Horizontal and vertical beamforming design for irs-assisted communications

Abstract

A method of operating a repeater (120) having a two-dimensional array (121) of emitter elements (122) with controllable phase shifts, wherein the method is performed by a network node (110) and comprises: obtaining a channel estimate of a channel between the network node and a wireless device via the repeater; maximizing, based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmitting data between the network node and the wireless device via the repeater while applying a result of said maximization, wherein the maximization of the received signal power comprises a first sub-maximization, in which a horizontal beamforming setting is varied, and an independent second sub-maximization, in which a vertical beamforming setting is varied, wherein the horizontal and the vertical beamforming setting include, respectively, a horizontal and a vertical evolution of the phase shifts of the repeater.

Description

HORIZONTAL AND VERTICAL BEAMFORMING DESIGN FOR IRS-ASSISTED COMMUNICATIONSTECHNICAL FIELD

[0001] The present disclosure is related to the field of cellular communications networks. In particular, it proposes a method by which a network node operates a repeater to provide a two-hop radio link. The repeater, an intelligent reflective surface (IRS), network-controlled repeater (NCR) or nodes with similar functionalities are equipped with a two-dimensional array of emitter elements with controllable phase shifts.BACKGROUND

[0002] An intelligent reflecting surface (IRS), also known as reconfigurable intelligent surface (RIS), is an emerging technology that is capable of intelligently manipulating the propagation of electro-magnetic waves. IRS is composed of a two-dimensional array of reflecting elements, where each element acts as a passive reconfigurable scatterer, i.e., a piece of manufactured material, which can be programmed to change an impinging electro-magnetic wave in a customized way. Such elements are usually low-cost passive surfaces that do not require dedicated power sources, and the radio waves impinged upon them can be forwarded without the need of employing power amplifier or RF chain. Moreover, IRS can, potentially, work in full duplex mode without significant self-interference or increased noise level and requires only low-rate control link or backhaul connections. IRS can be flexibly deployed due to its low weight and low power consumption. Specially, IRS is of interest in stationary or low-mobility networks, in which the transmission parameters can be well planned and, e.g., blockages / tree foliage is bypassed through IRS-assisted communication.

[0003] In 3GPP, IRS-assisted communication was suggested by some companies as a possible technology to be considered in the Release-18 network-controlled repeater (NCR) study-item. For instance, IRS has been discussed at the 3GPP TSG RAN Release-18 workshop, June 2021; see for instance RWS-210300, "NR repeaters and Reconfigurable Intelligent Surface”. However, it has been decided not to include the IRS in the 3GPP NCR study-item and leave it for possible discussions in the coming 3GPP releases. Therefore, while specification wise an NCR is likely to be a superset of the IRS, from an invention perspective, the IRS-specific ideas may be well applicable to the NCRs as well.

[0004] Beamforming design in IRS-assisted networks is a challenging task and efficient solutions for this problem are key factors for the current and next generations of wireless communication networks. In this context, the gNodeB (gNB) receives reference signals and / or feedback from user equipment (UE) devices to determine channel state information (CSI) of all channels (separately from each other or coupled). Beamforming by the gNB is then based on the determined CSI. Millimeter-wave (mmWave) and sub-terahertz (THz) communications facilitate a large number of antenna elements at network nodes, and, in addition, the IRS can also be composed of hundreds to thousands of reflecting elements. This large number of antenna and reflecting elements can significantly increase the overhead and computational complexity of the active (precoder and combiner) andpassive (IRS phase-shift design) beamforming optimization, at the gNB / UE and the IRS, respectively. The large computational complexity may ultimately increase the latency and / or the feedback and reference signal transmissions may require significant spectral resources. The increasing computational complexity and the required spectrum resources for CSI acquisition may degrade the overall spectral and energy efficiency and latency of the IRS-assisted systems.SUMMARY

[0005] One objective of the present disclosure is to make available a computationally efficient method for operating a repeater, one which has good scalability with respect to the number of controllable phase shift settings. A further objective is to make such a method available that additionally allows the operating of beamforming components in the network node or the UE, or both. In particular, it is desirable to operate a repeater in the form of an IRS or an NCR. A still further objective is to provide devices and software with these advantages.

[0006] At least some of these objectives are achieved by the invention as defined by the independent claims. The dependent claims relate to embodiments of the invention.

[0007] In a first aspect of the present disclosure, there is provided a method of operating a repeater (in particular, the method may include controlling the repeater) for a cellular communications network. The repeater is assumed to have a two-dimensional array of emitter elements with controllable phase shifts. The method is performed by a network node authorized to control these phase shifts, and it comprises at least the following steps: obtaining a channel estimate of a channel between the network node and a wireless device via the repeater; maximizing, based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmitting data between the network node and the wireless device via the repeater while applying a result of said maximization. According to the first aspect, the maximization of the received signal power comprises a first sub-maximization, in which a horizontal beamforming setting is varied, and an independent second sub-maximization, in which a vertical beamforming setting is varied. The horizontal and the vertical beamforming setting include, respectively, a horizontal and a vertical evolution of the phase shifts of the repeater. In the method, the first sub-maximization can be performed before or after the second submaximization, or the performance of the two sub-maximizations may overlap in time. Preferably, the first submaximization is restricted to the horizontal beamforming setting (i.e., it does not include any action related to the vertical beamforming setting), and the second sub-maximization is restricted in the same sense to the vertical beamforming setting.

[0008] In a second aspect of the disclosure, there is provided a network node arranged to operate a repeater for a cellular communications network. The network node has processing circuitry configured to: obtain a channel estimate of a channel between the network node and a wireless device via the repeater; maximize, based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmit data between the network node and the wireless device via the repeater while applying aresult of said maximization. The maximization of the received signal power comprises a first sub-maximization, in which a horizontal beamforming setting is varied, and an independent second sub-maximization, in which a vertical beamforming setting is varied. The horizontal and the vertical beamforming setting include, respectively, a horizontal and a vertical evolution of the phase shifts of the repeater.

