFradio frequency (RF) chains, is the number of physical antennas, and NRFis thenumber of RF chains, and T0is the number of beam training rounds and is a positive integer; and transmitting a compressed measurement report based on the sensing matrix and the set of indexes of the columns to be used in the dictionary.

[0006] Based on the proposed solution of the present application, channel estimation is done at the UE side, and limited prior knowledge about signal sub-space is known in advance.The network device (for example, a BS) shares a sensing matrix and a set of indexes of the columns to be used in a dictionary with the UE, so that the UE can make compressed channel estimation and provide a compressed report to the network device. The set of indexes of the columns to be used at the UE side is prior knowledge about the channel or the signal subspace which can be obtained by the network device in advance. The overhead of beam training in time domain can be reduced because o where T0is the number of beam training roundsin the proposed solution, and is the number of beam training rounds in the conventionalsolution.

[0007] In an implementation of the first aspect, where receiving the configuration of the sensing matrix further includes: receiving T0mapping matrixes corresponding to the T0analog precoders respectively and quantized phases of every column of each of the T0analog precoders.

[0008] In this implementation, sharing of a sensing matrix from a network device to a UE may result in larger overhead, so mapping matrixes corresponding to the analog precoders respectively and quantized phases of every column of each of the analog precoders are sent to the UE. The overhead of sharing the sensing matrix can be reduced.

[0009] In an implementation of the first aspect, where the method further includes: receiving a CSI-RS on T0time units, where the CSI-RS on the T0time units are precoded by the T0analog precoders respectively; and obtaining a sparse matrix based on the CSI-RS on the T0time units, the sensing matrix and the set of indexes of the columns to be used in the dictionary.

[0010] A network device sends CSI-RSs in downlink on T0time units (e.g., T0symbols), and the CSI-RSs on the T0time units are precoded by the T0analog precoders respectively.A UE receives the CSI-RSs to solve the OMP-like problem to recover the sparse channel.

[0011] In an implementation of the first aspect, where the method further includes: receiving a configuration of a sounding reference signal (SRS), where the configuration of theSRS includes time resources for the SRS, and a time resource comprises a few time units; and transmitting the SRS on the few time units, where the SRS on the few time units are used for effective channel estimation, and an effective channel is used to determine a baseband precoder.

[0012] In this implementation, an optional stage of sending a SRS for channel estimation can be done, which helps to obtain better accuracy of effective channels. Improvement of accuracy of the effective channels helps to improve accuracy of a total precoder for data transmission. Therefore, performance of the data transmission can be improved.

[0013] According to a second aspect, there is provided a method for communication. The method may be implemented by a second device. The second device may be a network function or a device installed in the network function such as a chip, a chip system, or a circuit, etc. the method includes: transmitting a configuration of a sensing matrix and a set of indexes of columns to be used in a dictionary, where the sensing matrix is determined according to a set of T0analog precoders and a channel structure descried with the dictionary, a first analog precoder is determined according to a first mapping matrix and a phase of ever column of the first analog precoder, the first analog precoder is any one of the T0analog precoders, the first mapping matrix is used to indicate a connection relationship between physical antennas andNRFradio frequency (RF) chains, is the number of physical antennas, NRFisthe number of RF chains, and T0is the number of beam training rounds and is a positive integer; and receiving a compressed measurement report obtained based on the sensing matrix and the set of indexes of the columns to be used in the dictionary.

[0014] Technical effects of the method in the second aspect or in some implementations of the second aspect can refer to those in the first aspect, therefore, they are not repeated.

[0015] In an implementation of the second aspect, where transmitting the configuration of the sensing matrix further includes: transmitting T0mapping matrixes corresponding to the T0analog precoders respectively and quantized phases of every column of each of the T0analog precoders.

[0016] In an implementation of the second aspect, where the method further includes: transmitting a CSI-RS on T0time units, where the CSI-RS on the T0time units are precoded by the T0analog precoders respectively.

[0017] In an implementation of the second aspect, where the compressed measurement report is obtained based on the CSI-RS on the T0time units, the sensing matrix and the set of indexes of the columns to be used in the dictionary; and the method further includes: obtaining an optimized analog precoder for data transmission according to the compressed measurement report by solving an optimization problem.

[0018] In an implementation of the second aspect, where after receiving the compressed measurement report, the method further includes: transmitting a configuration of sounding reference signal (SRS), where the configuration of the SRS includes time resources for the SRS, and a time resource includes a few time units; receiving the SRS on the few time units using the optimized analog precoder; and obtaining a baseband precoder for data transmission based on an effective channel that is calculated based on the SRS.

[0019] In an implementation of the second aspect, where the method further includes: determining a final matrix for data transmission, where the final matrix is related to the optimized analog precoder and the baseband precoder; and transmitting data based on the final matrix.

[0020] In an implementation of the first aspect or the second aspect, where each of the T0mapping matrixes is irregular, each irregular mapping matrix is obtained by performing analog pilot generation and phase modulation on every physical antenna.

[0021] In this implementation, irregular mapping matrixes provide better mutual coherence of a sensing matrix with a reduced number of training rounds, which provides better channel estimation with a small number of training rounds. The training overhead is reduced.

[0022] In an implementation of the first aspect or the second aspect, where the analog pilot generation comprises any one of: generating only a carrier grid with some carrier spacing on baseband, and using a same carrier grid for every radio frequency (RF) chain; performing modulation for every phase shifter such that antennas within one group generate the same phase modulation law; and adding constant phase shifts in time domain achieved from optimization of a Gram matrix to every antenna group.

[0023] In this implementation, irregular mapping matrixes can be artificially created by performing analog pilot waveform generation on every physical antenna if a phase shifter is assumed for every physical antenna. Different methods for analog pilot generation are provided, which does not increase complexity of the PC-HBF architecture as a complex RF switch network is not needed.

[0024] In an implementation of the first aspect or the second aspect, where each antenna is connected with more than one phase shifter and a power amplifier (PA) such that signals from every phase shifter is combined via a power divider (PD).