[0009] The present disclosure further relates to a computer program containing instructions for causing a computer, or the network node in particular, to carry out the above method. The computer program may be stored or distributed on a data carrier. As used herein, a "data carrier” may be a transitory data carrier, such as modulated electromagnetic or optical waves, or a non-transitory data carrier. Non-transitory data carriers include volatile and non-volatile memories, such as permanent and non-permanent storage media of magnetic, optical or solid-state type. Still within the scope of "data carrier”, such memories may be fixedly mounted or portable.

[0010] For the purposes of the present disclosure, a "wireless device” may be a smart phone, a mobile phone, a cell phone, a voice over IP (VoIP) phone, a wireless local loop phone, a desktop computer, a personal digital assistant (PDA), a wireless camera, a gaming console or device, a music storage device, a playback appliance, a wearable terminal device, a wireless endpoint, a mobile station, a tablet, a laptop, a laptop-embedded equipment (LEE), a laptop-mounted equipment (LME), a smart device, a wireless customer-premise equipment (CPE), a vehicle-mounted wireless terminal device and the like. A "network node” may be equipment capable, configured, arranged and / or operable to communicate directly or indirectly with a wireless device and / or with other network nodes or equipment in the wireless network to enable and / or provide wireless access to the wireless device and / or to perform other functions (e.g., administration) in the wireless network. Examples of network nodes include, but are not limited to, access points (APs), base stations (BSs) including radio base stations, evolved NodeBs (eNBs) and NR NodeBs (gNBs).

[0011] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a / an / the element, apparatus, component, means, step, etc.” are to be interpreted openly as referring to at least one instance of the element, apparatus, component, means, step, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

[0012] The characteristics of the method and the network node outlined above have been conceived based on the inventors' realization that the beamforming design problem can be split into more tractable vertical and horizontal sub-problems by exploiting the structure of uniform rectangular arrays (URA) at the network node, the repeater and / or the wireless device.

[0013] In one embodiment, when an IRS and a UE are connected to the network, they share their capabilities with a gNB informing about the number and / or the indexing of emitters or antenna elements along the horizontal and the vertical domains, etc. After estimating the channel at the gNB or receiving the estimated channel from the UE, as explained in the following, the gNB can exploit this structure and solve two independent optimization subproblems along horizontal and vertical directions, and the gNB determines the optimal precoder, combiner andIRS phase shifts. Finally, the gNB informs the IRS about the determined horizontal and vertical IRS phase shifts configurations and / or informs the UE about the determined horizontal and vertical combiners.

[0014] In one embodiment, the gNB solves the horizontal and vertical sub-problems with different optimization methods.

[0015] In another embodiment, depending on the channel variations or the received SNR variations along the azimuth and the elevation, the gNB can adapt different modulation order to transmit data along each domain. The data can be sent with higher modulation order in the domain which shows higher SNR and vice-versa.

[0016] In one embodiment, the gNB sends a control signal which determines the operation mode at the UE: i) Independent data transmission (i.e., horizontal and vertical domains can be used for the transmission of different data blocks) with the same modulation order along horizontal and vertical domains, ii) Independent data transmission with different modulation order along horizontal and vertical domains, iii) Data block is transmitted without exploiting the horizontal and vertical structure.This enables the decoupling of data transmission so that different modulation orders can be used depending on the quality and variations of the channels along different domains, or different SNRs along horizontal and vertical domains can be monitored whereby a more flexible system design and optimized data throughput become possible.

[0017] Further, by similarly exploiting the horizontal / vertical structure of the whole problem, it is possible to execute beam failure recovery (BFR) mechanisms in horizontal and vertical domains independently. Therefore, in the case of a signal quality drop (e.g., a drop in SNR) in one or both domains, it is possible to independently select the horizontal sub-beam, the vertical sub-beam or both of these, depending on the horizontal, the vertical or the total SNRs observed at the UE.

[0018] In another embodiment, the proposed scheme is applied for beam failure recovery, where independent backup beams are considered for the vertical and horizontal domains and, depending on the SNR observed in one or both of the domains, the beam(s) in one or both of the domains can be updated.

[0019] In another embodiment, the proposed idea is applied in a system, where different powers could be allocated independently to the horizontal and the vertical beamformers at the gNB, though possibly subject to a total power constraint. For example, if one of the domains shows better SNR compared to the other domain, more power could be allocated to the domain which shows higher SNR to have more flexible transmission (for instance adaptive modulation and coding along each domain independently).

[0020] In one embodiment, a network node performs a method comprising: obtaining a channel estimate of a channel between the network node and a wireless device (wherein the channel refers to a direct path or a pathvia a repeater), wherein the network node comprises a two-dimensional array of antenna elements with controllable antenna weights and / or the wireless device comprises a two-dimensional array of antenna elements with controllable antenna weights; maximizing, based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmitting data between the network node and the wireless device while applying a result of said maximization, wherein the maximization of the received signal power comprises a first sub-maximization, in which a horizontal beamforming setting including a horizontal evolution of the antenna weights of the network node and / or the wireless device is varied, and an independent second sub-maximization, in which a vertical beamforming setting including a vertical evolution of the antenna weights of the network node and / or the wireless device is varied. It is appreciated that "first” / ” second” does not necessarily refer to the succession of the sub-maximizations in time.

[0021] In a third aspect of the present disclosure, there is provided a method performed by a wireless device for a cellular communications network. The wireless device is assumed to have a two-dimensional array of antenna elements with controllable antenna weights. The method comprises: transmitting structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receiving configuration information from the network, which configuration information indicates a first factor and a second factor of a combiner to be used by the wireless device; and sending data to the network or receiving data from the network in accordance with the configuration information.

[0022] In a fourth aspect, there is provided a wireless device configured to perform this method.

[0023] In a fifth aspect, there is provided a method performed by a repeater (e.g., IRS, NCR) for a cellular communications network. The repeater is assumed to have a two-dimensional array of emitter elements with controllable phase shifts. The method comprises: transmitting structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receiving configuration information from the network, which configuration information indicates a first factor and a second factor of phase shifts to be used by the repeater; and facilitating a data transmission in the network in accordance with the configuration information.

[0024] In a sixth aspect, there is provided a repeater configured to perform this method.

[0025] The third through sixth aspects of the present disclosure facilitate the performing of the method according to the first aspect and / or support a network node according to the second aspect. More precisely, it is ensured that the network node is informed by the relevant capabilities of the wireless device and repeater, such as the horizontal and vertical size (or the number of rows and number of columns) of the two-dimensional arrays of these entities.

[0026] On a general level, the embodiments herein tend to reduce the computational complexity required for the beamforming design, by separating the beamforming problem into horizontal and vertical sub-problems and allowing parallelization.

[0027] By solving the optimization problem in horizontal and vertical domains separately, the gNB or other network node shares the optimal combiner and IRS / NCR phase shifts with the UE and the IRS / NCR, respectively, with less spectral resources. For example, in a scenario with a gNB and a UE having 16 antenna elements each and an IRS having 400 reflecting elements, the proposed solution has 1 / 40 of the computational complexity of the classical solution, and it requires about 1 / 9 of the spectral resources (to be detailed below).

[0028] By exploiting the horizontal and vertical structure of the gNB, the IRS / NCR and the UE, different codebook resolutions can be used along horizontal (azimuth) and vertical (elevation) domains.

[0029] It becomes possible to use different algorithms (e.g., different optimization solvers, or different instantiations of one solver) in respect of the independent sub-problems along horizontal and vertical domains to find beamforming and optimal IRS / NCR phase-shifts.

[0030] The proposed method can also exploit the variations of the channel along horizontal (azimuth) and vertical (elevation) domains in order to have different periodicities for active and passive beamforming designs along each domain.

[0031] The SNR can be independently observed with respect to the horizontal and vertical domains and, thereby, the gNB or other network node can be configured to transmit with different modulation order and / or coding for each domain.

[0032] The signaling associated with the reporting of the optimal combiners and IRS phase shifts along vertical and horizontal domains may have different periodicity depending on the SNR fluctuation in each domain, which adds to the flexibility of the system operation.

[0033] The beam failure recovery mechanism can be adapted in the vertical and horizontal domains separately where, depending on the SNR observed in different domains, the horizontal or the vertical beam, or both, are adapted.

[0034] In a real cellular communication, the gNB may have a larger number of antennas deployed in the vertical domain as compared to the horizontal domain due to sectorizing concept of a cell. The scheme proposed herein makes it possible to apply a codebook with a higher resolution (oversampled backup beams) in the vertical domain than in the horizontal domain.

[0035] Different transmit powers could be allocated along horizontal and vertical beamformers / precoders.

[0036] Finally, as explained in the following, the proposed method may be applicable to different massive-MIMO or line-of-sight MIMO networks as well, that is, without the repeater as a necessary component.BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Aspects and embodiments are now described, by way of example, with reference to the accompanying drawings, on which:figure 1 is system model of the radio links connecting the wireless device, repeater and network node; figure 2 is a flowchart of a method for operating a repeater, according to embodiments herein; figure 3 depicts a network node in the form of a base station with two-dimensional antenna arrays; figure 4 illustrates a data flow during an execution of the method illustrated in figure 2; figure 5 is a comparison of a classical beam failure recovery (BFR) design and a novel BFR according to embodiments herein; figure 6 is a plot of an empirically determined cumulative distribution function versus angular variation; figure 7 is a plot of computational complexity versus the number of reflecting elements in an IRS; figure 8 is a plot of spectral efficiency versus signal-to-noise ratio (SNR); and figure 9 is a sequence diagram illustrating an embodiment disclosed herein.DETAILED DESCRIPTION

[0038] The aspects of the present disclosure will now be described more fully hereinafter with reference to the accompanying drawings, on which certain embodiments of the invention are shown. These aspects may, however, be embodied in many different forms and should not be construed as limiting; rather, these embodiments are provided by way of example so that this disclosure will be thorough and complete, and to fully convey the scope of all aspects of the invention to those skilled in the art. Like numbers refer to like elements throughout the description.

[0039] In current wireless communication systems, the design of active and passive beamforming for the gNB / UE and the IRS does not exploit the geometry of the transmit and the receive antenna arrays as well as the structure of the IRS. In addition, the current state-of-the art solutions do not exploit the fact, realized by the present inventors, that the channel has a separable structure in active and passive beamforming designs. For example, a precoding codebook is used based on the precoding matrix indicator (PMI) to enhance the system performance in 3GPP, which does not utilize the geometry of the transmit and receive arrays of the base station and the IRS.

[0040] As the number of antennas and / or the IRS elements increases, the cost and complexity of active and passive beamforming, and the IRS's corresponding signaling cost increases. Such a problem affects both the active / passive beamforming design at the IRS as well as the beam failure recovery process. This may affect the overall efficiency of IRS-assisted networks. Therefore, this disclosure addresses the need for techniques that reduce the cost and / or complexity of the active and passive beamforming design and its corresponding signaling in IRS-assisted systems.

[0041] The 3D beamforming design specified by the 3GPP is based on a predefined codebook (see Technical Specification (TS) 38.211 "NR; Physical channels and modulation” and TS 38.214 "NR; Physical layer procedures for data” within Release 17). Nor does the beam failure recovery (BFR) mechanism exploit the separablegeometrical structure to split the backup-beams problem into horizontal and vertical sub-beams. If the SNR falls below or the BLER exceeds a certain threshold at the UE, then according to the classical approach, the next beam from a set of pre-determined backup beams will be selected.

[0042] Through the present disclosure, the inventors propose a beamforming design strategy that considers the geometry of the uniform rectangular array (URA), at both the transmitter and the receiver, as well as the separable structure of the IRS to split the beamforming design problem into horizontal and vertical domains subproblems with reduced dimensions. This way, the beamforming design determines the active and passive beamforming at the gNB / UE and the IRS, respectively. The beamforming design problem is split into horizontal and vertical sub-problems, which are solved independently and / or with different periodicities, where each domain can use different optimization methods. Specifically, the final combiner (RX beamformer), precoder (TX beamformer) and IRS phase shifts are expressed as Kronecker products of the horizontal and vertical combiners, precoders and IRS phase shifts, respectively.