[0025] Note that, each physical antenna is connected with more than one phase shifter, but the same signal goes to an input of every phase shifter from RF chain, because this signal is divided by a small power divider (PD). Then this signal is modulated differently by every PS because we apply different modulation law on every PS such to achieve multiple pilots wave forms at an output of every phase shifter and have multiple pilots per physical antenna despite the fact that every physical antenna is connected to only one RF chain. In classical solution it can have only 1 pilot, but multiple phase shifter allows to increase pilots and allows to make overlapped sub-arrays virtually in the present application.

[0026] Therefore, in this implementation, overlapped irregular analog matrixes (i.e., overlapped sub-arrays) during pilot transmission can be achieved without significant hardware increasement, just by putting more phase shifters for every physical antenna and constrain gainpower amplifier (PA), so complexity and size would not be increased for power dividers.

[0027] In an implementation of the first aspect or the second aspect, where the number of time units for the SRS is one or two. In this implementation, channel estimation with a small number of training rounds is allowed. Therefore, a training overhead is reduced.

[0028] In an implementation of the first aspect or the second aspect, where a time unit is a symbol.

[0029] According to a third aspect, there is provided a communication apparatus having a function or module to perform the method in the first aspect or the second aspect, or any one of the implementations in these aspects.

[0030] According to a fourth aspect, there is provided an integrated circuit. The integrated circuit includes at least one processor, where the at least one processor is coupled to at least one memory. The at least one memory is configured to store one or more instructions and / or executable computer code. The at least one processor is configured to invoke the one or more instructions and / or executable computer code, so that a communication apparatus comprising the integrated circuit can perform the method of the first aspect or the second aspect, or any one of the possible implementations in these aspects. Optionally, the integrated circuit may further include the at least one memory. Optionally, the integrated circuit may further include a communication interface, and the communication interface is configured to input and / or output information or data.

[0031] According to a fifth aspect, there is provided a communication apparatus. The communication apparatus includes one or more circuits and one or more communication interfaces. The one or more communication interfaces may include a first interface for receiving(that is, inputting) information and / or data that is to be processed by the one or more circuits and a second interface for transmitting (that is, outputting) information and / or data processed by the one or more circuits. The one or more circuits are configured to process the information and / or data that is to be processed so that the communication apparatus performs the method of the first aspect or the second aspect, or any one of the implementations in these aspects.

[0032] According to a sixth aspect, there is provided a communication system. The communication system may include at least one communication apparatus in the first aspect and a communication apparatus in the second aspect.

[0033] According to a seventh aspect, there is provided a computer storage medium that stores executable computer code, and the executable computer code is used to execute one or more instructions for the method according to the first aspect or the second aspect, or any one of the possible implementations in these aspects.

[0034] According to an eighth aspect, there is provided a computer program product including one or more instructions, and when the computer product program runs on a computer, the computer performs the method according to the first aspect or the second aspect, or any one of the possible implementations in these aspects.DESCRIPTION OF DRAWINGS

[0035] One or more embodiments are exemplarily described by corresponding accompanying drawings, and these exemplary illustrations and accompanying drawings constitute no limitation on the embodiments. Elements with the same reference numerals in the accompanying drawings are illustrated as similar elements, and the drawings are not limited to scale, in which:

[0036] FIG. 1 illustrates a PC-HBF architecture.

[0037] FIG. 2 illustrates a DL beam training procedure using CSI-RS.

[0038] FIG. 3 illustrates an example of an incomplete signal subspace in beam domain which is assumed as prior knowledge based on previous historical measurements.

[0039] FIG. 4 are examples of measurement matrixes of conventional solutions which are:(a) - random phases for fixed mapping HBF, (b) - uniformly random antenna sampling, one antenna per RF chain

[0040] FIG. 5 is a schematic flow chart of a method (200) for communication according to an embodiment of the present application.

[0041] FIG. 6 is an example of a channel estimation procedure according to an embodiment of the present application.

[0042] FIG. 7 are examples of difference shapes of mapping matrixes.

[0043] FIG. 8 illustrates Gram matrix comparison of proposed mapping optimization (left) and a conventional solution (right).

[0044] FIG. 9 illustrates correlation of recovered signals for 12% overhead (left) and Gram matrix properties (right).

[0045] FIG. 10 illustrates an example of a sensing matrix sharing process.

[0046] FIG. 11 illustrates examples of analog pilot generation.

[0047] FIG. 12 illustrates a switching speed, a pilot spectrum, and an autocorrelation function of an analog pilot.

[0048] FIG. 13 illustrates overlapped sub-arrays with multiple phase shifters and a simple power divider according to an example of the present application.

[0049] FIG. 14 is a schematic block diagram of a communication apparatus according to some embodiments of the present application.

[0050] FIG. 15 is a schematic block diagram of a communication apparatus according to some embodiments of the present application.DESCRIPTION OF EMBODIMENTS

[0051] In order to understand features and technical contents of embodiments of the present application in detail, implementations of the embodiments of the present application will be described in detail below with reference to the accompanying drawings, and the attached drawings are only for reference and illustration purposes, and are not intended to limit the embodiments of the present applications. In the following technical descriptions, for ease of explanation, numerous details are set forth to provide a thorough understanding of the disclosed embodiments.

[0052] Related technical concepts and technologies are introduced first in order to understand the proposed solution of the present application.

[0053] A problem formulation of compressed channel estimation starts with restrictions related to a HBF-MIMO architecture shown in FIG. 1.

[0054] FIG. 1 illustrates a PC-HBF architecture. A transmitter transmits data precoded by a precoder consisting of two parts, that is, a baseband precoder and an analog precoder. In FIG.1 , K denotes a dimension of a transmitting signal, or we can say the transmitter transmits K streams of the transmitting signal, NRFdenotes the number of RF chains, and Ntxdenotesthe number of antennas. FBBis a transmitting digital beamformer matrix, and NRFis a transmitting analog beamformer matrix.