[0043] The inventors' proposed scheme is well applicable for a scenario where the channel may show more variations in one of the vertical or horizontal domains as compared to the other domain. Even if the gNB, the IRS and the UE are at fixed positions, with or without unobscured lines of sight (LoS) therebetween, in most of the cases, one of the domains shows lower variations (angular spread) compared to the other. In real scenarios, where the UE is moving, more variation is observed in one angular domain compared to the other angular domain. It is noted that specialized terminology exists in which the horizontal domain corresponds to the so-called azimuth spatial frequencies, while the vertical domain represents the so-called elevation spatial frequencies.

[0044] The inventors' proposed method exploits the variations of the channel along the horizontal (azimuth) and the vertical (elevation) domains, which are due to the movement of the UE. Particularly, depending on the network characteristics, one may perform the beamforming design of different domains with different periodicities. Correspondingly, the beam failure detection and recovery process can be performed independently in horizontal and vertical domains, where the backup beams of each domain are determined independently based on the SNR of each domain. Therefore, according to the invention, in case of SNR drop, the network node, such as the gNB, analyzes (determines) the individual SNR of each domain, and based on this determination, switches the beams in one or both domains.

[0045] In power allocation, the inventors' proposed procedure enables to allocate power separately in each domain without requiring a greater total supply of power to the system. In one embodiment, the domain that shows less variations is allocated more power compared to the domain that shows more variations, and vice versa.

[0046] Turning to the detailed technical description of the proposals, figure 1 shows an example repeater- assisted Ml MO communication system 100 adapted for downlink communication or uplink communication, or both. Here, a gNB 110 is equipped with a uniform rectangular array (URA) 111 having Myantenna elements along a substantially horizontal axis y and Mzantenna elements along a substantially vertical axis z, hence, witha total number of antenna elements M = MyMz. The horizontal and vertical axes may correspond to a row and a column direction of the URA 111, or vice versa. Similarly, a UE 130 is equipped with a URA 131 having Kyantenna elements along a first axis and Kzantenna elements along the second axis, hence, with a total number of antenna elements K = KyKz. The first and second axes, depending on a current orientation of the UE 130, may correspond approximately to a vertical and a horizontal direction, or vice versa. Alternatively, the first and second axes may correspond approximately to the respective axes that are identified as "vertical” and "horizontal” in the gNB 110.

[0047] The repeater 120 for its part is equipped with an array 121 having Nyemitter elements 122 along a horizontal axis, and Nzemitter elements 122 along the vertical axis, hence, a total number of N = NyNz. If the repeater 120 operates like an IRS, the emitter elements 122 are reflecting elements configured to apply a controllable phase shift between an incident radio-frequency wave and a reflected radio-frequency wave. If the repeater 120 operates like an NCR, the emitter elements 122 are antenna elements configured to apply a controllable phase shift between a radio-frequency input signal and a transmitted radio-frequency wave; this may be alternatively described as beamforming.

[0048] IRSs represent an emerging technology that is capable of intelligently manipulating the propagation of electro-magnetic waves. Alternative terminology includes reflective intelligent surface (RIS), large intelligent surface, smart reflect-array, passive intelligent mirror, artificial radio space, and meta-surface. IRSs offer an opportunity for improved wireless communication. Specifically, significant gains are envisioned to be made for millimeter-wave spectrum, which is the spectrum used in fifth-generation telecommunication systems and sixthgeneration telecommunication systems. This spectrum has serious challenges when it comes to propagation and coverage, e.g., due to its support for very high frequency ranges in tens of GHz. The challenges are larger compared to challenges for spectrum with lower frequencies e.g., for so-called sub-6GHz frequency bands.

[0049] In general terms, an IRS is composed of a two-dimensional array of reflecting antenna elements, such as patch antennas, where each antenna element acts as a passive reconfigurable scatterer, i.e., a piece of manufactured material, which can be programmed to change an impinging electro-magnetic wave in a customizable way. Such antenna elements are commonly provided as low-cost passive surfaces that do not require dedicated power sources, and the radio waves impinging upon them can be forwarded without the need for power amplifiers or radio chains. Moreover, an IRS can, potentially, operate in full duplex mode without significant self-interference or increased noise level and requires only low-rate control link or backhaul connections. IRSs can be flexibly deployed in many different locations due to their low weight and low power consumption.

[0050] The system model according to figure 1 further includes G e C7€xW, the Ml MO channel between the gNB 110 and the IRS, which according to the inventors' separation approach is written as a Kronecker product G = Gy® Gzof a horizontal factor Gye CKyxNy and a vertical factor Gze CKzXNz. The nonstandard notation (CKyXJVyrefers sejofmatrjcesof dimension Kyx Ny, which is isomorphic toKyNy. Similarly, H ematrix between the IRS 120 and the UE 130, with H = Hy® Hz, where Hye

[0051] With continued reference to figure 1, 0 e NXNdenotes adiagonal matrix holding the phase shifts of the IRS reflecting elements 122, where 0 = Qy® 0Z, where Qye Cwxxwy represents the horizontal evolution of the phase shifts and 0Ze CWzXWzrepresents a vertical evolution of the phase shifts. If, as in the above example, the horizontal and vertical axes correspond to a row and a column direction of a URA 111 , then the horizontal evolution may correspond to a row-wise progression of the phase shifts, and the vertical evolution may correspond to a column-wise progression of the phase shifts. In this connection, an evolution of phase shifts may correspond to sequences of values of the individual phase shift. In the case of a uniform variation of the phase shifts, the evolution could further correspond to an additive or multiplicative increment between pairs of consecutive emitter elements.

[0052] Further still, the horizontal evolution may correspond to an azimuth angle pbs(figure 1) of a main lobe of the antenna array 111 of the gNB 110, an azimuth angle <pirsDof the array 121 of the repeater 120 or an azimuth angle pueof the array 131 of the UE 130. Similarly, the vertical evolution may correspond to an elevation angle 9bs, 9irsD, 9ueof said main lobe.

[0053] Once the gNB 110 has the estimated channels G and H and is informed about the IRS and / or UE structure (respective number of elements along the vertical and horizontal domains), it can split the optimization problem into horizontal and vertical domains and solve each subproblem independently. Then, the gNB 110 sends to the IRS 120 the optimal horizontal and vertical phase shifts 0y, 0Z. In some embodiments, where the beamforming settings of the gNB 110 and the UE 130 are processed as well, the gNB 110 sends optimal horizontal and vertical combiners wye! Kyxwze C7€zX1to the UE 130 and / or applies optimal horizontal and vertical precoders qye CMxxl, qze CMzX1at its own end. The horizontal and vertical combiners wy, wzdefine a K x 1 combiner matrix given as w = wy® wz, and the horizontal and vertical precoders qy, qzdefine a M x 1 precoder matrix given as q = qy® qz. Correspondingly, the IRS 120 and the UE 130 will be configured according to the received configurations.

[0054] With reference now to figure 2 and in the example case of downlink data communication, there will be described a method 200 for operating a repeater (e.g., IRS, NCR) 120 with a two- or higher-dimensional array 121 of emitter elements, with the general structure suggested in figure 1. The method 200 is to be implemented in a network node 110, such as a gNB in 3GPP NR. As shown in figure 3, the network node 110 may comprise, in addition to the antenna array 111, processing circuitry 113 and a memory 114 suitable for storing a computer program 115 with instructions for carrying out the method 200. The processing circuitry 113 is provided using any combination of one or more of a suitable central processing unit (CPU), multiprocessor, microcontroller, digital signal processor (DSP), etc., capable of executing software instructions stored in a computer program 115. Theprocessing circuitry 113 may further be provided as at least one application-specific integrated circuit (ASIC), or field programmable gate array (FPGA).

[0055] In a first step 210, a channel estimate of a channel between the network node 110 and a wireless device 130 via the repeater 120 is obtained. The channel estimate may be obtained by measurements performed, at least in part, by the network node 110 itself, or it may be received from another entity in the network 100. In the downlink example, the received channel (at the UE 130) can be written as r = wHG0Hqx + n, (1) where w e C7€xlis the combiner, q e CMxlis the precoder with total power p, x is the transmitted symbol and n is a noise term (e.g., additive white Gaussian noise). The channels can be written in factorized form as a sum over V paths:where G^represent the horizontal and vertical channel matrices of the vth path of the UE-IRS link 150, respectively. Similarly,e cNyxMy and HZ(V)e CWzXMzare the horizontal and vertical channel matrices for this path of the gNB-IRS link 140. In most of the two-dimensional mobile communications scenarios, the channel in the vertical domain shows lower variations as compared to the azimuth variations. Therefore, in these scenarios, the channels described in equations (2) and (3) can be approximated as:

[0056] It is noted that step 210 may include obtaining the first channel estimate Hy, Hzfor a channel between the network node 110 and the repeater 120 and obtaining, separately from the first channel estimate, a second channel estimate Gy, Gzfor a channel between the repeater 120 and the wireless device 130. This may be necessary in embodiments where certain types of optimization methods are used. Alternatively, in embodiments where knowledge of the individual channels is not needed (e.g., with optimization based on cascaded channels), the first and second channel estimates can be obtained together, e.g., by the gNB 110 receiving reporting from the UE 130. For example, a horizontal part of the first and second channel estimates can be obtained in the formGy0yHy (wh I ch is equivalent to GyHy0ybecause 0yis diagonal), and a vertical part of the first and second channel estimates can be obtained in the form GZQZHZ= GZHZQZ.

[0057] In a next step 212, the power of the received signal (1) is maximized by varying a beamforming setting.Using the approximations in equations (4) and (5), it is possible to rewrite equation (1) as:Assuming that the combiner vector, IRS matrix and precoder can be factorized into horizontal and vertical components equation (6), in turn, can be rewritten as:Applying the identity (.4 (g) B)(C (g) Z>) = AC (g) BD transforms the first term of equation (7) into a product of scalars, as follows:where qyand qzare horizontal and vertical beamformers, as defined above, with respective transmit powers pyand pz, such that the total transmit power is p = pypz. Here, ryand rzmay be described, respectively, as a horizontal contribution and a vertical contribution to the received signal power.

[0058] It follows from this expression that the spectral efficiency can be rewritten by applying equation (8), namely:where a2denotes noise variance. This allows the spectral efficiency to be expressed asSE = log2(l + SNRySNRz) (10) where the SNR factors are defined asWithout the proposed factorization method, the problem of maximizing SNR at the UE 130 (to be solved by the gNB 110) would be stated: max|ivHG0Hq\ (12a) w,q,Qfor all n, where 0 = diag(01, 02, ... ) and || • || is an €2norm. Exploiting the factorization of the channels, the problem (12) can be reformulated into horizontal and vertical sub-problems. This can be done once the gNB hasinformation about the number of elements along horizontal and vertical domains, at the gNB, at the IRS and at the UE. Also, in Equation (8) the received signal along horizontal and vertical domains, ryand rz, respectively, are independent, which means that maximizing (8) is the same as maximizing both horizontal and vertical components. This corresponds to substeps 212H and 212V, respectively. More precisely, the beamforming design problem in the horizontal domain, where the horizontal beamforming setting is varied, is given by:This problem depends parametrically on the channel estimates Gy, Hy. The beamforming design problem in the vertical domain, where the vertical beamforming setting is varied, becomes:This problem depends parametrically on the channel estimates Gz, Hz. The problems in (13) and (14) can be independently solved, preferably at the gNB 110. In particular, wherein the horizontal beamforming setting wy, qy, Qycan be varied at a horizontal codebook resolution and the vertical beamforming setting wz, qz, 0Zcan be varied at a vertical codebook resolution, which is independent of the horizontal codebook resolution.