[0055] FIG. 2 illustrates an example of a conventional channel estimation beam training procedure in downlink (DL) using CSI-RS. The BS sends a synchro signal block (SSB) using symbols to roughly measure a link budget and make frame synchronization and roughdoppler compensation. The UE sends back receiver power, for example, reference signal receiving power (RSRP), per discrete Fourier transform (DFT) analog beam. In order to obtain channel state information (CSI), BS should make beam refinement either by SRS in UL or CSI-RS pilots in DL, here we will consider CSI-RS in DL for example. The BS makes beam refinement and sends a CSI-RS burst for channel estimation using symbols. The UEmakes channel estimation (CE) after the CSI-RS burst, and then sends a precoding matrix indicator (PMI) and channel quality indicator (CQI) using for example enhanced codebook type-II (beam-tap domain sparse coefficients). The BS recovers a full channel matrix according to feedback of theHBF optimization further can be done like and a total precoder for data transmission is determined like

[0056] It can be seen from FIG. 2, channel estimation for PC-HBF is done via a beam training procedure consuming a number of symbols in time domain, specifically, symbols.

[0057] However, in the present, we assume that a signal can be sparse in beam domain. As an additional source of prior art information, we can assume knowledge of signal subspace (or simply speaking, possible directions or angle of arrival (AoA) or angle of departure (AoD) where most of UE channel power can be located can be known in advance) as shown in FIG. 3.

[0058] It is seen from FIG. 3 that signal subspace may not necessarily occupy all possible angles from semi-sphere of antenna array, meaning that we should not expect signal to come from arbitrary direction but from specific set of angels which restrict search area and potentially reduce search space and beam training overhead. Such beam space knowledge can be prior achieved on BS based on some historical measurements or long-term statistic accumulations or channel learning procedure or quasi-deterministic geometrical channel models. Therefore, it isassumed that search space is not full and we will call such basis incomplete reduced dictionary.In such assumptions, compressed channel estimation problem is formulated as follows:

[0059] 1) channel in some sparsifying basis, for example (not necessary) DFT basis (sparse representation assumption) a) where H represents the full channel, A represents areceiving sparsifying basis or dictionary (typically is full basis due to random nature of UE rotation), represents a sparse channel matrix or compressible channel matrix in someproper basis, represents a conjugate transposed matrix of a transmitting sparsifying basisor reduced dictionary .

[0060] 2) received signal on t-th round at UE side a)whereYtrepresents a received signal ont-th round at UE side, represents an analog precoder used for precoding the pilots in t-th round at BS side, represents pilots in allocated in frequency or time domain, Nrepresents noise signal, represents the number of physical antennas adopted at thereceiving device (e.g., the UE), and Nrfrepresents the number of the RF chains.

[0061] 3) Each RF chain sends orthogonal pilots in frequency or time domain or pilots allocated on orthogonal resources using regular resource element grid (COMB structure) precoded with columns of UE makes estimation of effective channel using conventionalleast squares (LS) approach. a)where represents the effective channel at UE side, Ntxrepresents the number ofphysical antennas adopted at the transmitting device (e.g., the BS), and Nbis the number of the columns of an incomplete dictionary Atx(which is used to describe a channel structure).

[0062] 4) Using conventional multilinear CS approach sparse channel matrix Hspcan be estimated after applying vectorization using conventional mathematical expressiona)) (( ) where represents avectorized result of estimation of the effective channel of

[0063] 5) In order to improve accuracy of compressed channel estimation accumulation of measurements and sensing matrixes Stfor every t-th roung is done a) represents a transposed vector ofb) jstotal sensing matrix if

[0064] 6) solve inverse regularized problem under sparsity constraint using for exampleOMP a)

[0065] 7) sends sparse matrix Hspto BS using similar method like PMI index (or codebook type-II).

[0066] Important to mention that successful recovery of sparse channel Hsptightly depends on quality of total sensing matrix S which is in turn depends on total measurement matrixused to transmit precoded pilots signal in DL CSI-RS.Also, in order to solve OMP problem at UE side, matrix FRFand dictionary matrix Atxshould be known at UE side (in case where full sparsifying basis is known at UE, for exampleDFT basis, only indexes of columns which are used for channel recovery are needed on UE side). In conventional approach measurement matrix is not optimized for cell specific scenario, but just stored as pseudo-random sequence which is suitable for general case, where now additional prior knowledge about beam domain sub-space is available at BS side.

[0067] To improve accuracy of compressed channel recovery in presence of prior knowledge, total sensing matrix 5 should be optimized by changing total measurement matrix in order to minimize so called mutual coherency of the sensing matrix. Such optimization function directly minimizes maximum correlation between any columns of total sensing matrix such that more spatial directions can be estimated without ambiguity appearance. This means that optimal sensing matrix with small mutual coherence increases number of sparse channel coefficients which can be estimated by small number of measurements in comparison with sensing matrix with large mutual coherence:

[0068] Additionally, mutual coherence can be optimized using indirect methods by optimizing corresponding Gram matrix , trying to solve following problem:

[0069] Where G is Gram matrix which should be as close as possible to Identity matrix in order to minimize mutual coherence,arbitrary measurement matrix with all non-zero elements and without constraints, - practical HBF measurement matrix with modulusconstraint and with structural constraint of mapping matrix (positions of zero elements in FRFare correspond to position of zeros in mapping matrix M), M - total mapping matrix showing physical antenna mapping configuration during every round for every RF chain, mapping matrix is binary matrix containing only ones and zeros such that It is knownthat mutual coherency and Gram matrix optimization depends on prior knowledge ofdictionary structure or channel structure

[0070] Mutual coherence minimization allows to recovery signal with smaller sparsity with smaller number of measurements, or another words, smaller number of measurements are needed to recovery sparse signal if the measurement matrix provides better mutual coherency of the total sensing matrix. In embodiments of the present application, measurement reduction means training overhead reduction.

[0071] Important to say that, in prior art, the measurement matrix FRFis usually taken as random uniformly sampled antennas where on every round each RF chain is connected to only single antenna or random phase sub-arrays where on every training round every RF chain is connected to sub-set of antennas via power divider and set of phase shifter which generate uniformly random phase distribution across sub-array. Both measurement matrixes proved to satisfy restricted isometry property (RIP) with small RIP constants which is similar to minimized mutual coherency. Examples of such measurement matrixes are shown on FIG. 4.

[0072] FIG. 4 are examples of measurement matrixes of conventional solutions, where FIG.4 (a) is an example of random phases for fixed mapping HBF, and FIG. 4 (b) is an example of uniformly random antenna sampling one antenna per RF chain.