[0059] In a next step 214 of the method 200, a transmission of data (e.g., user data, control data, system information) is performed while applying the result of the signal power maximization in step 212. The result of the maximization (performed as one vertical and one horizontal sub-maximization, steps 212V, 212H) is the beamforming setting, which may include a setting for one or more of the gNB 110, the repeater 120, and the UE 130. The optimal combiner vector (to be applied in the UE 130 in the downlink example) is given by a Kronecker product of the combiner vectors in the vertical and in the horizontal domains, which product is evaluated at the gNB 110 or the UE 130. A similar procedure can be applied to obtain the optimal IRS matrix and precoder vector, as follows: q = qy® qz(15a)0 = 0y® 0Z(15b)W = Wy ® wz. (15c)Additionally, a receive-side beamforming setting for an NCR may be provided. From the point of view of signaling, to save overhead, each node preferably receives its optimal horizontal and vertical operation vectors and the Kronecker product between those vectors should be done at each node.

[0060] Figure 4 illustrates the data flows occurring during an execution of the method 200. In figure 4, box 410 corresponds to the problem (12), which is split and solved as the horizontal sub-problem in equation (13) and theindependent vertical sub-problem in equation (14), to which boxes 41 OH and 410V correspond. By combining 490 the solutions from the sub-problems, the optimal beamformer, repeater and combiner matrices in accordance with equations (15).

[0061] Figure 9 is a sequence diagram illustrating the signaling exchanged during an execution of the method. Here, the three columns represent the gNB 110, the IRS (or a general repeater) 120 and the UE 130, respectively.

[0062] Step 910: After the IRS connects to the network, the IRS reports to the gNB its capabilities, i.e., number and indices of reflecting elements etc. (structure information). The capability reporting is an optional step.

[0063] Step 912: The gNB sends a pilot block to the UE via IRS.

[0064] Step 914: The UE optionally estimates the coupled channel.

[0065] Step 916: The UE sends the estimated coupled channel to the gNB and / or sends its own number and / or indices of antenna elements etc. (structure information).

[0066] Step 918: The gNB optionally estimates the channel parameters.

[0067] Step 920: The gNB splits the optimization problem in vertical and horizontal sub-problems using the estimated channels.

[0068] Step 922: The gNB solves each sub-problem independently to obtain the optimal vertical and horizontal precoders, the IRS phase shifts and the combiners. This corresponds to above-described sub-maximizations 212H, 212V.

[0069] Step 924: The gNB communicates the vertical and horizontal optimal phase shifts to the IRS.

[0070] Step 926: The gNB communicates the vertical and horizontal optimal combiners to the UE.

[0071] Step 928: The gNB informs the UE about the operation mode to be used. The modes can be selected from: I) independent data transmission with the same modulation order along horizontal and vertical domains; II) independent data transmission with different modulation order along horizontal and vertical domains; ill) data block is transmitted without exploiting the horizontal and vertical structure.

[0072] In some embodiments of the method 200, the signal power maximization 212 solves the problems (13) and (14) by singular value decomposition (SVD) algorithms. In a classical solution to solve (12), two big SVDs are applied, one on H and one on G to obtain the optimal precoder, combiner and IRS phase shifts. In the inventors' proposed method for solving (13) and (14), four small SVDs are applied on Hy, Hz, Gy, Gzto obtain the optimal precoders, combiners and IRS phase shifts along the horizontal and the vertical domains. This split results in lower computational complexity and occupies less spectral resources.

[0073] In some embodiments of the method 200, the horizontal optimization sub-problem (13) is solved not by numerical optimization but by selecting, based on measurements on reference symbols differing with respect tothe horizontal beamforming setting, a horizontal beamforming setting to be applied. The vertical problem (14) is solved by selecting, based on measurements on reference symbols differing with respect to the vertical beamforming setting, a vertical beamforming setting to be applied.

[0074] In some embodiments, the factorization (8) or (11) of the received signal power into a horizontal contribution and a vertical contribution is used to monitor (step 216H) the horizontal contribution independently of the vertical contribution and / or monitor (step 216V) the vertical contribution independently of the horizontal contribution. The horizontal and the vertical contribution to the received signal power can be monitored as part of a signal-to-noise metric for the corresponding domains. In implementations, these expressions (8) and (11) are normally evaluated using a channel estimate. Specifically, if the vertical domain (for instance) is varying slowly, SNRZcan be considered to be constant for a certain period, whereas the horizontal factor SNRyis computed based on the received SNR. Accordingly, the estimated channels are used, depending on the circumstances, for the vertical or the horizontal or both SNRs.

[0075] Independent power allocation techniques can be applied in (13) and (14). More precisely, in a step 214H, a transmit power is assigned to a horizontal precoder on the basis of the monitored horizontal contribution to the received signal power. In a step 214V, independent of this assignment (though possibly subject to a total power constraint), a transmit power is assigned to a vertical precoder on the basis of the monitored vertical contribution to the received signal power. This may lead to a situation where a higher pyis allocated if the horizontal SNRy(for example) is higher; meanwhile in the vertical domain, for which the SNRZis lower, less power pzis allocated, and the vertical domain can be used for pilot transmission.

[0076] In some embodiments, the method 200 comprises in addition to steps 216H, 216V a step 218H of adapting a first modulation scheme, to be applied with respect to a horizontal spatial diversity direction, on the basis of the monitored horizontal contribution to the received signal power. Such embodiments may further comprise a step 218V of independently adapting a second modulation scheme, to be applied with respect to a vertical spatial diversity direction, on the basis of the monitored vertical contribution to the received signal power. Each of the modulation schemes may include a modulation order.

[0077] In some embodiments, the method 200 comprises in addition to steps 216H, 216V a step 220H of performing horizontal beam management on the basis of the monitored horizontal contribution to the received signal power; and a step 220 V of independently performing vertical beam management on the basis of the monitored vertical contribution to the received signal power. In particular, the horizontal beam management (step 220H) and vertical beam management (step 220V) can be performed concurrently using different reporting periodicities. The horizontal reporting periodicity can be based on observed fluctuation of the horizontal contribution ryto the received signal power, and the vertical reporting periodicity the observed fluctuation of the vertical contribution rzto the received signal power.

[0078] In some embodiments, the method 200 comprises in addition to steps 216H, 216V a step 222H of performing horizontal beam failure detection on the basis of the monitored horizontal contribution to the receivedsignal power; and a step 222V of performing vertical beam failure detection on the basis of the monitored vertical contribution to the received signal power. In these embodiments, when a horizontal beam failure has been detected (step 222H), a partial beam recovery procedure (step 224H), which is restricted to adapting the horizontal beamforming setting, is initiated. Additionally or alternatively, when a vertical beam failure has been detected (step 222V), a partial beam recovery procedure (step 224V), which is restricted to adapting the vertical beamforming setting, is initiated.

[0079] The following facts about the beam recovery procedure in 3GPP NR are recalled. At higher frequencies above 6 GHz, blocking is expected to be common due to the narrow beams used at both the transmit receive points (TRP) and the UE, in addition to high penetration loss and diffraction loss. To handle blocking at higher frequencies, in NR, a beam recovery procedure has been standardized. The NR beam recovery procedure is more efficient than the radio link failure (RLF) procedure used by previous generation, which is a much more costly and time-consuming procedure. The purpose of the beam recovery procedure is to find an alternative beam pair link (BPL) in case the active beam pair link is blocked. The NR beam recovery procedure consists of the following four stages.

[0080] 1. Beam failure detection: The UE detects beam failure by monitoring a dedicated reference signal (CSI- RS or synchronization signal block, SSB) that is spatially quasi co-location (QCL) with physical downlink control channel (PDCCH) and assesses if a trigger condition has been met. The trigger condition is based on block error rate (BLER) for a hypothetical PDCCH based on the measurements on the dedicated DL-RS. A trigger condition is met (i.e., a beam link failure is declared) if the BLER for the hypothetical PDCCH is above a given threshold for X consecutive occasions, where X is a configurable number.

[0081] 2. New candidate beam identification: To quickly find a candidate BPL after a beam link failure, the UE constantly monitors (i.e., measures RSRP on) beam identification reference signal (RS), which for example can be SSB (or periodic CSI-RS if configured). Since SSB is expected to be beamformed at higher frequencies to attain coverage, the UE can determine a preferred candidate TRP SSB beam based on these measurements. Since each SSB consists of four OFDM symbols, the UE can also perform a UE RX beam sweep during each SSB transmission, and hence it is possible for the UE to determine both a suitable TRP beam and UE beam for the candidate BPL.

[0082] 3. Beam Failure Recovery Request transmission: When the UE has declared a beam link failure and a new candidate beam has been determined, the UE transmits a beam failure recovery request (BFRQ) on UL to notify the network about the beam link failure. The BFRQ is a Physical random access channel (PRACH) which implicitly informs the TRP about the preferred TRP SSB beam.

[0083] 4. UE monitors gNB response for beam failure recovery request response: UE monitors for a beam failure recovery request response from the network on the new candidate beam pair link to finalize the beam link recovery procedure.

[0084] A classical BFR design, which is to be extended to the cases with IRSs, is shown in figure 5A. The triplet consisting of L beamformers, U IRS phase shifts and T combiners may be chosen as candidates, where L, P and T represent, respectively, the codebook size of the beamformer, the IRS phase shifts and the combiner. On the other hand, in the inventors' proposed BFR design, shown in figure 5B, the list of candidates is given by six elements, since the beamformers, IRS phase shifts and combiners are split into horizontal and vertical domains, where Lyand Lzrepresent the codebook size of the horizontal and vertical beamformers, respectively, Uyand Uzrepresent the codebook size of the horizontal and vertical IRS phase shifts, respectively, and Tyand Tzrepresent the codebook size of the horizontal and vertical combiners, respectively. This splitting allows, for example, that the domain that shows higher variations can have more elements (oversampled or higher resolution of codebook) compared to the domain with lower variations. Furthermore, the total number of beams (beamformers, IRS phase shifts and combiners) in horizontal and vertical domains is at least equal to the total number of beams in the traditional procedure, since LyLz= L, UyUz= U and TyTz= T. For example, if L = U = T = 4 and Lv= L7= Uv= U7= Tv= T7= 2, the total number of beams in both cases is 12.Example 1

[0085] To illustrate the benefits of the inventors' proposed approach, it is assumed that M = 16, My=4, Mz= 4 (for gNB 110), that K = 16, Ky= 4, Kz= 4 (for repeater 120) and N = 400, Ny= 20, Nz= 20 (for UE 130). To obtain the optimal precoders, the proposed scheme is used to solve (13) and (14) by four small SVDs applied on Hy, Hz, Gy, Gz.

[0086] The computational complexity required to solve the optimization problem is given by the complexity of the four SVDs required, where the complexity of each one is equal to O [4 x 20 x min{4,20}]. Then, the total complexity of the proposed method is 40(320). Meanwhile, in the classical approach, this complexity is given by the complexity of the two big SVDs required, where the complexity of each one is equal to [16 * 400 * min{16,400}]. Then, the total complexity of the classical solution is 20(102400). In this scenario, the complexity of the proposed method is 1 / 160 of that the classical method. Furthermore, in the proposed method, the lengths of the optimal precoder, combiner and IRS phase shift vectors are 8, 8 and 40, respectively. Meanwhile in the classical approach, those lengths are 16, 16 and 400, respectively. In this example, the proposed method requires approximately 1 / 8 of the spectral resources occupied by the classical method.

[0087] As described above, more power can be allocated to the domain which is less varying and / or used for data transmission; meanwhile, less power is allocated to the domain that shows more variation and may be used for pilot transmission to track the variations. Also, more power will be allocated to the domain that uses a higher constellation order, so as to achieve better detection, and less power will be allocated to the domain which uses a lower constellation order. If it is furthermore assumed that the total power at the gNB is p = pypz= 0.1 (in units of watt or another power unit) and the vertical domain is the less varying one, then the power allocated to the vertical domain is pz= 4, so that0.1 — = 0.0254 is allocated to the horizontal domain.Example 2

[0088] To further clarify and show the channel variations along azimuth and elevation domains, a Quadriga Channel Model (QuaDRIGa) is used to verify the claim of factorization of the propagation channel. A moving UE is assumed, which travels from one point in the (y, z) domain. The propagation scenario considered is an urban- micro (UMi) and the central frequency is 28 GHz. (The UMi scenario is discussed, for instance, in Weiler et al., "Measuring the busy urban 60 GHz outdoor access radio channel”, Proceedings of the IEEE International Conference on Ultra-Wideband (ICUWB), Paris, France 2014; pp. 166-170; and Weiler et al., "Simultaneous millimeter-wave multi-band channel sounding in an urban access scenario”, Proceedings of the 9th European Conference on Antennas and Propagation (EuCAP 2015), Lisbon, Portugal, 2015, pp. 1-5.)