[0073] Using prior art measurement matrixes for compressed sensing provides near to optimal solution for a given sparsity scenario where no additional prior knowledge about signal sub-space is assumed. We will call such measurement matrixes conventional or compressed sensing (CS) prior art such measurement matrixes are optimal and easy solution in absence of any additional prior knowledge about signal subspace. Such matrixes can be once stored at UE side by using some kind of quasi random realization or random seed which is suitable for measurement in general case of compressed sensing (CS) problem. But such random measurement matrixes become rather non-optimal solution if any prior knowledge about signal sub-space exists. For example it is clear that random phase sub-array provides no beamforming gain as some of complex exponents in average will give zero if uniform random angles will be used for every term So additional optimization of themeasurement matrix becomes cell specific and optimal measurement matrix FRFbecomes unique for every specific cell deployment scenario. Therefore, in embodiments of the present application, it is necessary to send such information to UE in order to solve CS problem. In additional one can notice that total measurement matrix FRFdepends on mapping matrix M, so in proposed solution we claim that mapping matrix optimization significantly improves sensing matrix and channel estimation quality in case of prior knowledge assumption. In fact, practical implementation of dynamic mapping matrix is very complex hardware solution due to multiple switch network for reconfiguring of physical antenna subset connected to every RF chain. An embodiment of the present application relates to specific architecture of PC-HBF, and provides specific way to generate pilots in DE to virtually achieve reconfigurable mapping structure of mapping matrix M and total measurement matrix FRFspecific way to send measurement matrix to UE (share sensing matrix), which will be further described in the following.

[0074] As what was said previously, most of prior art solutions are related to pseudo sequences or random beams with random phase distribution which are used to obtain good sensing matrixes, such solutions may be good for general compressed sensing formulation where there is any prior knowledge about channel, so pseudo random antenna sampling patterncan be generated in advance at UE and BS sides and shared only once, and such solutions will work sub-optimally if channel have some structure (incomplete dictionary case)

[0075] In fact, main disadvantage of prior art solution is that conventional measurement matrixes like random antenna sampling, random phases beamforming for evert sub-array and random beam sweeping are not optimal for cell specific scenario, for UE group specific scenario, where prior knowledge like incomplete dictionary structure can exist due to available of longterm statistics, AoA / DoA to main scatters, geometrical channel model or historical measurements. In case of such knowledge additional optimization of measurement matrix should be done in order to minimize mutual coherence and additional sharing of measurement matrix from BS to UE should be done to make compressed channel estimation on UE side and make compressed report (only sparse coefficients and their indexes in vector v to theBS.

[0076] Main idea of the present application is to provide a new method of compressed measurements which comprises: 1) optimized structure of ABF measurement matrix for every round by irregular mapping matrix which maps physical antennas with RE chain on every training round to minimize mutual coherence; 2) method for achieving irregular mapping matrix using analog pilot generation with time modulated phase shifters according to waveform law on every physical antenna which belong to optimized group of antennas (indexes of ones in j- th column of M describes antenna sub-set which belongs to j-th RF chain or j-th effective subarray). Mapping optimization can be done by two ways: 1) by physical connection optimization using RF switch network; 2) by modulating carrier frequency on every phase shifter independently to achieve good autocorrelation properties of analog pilots and to provide artificial pilots from specific sub-set of antennas which provides irregular sub-array structure optimal for compressed channel estimation. Detailed description is given below.

[0077] FIG. 5 is a schematic flow chart of a method (200) for communication according to an embodiment of the present application.

[0078] At step 210, a network device transmits a configuration of a sensing matrix and a set of indexes of columns to be used in a dictionary.

[0079] The sensing matrix is determined according to a set of T0analog precoders and achannel structure descried with the dictionary. Assuming that the T0analog precoders include a first analog precoder, the first analog precoder is determined according to a first mapping matrix and a phase of every column of the first analog precoder. The first analog precoder is any one of the T0analog precoders. The first mapping matrix is used to indicate a connection relationship between physical antennas and NRFRF chains, is the number ofphysical antennas, and NRFis the number of RF chains, and T0is the number of beam training rounds.

[0080] Correspondingly, the UE obtains the sensing matrix and the set of indexes of the columns to be used in the dictionary.

[0081] At step 220, the UE transmits a compressed measurement report based on the sensing matrix and the set of indexes of the columns to be used in the dictionary.

[0082] The UE forms the compressed measurement report by measuring pilots (for example,CSI-RS) from the BS using the sensing matrix and the set of indexes of the columns. Then, theUE transmits the compressed measurement report to the BS. For examples, the compressed measurement report could be a wide band PMI, or is sent to the BS based on a codebook type¬II, more accurately beam-tap domain coefficients.

[0083] The BS receives the compressed measurement report, and estimates an optimized analog precoder for data transmission by solving some optimization problem.

[0084] Optionally, before the BS transmits the sensing matrix and the set of indexes of the columns to be used in the dictionary, the method may further include step 230.

[0085] At step 230, the BS determines the number of training rounds T0, the T0analog precoders and the set of indexes of the columns to be used in the dictionary, for example, the set of indexes may be determined according to prior knowledge about the channel. The number of training rounds and the T0analog precoders may be determined according feedback of the latest received power of SSB from the BS per DFT analog beam. The number of training rounds and the sensing matrix determined by the BS are optimal for learnt sub-space of whichwill give more details in the following embodiments.

[0086] Further, the method may include step 240.

[0087] At step 240, the UE solves an OMP-like problem using the sensing matrix and theset of indexes of the columns to be use in the dictionary to obtain a sparse matrix based on CSI-RS form the BS.

[0088] In the proposed solution, in order to solve the OMP-like problem at the UE side, the sensing matrix and the set of indexes of the columns to be used in the dictionary should be known at the UE side in order to recover sparse channel coefficients. Therefore, the BS shares this information (i.e., the sensing matrix and the set of indexes of the columns to be used) with the UE. The set of indexes of the columns to be used at the UE side is prior knowledge about the channel or the signal subspace. Based on the proposed solution, overhead of beam training in time domain can be reduced because of where T0is the number of beam trainingrounds in the proposed solution, and is the number of beam training rounds in theconventional solution. The convention solution is described previously, and it can refer to FIG.2.

[0089] More details of the proposed solution will be described in the following embodiments.

[0090] 1) Channel estimation procedure

[0091] FIG. 6 is an example of a channel estimation procedure according to an embodiment of the present application.

[0092] At step 1, a BS sends a synchro signal block (SSB) burst to roughly measure a link budget and make frame synchronization.

[0093] Before the BS implements step 1, the BS collects long term statistic by historical measurements and accumulation measurement data, and this is a process of learning signal sub- space in beam domain by the BS. The BS decomposes all data in some basis A tx asma ll numberof columns to achieve a set of indexes10094] At step 2, a UE sends back received power, for example, RSRP, per DFT analog beam.

[0095] At step 3, the BS optimizes the number of training rounds and measurement matrixes both by quantized phasesand mapping matrixes [ which are optimal for learnt sub-space Atxandminimizes coherence arg

[0096] At step 4, the BS shares optimized measurement matrixes as well as a configuration of the measurement matrix with the UE

[0097] The configuration of the measurement matrix includes: mapping matrixes quantized phases of every column of analog matrixesand the set of indexes

[0098] At step 5, the BS sends CSI-RS pilots precoded with an irregular analog matrix in DL using T0symbols, where The irregular analogmatrix is achieved using the analog pilot procedure by phase modulation on every physical antenna, which will be further described in the following embodiments.

[0099] Correspondingly, the UE receives the CSI-RS and solves the OMP-like problem to recover sparse channel coefficients.

[0100] At step 6, the UE generates a compressed report, , that includeswideband PMI and / or codebook type-II feedback.

[0101] Correspondingly, the BS receives the compressed report.

[0102] At step 7, using the compressed report, the BS estimates an analog precoder for data transmission by solving some optimization problem, for example,solves the minimization problem.

[0103] At step 8, the BS allocates resources for SRS and sends a configuration of the SRS to chosen UEs.

[0104] At step 9, during a small number of symbols, the BS receives SRS in UL from UEs using an optimized analog matrixand estimates a baseband precoder based on effective channels where is the baseband precoder,represents an effective channel in the z-th training round.

[0105] Note that, step 8 and step 9 are optional. These two steps can be done for better accuracy of the effective channels. That is to say,

[0106] At step 10, the BS transmits data with a total precoder that is equal to

[0107] It can be seen according to the channel estimation procedure shown in FIG. 6 that,the present application proposes a 2-stage channel estimation approach, that is, a CSI-RS stage and an optional SRS stage. The BS sends CSI-RS in DL at the first stage (i.e., a CSI-RS stage). After the CSI-RS stage, the UE sends SRS in UL during a few symbols, for example, one or two symbols, while the BS already sets an analog combiner optimized based on the CSI-RS stage in order to achieve a good baseband precoder.

[0108] 2) Irregular mapping for ABF.

[0109] To optimize a sensing matrix 5, an irregular mapping matrix M, for every round is proposed. The ABF needs to be optimized not only in terms of phases but also the mapping matrix:

[0110] where different shapes of the mapping matrix are shown in FIG. 7.

[0111] FIG. 7 are examples of different shapes of a mapping matrix. Structures of ABF measurement matrixes provided by embodiments of the present application are listed as: (a) regular fixed sub-arrays; (b) an optimized irregular analog matrix; (c)a regular analog matrix with overlapped columns; and (d) an optimized irregular analog matrix with overlapping.

[0112] Every column ofmatrix represents a RF chain and every row represents a physical antenna element which is connected to a given RF chain. In FIG. 7, it is seen that, for example, for the matrix (a), some physical antennas belong to first column. The structure of such column can be regular, which means that the physical sub-array has a regular structure, but for the channel estimation such regularity can have drawbacks, so structures (b) and (d) are better ones.

[0113] Simulation results show that the measurement matrix FRFwith the optimized mapping matrix (for example, the structures (b) and (d)) for every training round provides significant improvement of the sensing matrix. Such conclusion can be demonstrated in FIG. 8 by comparison of the conventional Gram matrix and the new proposed measurement matrix.For the new proposed measurement matrix, a total Gram matrix after all rounds is defined as a sum of Gram matrixes of all rounds:

[0114] FIG. 8 illustrates Gram comparison of proposed mapping optimization (left) and aconventional solution (right). From FIG. 8 it is seen that optimized mapping provides smaller values of non-diagonal elements of the total sum Gram matrix, which corresponds to smaller mutual coherence of a total sensing matrix

[0115] Optimization of phases and irregular mappings jointly provides a better sensing matrix S in terms of non-diagonal elements of the Gram matrix and better diagonal elements, which provides smaller mutual coherence and higher effective SNR during compressed measurements, as measurement beams provide higher correlation with dictionary words (i.e., diagonal elements of the Gram matrix). The proposed method can be applied in time division duplexing (TDD) in DL and UL pilots, and also can be applied in frequency division duplexing(FDD).

[0116] FIG. 9 demonstrates performance of compressed channel estimation for the case of APM, FIX and a random phase measurement matrix as well as a cumulative distribution function for diagonal and non-diagonal elements of the Gram matrix. An optimized sensing matrix provides improvement in correlation of a vectorized estimated channel and a true channel.

[0117] 3) Sensing matrix feedback

[0118] As what has been described in previous embodiments, before sending CSI-RS, theBS shares indexes of columns to be used, n the dictionaryand the sensing matrix S. Sharing the compressed sensing matrix S directly can have large overhead, so the present application proposes sending of quantized phases of every column of each of the analog precoders, mapping matrixesand indexes of columns to be used, The process of sharing measurementmatrixes is shown in FIG. 10.

[0119] The proposed solution reduces training overhead. Here is an example of overhead calculation.

[0120] Assuming that the number of BS antennas istx, the number of RF chains is NRF= 32, the number of columns in the dictionary is and the number ofrounds is T0= 4.

[0121] An example of sensing matrix feedback calculation:32x200x4 complex numbers, and 4...8 bits for real and imaginary parts, so totally 2x(4...8) x32x200x4=204800...409600 bits for adaptive codebook sharing;200 natural numbers from 1 ... 1024, and 4...8 bits for real and imaginary parts, so totally 2x(4...8) x32x200x4=204800...409600 bits for adaptive codebook sharing;

[0122] Therefore, total overhead is: 204800...411600 bits for the sensing matrix and dictionary.

[0123] An example of RF matrix feedback calculation:32 X 200 X 4 natural numbers, and 4...5 bits for a phase shifter, so(4...5) x32x 32x4=16384...20480 bits for adaptive codebook sharing using a sub-array phase;32x32x4 natural numbers from 1 ... 1024 to describe a mapping pattern for every sub-array on every round, so 32x32x4x10=40960 bits for mapping sharing;200 natural numbers from 1... 1024, and 4...8 bits for real and imaginary parts, so 200x10 bits for dictionary.

[0124] Therefore, total overhead is: 16384...20480+40960+2000=59344...63440 bits for long-term statistics and dictionary. The overhead of sensing matrix sharing will be reduced by

[0125] 4) Analog pilot generation

[0126] An irregular analog matrix can be achieved by huge hardware complexity, which means that a huge RF switch network is need. Physical realization of the irregular analog matrix can be done by forming a DL pilot directly on every physical antenna by phase modulation of a phase shifter during the symbol time.

[0127] Analog phase shifters cannot provide amplitude modulation and only can provide phase modulation of the signal, so the number of analog waveforms which can be created on the phase shifter is limited with CAZAC types of waveforms, for example, Hadamard, DFT, Zadoff-Chu. Examples of analog pilot generation are shown in FIG. 11.

[0128] The present application provides the following methods for analog pilot generation: a) generating only a carrier grid with some carrier spacing on baseband, and using the same carrier grid for every RF chain; b) performing modulation on every phase shifter (PS) such that antennas withinone group should generate the same phase modulation law; and c) adding constant phase shifts achieved from optimization of the Gram matrix to every group (which corresponds to artificial RF chain).

[0129] FIG. 11 illustrates examples of analog pilot generation. From FIG. 11 , it is seen that if a pilot is generated on every PS, we can effectively achieve an irregular sub-array structure corresponding to some optimal mapping matrix. Quality of measurement pilots after analog pilot generation depends on the RF phase shifter switching speed as shown in FIG. 12. It is seen that if the switching speed is 25ns, sidelobes of the linear modulated signal spectrum reach up to -12dB, which is far from good. But with a higher switching speed, more accurate spectrum, closer to ideal can be achieved. Comparison of two different pilot signals shows that even if large spurs in the spectrum exist, signals still can have good autocorrelation properties that can distinguish from each other at the UE side.

[0130] FIG. 12 illustrates a switching speed, a pilot spectrum, and an autocorrelation function of an analog pilot.

[0131] 5) Overlapped sub-arrays

[0132] As a practical embodiment of the invention, the new type of HBF architecture with multiple phase shifters per antenna can be considered. Additional improvement of a Gram matrix and a sensing matrix can be achieved if more phase shifters per antenna will be used. As it was said, such architecture is called overlapped sub-arrays. If the APM-HBF technique is applied to overlapped sub-arrays, additional improvement can be achieved.

[0133] Besides, since practical implementation of overlapped sub-arrays and dynamic overlapped sub-arrays is extremely difficult because of a huge switch network, the present application proposes an architecture which can implement dynamic overlapped sub-arrays with small complexity because the proposed solution only increases the number of PS per antenna but does not add a switch network.

[0134] The present application proposes implementation of dynamic overlapping by generating waveform directly by every PS by applying phase modulation during the symbol.Because mapping adaption with overlapped sub-arrays is even more difficult in realization from hardware point of view, analog pilot generation methods for overlapped sub-arrays is a very attractive solution. In order to be able to make overlapping as shown in FIG. 13, several PS perantenna should be connected and also an additional power amplifier should be installed in order to compensate losses from combining, but a power divider can still have a FIX structure and can be passive. It means that from power divider point of view, complexity is similar to PC-HBF. The proposed new architecture is shown in FIG. 13.

[0135] FIG. 13 illustrates overlapped sub-arrays with multiple phase shifters and a simple power divider according to an example of the present application. Additional improvement of a Gram matrix and a sensing matrix can be achieved if more phase shifters per antennas will be used as it was said that such architecture is called overlapped sub-arrays. If the APM HBF technique is applied to overlapped sub-arrays, benefits can be obtained as follows: sub-arrays with overlapping adapted mapping (APM) provide a better shape of sensing patterns, which means smaller gain loss and smaller mutual coherence; overlapping is provided without a need of an additional power divider, that is, the overlapped sub-arrays can be created just by analog pilots artificially and mathematically; and performance of the overlapped sub-arrays is close to that of the FC-HBF compressed sensing (CS) optimization.

[0136] If APM-HBF is applied to overlapped sub-arrays, additional phase shifters need to be placed for nearly every antenna. Besides, small power divider (PD) with a constant gain are needed to compensate for Wilkinson power divider losses.

[0137] In FIG. 13, means the number of phase shifters connected to each physicalantenna. The small PD means a small power divider, typically, Wilkinson power divider, or any other microwave power divider. Besides, a PS block contains multiple phase shifters. A gain block contains small combiner to combine signals from several phase shifters to single physical antenna and final power amplifier.

[0138] The method proposed in the embodiments of the present application is described in detail above, and a communication apparatus provided by the present application will be described in detail below.

[0139] FIG. 14 is a schematic block diagram of an apparatus 10 according to some embodiments of the present application. The apparatus may be a communication device or anapparatus implemented in a communication device and capable of realizing corresponding functions of any one of the embodiments of the present application. For example, the apparatus implemented in a communication device may be an integrated circuit, which in some contexts may be known by other colloquial names, such as chip, modem, modem chip, baseband chip, or baseband processor. In some implementations, one or more integrated circuits can be packaged into a system-on-chip, a system-in-package, or a multi-chip module. The apparatus may comprise one or more integrated circuits or comprise one or more integrated circuits and other discrete components. The communication device may be the signal transmitter, the signal receiver, or the apparatus implemented in any one of these communication devices.

[0140] The apparatus 10 includes a processing module 1001. The processing module 1001 may be a processor, a processing circuit, a processing board, a processing unit, or a processing device, et al. The processing module 1001 is configured to implement processing and / or operations implemented inside the communication apparatus except sending or receiving actions.

[0141] The apparatus 10 may further include a communication module 1002. The communication unit 1002 is configured to implement a sending action and / or a receiving action.The communication module 1002 also may be called as a transceiver module, a transceiver, or a transceiver device, et al, and is configured to implement operations of receiving (which may be referred to as inputting) and / or sending (which may be referred to as outputting).

[0142] For example, if the apparatus 10 corresponds to the network device in FIG. 5, the communication module 1002 is configured to implement a sending action at step 210 and a receiving action at step 220. The processing module 1001 is configured to implement step 230. If the apparatus 10 corresponds to the terminal device in FIG. 5, the communication module1002 is configured to implement receiving action at step 210 and a sending action at step 220.The processing module 1001 is configured to implement step 240.

[0143] Briefly, the operations and / or functions of the apparatus 10 are intended to implement corresponding steps of the foregoing method embodiments.

[0144] FIG. 15 is a schematic block diagram of an apparatus according to some embodiments of the present application. The apparatus 20 includes at least one processor 21. The at least one processor 21 is coupled to at least one memory 22. The at least one memory 22is configured to store one or more instructions and / or executable computer code. The at least one processor 21 is configured to invoke the one or more instructions and / or executable computer code, so that the apparatus 20 implements the method provided in the embodiments of the present application. Optionally, the apparatus 20 may further include the at least one memory 22. Optionally, the apparatus 20 may further include at least one communication interface 23, and the at least one communication interface 23 is configured to input and / or output information or data.

[0145] In an implementation, the apparatus 20 may be any one of the communication devices in the method embodiments. For example, the apparatus 20 may be the network device or the terminal device. In this implementation, the processor 21 may be a baseband apparatus, and the communication interface 23 may be a radio frequency apparatus.

[0146] In another implementation, the apparatus 20 may be installed in a communication device such as the network device and the terminal device. In this case, the apparatus may be an integrated circuit, which in some contexts may be known by other colloquial names, such as chip, modem, modem chip, baseband chip, or baseband processor. In some implementations, one or more integrated circuits can be packaged into a system-on-chip, a system-in-package, or a multi-chip module. The apparatus may comprise one or more integrated circuits or comprise one or more integrated circuits and other discrete components. In this implementation, the processor 21 may be a logical module or circuit that is part of the integrated circuit. The communication interface 23 may be a transceiver, an interface circuit, an input / output interface, a bus, a module, a pin, or other types of interfaces.

[0147] An embodiment of the present application further provides a communication system.The communication system may include a transmitting device and a receiving device, for example, a network device and a terminal device as shown in FIG. 5.

[0148] An embodiment of the present application further provides a computer storage medium, and the computer storage medium may store one or more instructions for executing any of the foregoing methods.

[0149] An embodiment of the present application further provides a computer program product, and the computer program product may store one or more instructions for executing any of the foregoing methods.

[0150] In the embodiments of this application, “and / or” describes an association relationship between associated objects and represents that three relationships may exist. For example, A and / or B may represent the following three cases: Only A exists, both A and B exist, and only B exists. The character “ / ” generally indicates an “or” relationship between the associated objects. “At least one” means one or more. “At least one of A and B”, similar to “A and / or B”, describes an association relationship between associated objects and represents that three relationships may exist. For example, at least one of A and B may represent the following three cases: Only A exists, both A and B exist, and only B exists.

[0151] Besides, the use of a singular form of “a”, “an” and “the” in the embodiments of the present application and the claims appended hereto is also intended to include a plural form, unless otherwise clearly indicated herein by context.

[0152] A person of ordinary skill in the art will be aware that, in combination with the examples described in the embodiments disclosed in this specification, units and algorithm steps may be implemented by using electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed by using hardware or software depends on particular applications and design constraint conditions of the technical solutions.A person skilled in the art may use different methods to implement the described functions for each particular application, but it should not be considered that the embodiment goes beyond the scope of this application.

[0153] It would be understood by a person skilled in the art that, for the purpose of convenience and brevity, in a detailed working process of the foregoing system, apparatus, and unit, reference may be made to a corresponding process in the foregoing method embodiments, and details are not described herein again.

[0154] In the several embodiments provided in this application, the disclosed system, apparatus, and method may be implemented in other manners. For example, the described apparatus embodiment is merely an example. For example, the unit division is a logical function division and other methods of division may be used in an actual embodiment. For example, a plurality of units or components may be combined or integrated into another system, or some features may be ignored or not performed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections may be implemented using variouscommunication interfaces. The indirect couplings or communication connections between the apparatuses or units may be implemented in electronic, mechanical, or other forms.

[0155] In addition, function units in the embodiments of this application may be integrated into one processing unit, each of the units may exist alone physically, or two or more units may be integrated into one unit.

[0156] When the functions are implemented in the form of a software functional unit and sold or used as an independent product, the functions may be stored in a computer-readable storage medium. The technical solutions of this application may be implemented in the form of a software product. The software product is stored in a storage medium, and includes several instructions for instructing a computer device (which may be a personal computer, a server, a network device, or the like) to perform all or some of the steps of the methods described in the embodiments of this application. The foregoing storage medium includes any medium that can store program code, such as a USB flash drive, a removable hard disk, a ROM, a RAM, a magnetic disk, an optical disc or the like.

[0157] The units described as separate parts may be or may not be physically separate, and parts displayed as units may be or may not be physical units, may be located in one position, or may be distributed on a plurality of network units. Some or all of the units may be selected based on actual requirements to achieve the objectives of the solutions of the embodiments. In addition, functional units in the embodiments of this application may be integrated into one processing unit, or each of the units may exist alone physically, or two or more units are integrated into one unit.

[0158] The foregoing descriptions are merely specific implementations of this application, but are not intended to limit the protection scope of this application. Any variation or replacement readily figured out by a person skilled in the art within the technical scope disclosed in this application shall fall within the protection scope of this application. Therefore, the protection scope of this application shall be subject to the protection scope of the claims.

Claims

CLAIMS1. A method for communication, comprising: receiving a configuration of a sensing matrix and a set of indexes of columns to be used in a dictionary, wherein the sensing matrix is determined according to a set of T0analog precoders and a channel structure descried with the dictionary, a first analog precoder is determined according to a first mapping matrix and a phase of every column of the first analog precoder, the first analog precoder is any one of the T0analog precoders, the first mapping matrix is used to indicate a connection relationship between physical antennas and NRFradio frequency (RF) chains, is the number of physical antennas, and NRFis thenumber of RF chains, and T0is the number of beam training rounds and is a positive integer; and transmitting a compressed measurement report based on the sensing matrix and the set of indexes of the columns to be used in the dictionary.

2. The method according to claim 1, wherein receiving the configuration of the sensing matrix further comprises: receiving T0mapping matrixes corresponding to the T0analog precoders respectively and quantized phases of every column of each of the T0analog precoders.

3. The method according to claim 1 or 2, wherein the method further comprises: receiving a channel state information (CSI)-reference signal (RS) on T0time units, wherein the CSI-RS on the T0time units are precoded by the T0analog precoders respectively; obtaining a sparse matrix based on the CSI-RS on the T0time units, the sensing matrix and the set of indexes of the columns to be used in the dictionary.

4. The method according to claim 2 or 3, wherein each of the T0mapping matrixes is irregular, each irregular mapping matrix is obtained by performing analog pilot generation and phase modulation on every physical antenna.

5. The method according to claim 4, wherein the analog pilot generation comprises any one of:generating only a carrier grid with some carrier spacing on baseband, and using a same carrier grid for every RF chain; performing modulation for every phase shifter such that antennas within one group generate the same phase modulation law; and adding constant phase shifts in time domain achieved from optimization of a Gram matrix to every antenna group.

6. The method according to any one of claims 2 to 5, wherein each antenna is connected with more than one phase shifter and a power amplifier (PA) such that signals from every phase shifter is combined via a power divider (PD).

7. The method according to any one of claims 1 to 6, wherein the method further comprises: receiving a configuration of a sounding reference signal (SRS), wherein the configuration of the SRS comprises time resources for the SRS, and a time resource comprises a few time units; and transmitting the SRS on the few time units, wherein the SRS on the few time units are used for effective channel estimation, and an effective channel is used to determine a baseband precoder.

8. The method according to claim 7, wherein the number of time units for the SRS is one or two.

9. The method according to any one of claims 3, 7 and 8, wherein a time unit is a symbol.

10. A method for communication comprising: transmitting a configuration of a sensing matrix and a set of indexes of columns to be used in a dictionary, wherein the sensing matrix is determined according to a set of T0analog precoders and a channel structure descried with the dictionary, a first analog precoder is determined according to a first mapping matrix and a phase of ever column of the first analog precoder, the first analog precoder is any one of the T0analog precoders, the first mapping matrix is used to indicate a connection relationship between physical antennas and NRFradio frequency (RF) chains, is the number of physical antennas, is the numberof RF chains, and T0is the number of beam training rounds and is a positive integer; and receiving a compressed measurement report obtained based on the sensing matrix and theset of indexes of the columns to be used in the dictionary.

11. The method according to claim 10, wherein transmitting the configuration of the sensing matrix further comprises: transmitting T0mapping matrixes corresponding to the T0analog precoders respectively and quantized phases of every column of each of the T0analog precoders.

12. The method according to claim 10 or 11, wherein the method further comprises: transmitting a CSI-RS on T0time units, wherein the CSI-RS on the T0time units are precoded by the T0analog precoders respectively.

13. The method according to claim 12, wherein the compressed measurement report is obtained based on the CSI-RS on the T0time units, the sensing matrix and the set of indexes of the columns to be used in the dictionary; and the method further comprises obtaining an optimized analog precoder for data transmission according to the compressed measurement report by solving an optimization problem.

14. The method according to any one of claims 11 to 13, wherein each of the T0mapping matrixes is irregular, each irregular mapping matrix is obtained by performing analog pilot generation and phase modulation on every physical antenna.

15. The method according to claim 14, wherein the analog pilot generation comprises any one of: generating only a carrier grid with some carrier spacing on baseband, and using a same carrier grid for every RF chain; performing modulation on every phase shifter such that antennas within one group generate the same phase modulation law; and adding constant phase shifts achieved from optimization of a Gram matrix to every antenna group.

16. The method according to any one of claims 11 to 13, wherein each antenna is connected with more than one phase shifter and a power amplifier (PA) such that signals from every phase shifter is combined via a power divider (PD).

17. The method according to any one of claims 13 to 16, wherein after receiving the compressed measurement report, the method further comprises:transmitting a configuration of sounding reference signal (SRS), wherein the configuration of the SRS comprises time resources for the SRS, and a time resource comprises a few time units; receiving the SRS on the few time units using the optimized analog precoder; and obtaining a baseband precoder for data transmission based on an effective channel that is calculated based on the SRS.

18. The method according to claim 17, wherein the number of time units for the SRS is one or two.

19. The method according to any one of claims 12, 17 and 18, wherein a time unit is a symbol.

20. The method according to any one of claims 17 to 19, wherein the method further comprises: determining a final matrix for data transmission, wherein the final matrix is related to the optimized analog precoder and the baseband precoder; and transmitting data based on the final matrix.2 1. A communication apparatus, comprising a processor configured to enable the apparatus to perform the method according to any one of claims 1 to 9 or any one of claims 10 to 20.

22. The communication apparatus according to claim 21, further comprising a memory for storing instructions to be executed by the processor.

23. The communication apparatus according to claim 21 or 22, further comprising an interface configured to input and / or output information or a signal.

24. A communication system, comprising: a communication apparatus that performs the method according to any one of claims 1 to 9; and a communication apparatus that performs the method according to any one of claims 10 to 20.

25. A computer readable storage medium, comprising one or more instructions, wherein when the one or more instructions are run on a computer, the computer performs the method according to any one of claims Ito 9 or any one of claims 10 to 20.

26. A computer program product, comprising one or more instructions, wherein when the one or more instructions are run on a computer, the computer performs the method according to any one of claims 1 to 9 or any one of claims 10 to 20.

Images