[0089] The angular variations along elevation and azimuth domain are calculated and plotted as a cumulative distribution function (CDF), as shown in figure 6. It is clear from figure 6 that during the UE movement from one point to another point, the variations in the azimuth domain happens more frequently as compared to the elevation domain, which leads to the approximation H « Hy® Hz. The approximation reaches the exact solution as the elevation spread decreases, at the limit. When there is no elevation spread, the equality is achieved. This further concludes that these scenarios can be exploited for the proposed beamforming design. Furthermore, the estimated elevation spread is 1.8 degrees.

[0090] To illustrate the benefits of the proposed beamforming solution, figure 7 shows the computational complexity of the inventors' proposed beamforming design and the classical beamforming approach, where no factorization of the channels is considered. Each time, the computational complexity is plotted versus the number of the IRS 120 reflecting elements in a scenario where the BS 110 and the UE 130 are equipped with 16 antennas each. By exploiting the Kronecker factorization structure of the channels the proposed design, a computational complexity significantly lower than for the classical design can be achieved by splitting the problem into horizontal and vertical sub-problems with smaller dimensions.

[0091] Figure 8 shows the spectral efficiency of the proposed beamforming design (circles, dashed trend line) compared to the classical design (triangles, solid trend line). In this scenario, the elevation spread considered is 2.5 degrees, as per the output from the QuaDRIGa channel model. Further, four paths are considered with low elevation spread, i.e. , 2.5 degrees. Although an approximation is used to obtain the factorized version of the channels, the performance of the proposed method is nearly the same as the classical method, i.e., a spectral efficiency of 0.2 bit / s / Hz less than the classical solution. On a system level, this negligible performance loss is by far compensated by the computational complexity reduction demonstrated above.

[0092] Furthermore, it is important to recall that in the mmWave channels and pmWave channels (THz channels), there is a strong probability to have only few multi-paths with a dominant line-of-sight (LoS) path. In this scenario, the proposed approach can nearly reach the exact solution with less computational complexity since the approximation becomes nearly exact for low-rank channels with a dominant LoS.Closing remarks

[0093] The aspects of the present disclosure have mainly been described above with reference to a few embodiments. However, as is readily appreciated by a person skilled in the art, other embodiments than the ones disclosed above are equally possible within the scope of the invention, as defined by the appended patent claims.

[0094] In particular, the approached disclosed herein can also be applied to any point-to-point MIMO communication system, e.g., mMIMO or LoS MIMO (without the assistance of an IRS) as long as the channel can be factorized into horizontal and vertical domains with high accuracy.

[0095] This approach can also be applied for codebook-based beamforming design to allow for different codebook resolutions in each domain independently.

[0096] In a beam failure context, the disclosed approach could be explored to change the sub-beams in each domain independently to form the backup beam. Particularly, one can follow the same approach as in equations (1 )-(14) and find a number of alternative backup beams in each domain. Then, based on the SNR drop in one or both of the domains, one can check different beam backup combinations of the y and z domains, and determine the appropriate backup beams to be used in the cases with a low SNR in one or both of the domains. The SNR drops can be understood using (11), i.e.,SE = log2(l + SNRySNRz) « log2SNRy+ log2SNRZThis provides an additional degree of freedom.

[0097] In power allocation, the disclosed approach can be explored to apply independent power allocation strategies for each domain (horizontal and vertical beamforming). For instance, more power can be allocated to the domain which shows less variations, while less power will be allocated to the domain which shows more variations.

Claims

CLAIMS1. A method (200) of operating a repeater (120) for a cellular communications network (100), said repeater having a two-dimensional array (121) of emitter elements (122) with controllable phase shifts, wherein the method is performed by a network node (110) authorized to control the phase shifts and comprises: obtaining (210) a channel estimate of a channel between the network node and a wireless device via the repeater; maximizing (212), based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmitting (214) data between the network node and the wireless device via the repeater while applying a result of said maximization, wherein the maximization of the received signal power comprises a first sub-maximization (212H), in which a horizontal beamforming setting is varied, and an independent second sub-maximization (212V), in which a vertical beamforming setting is varied, wherein the horizontal and the vertical beamforming setting include, respectively, a horizontal and a vertical evolution of the phase shifts of the repeater.

2. The method (200) of claim 1 , wherein: the network node (110) comprises a two-dimensional array (111) of antenna elements (112) with controllable antenna weights and / or the wireless device (130) comprises a two-dimensional array (131) of antenna elements with controllable antenna weights; the horizontal beamforming setting further includes a horizontal evolution of the antenna weights of the network node and / or the wireless device; and the vertical beamforming setting further includes a vertical evolution of the antenna weights of the network node and / or the wireless device.

3. The method (200) of claim 1 or 2, wherein the horizontal evolution corresponds to an azimuth angle (< >bs, <PirsD> <Pue) ofa mainlobe and / or the vertical evolution corresponds to an elevation angle 6bs> 6irsD> 6ue) of the main lobe.

4. The method (200) of any of the preceding claims, wherein: the phase shifts (0y® 0Z) of the repeater (120) and / or the antenna weights (qy® qz) of the network node (110) and / or the antenna weights (ivy0 wz) of the wireless device (130) are represented as a Kronecker product of a first factor (0y, qy, ivy) and a second factor (0Z, qz, wz)', and the horizontal and the vertical beamforming setting include, respectively, the first and the second factor.

5. The method (200) of claim 4, further comprising: transmitting configuration information to the repeater (120) comprising the first factor (0y) and the second factor (0Z) of the phase shifts to be used during said data transmission (214); and / or transmitting configuration information to the wireless device (130) comprising the first factor (ivy) and the second factor (ivz) of the antenna weights to be used during said data transmission (214).

6. The method (200) of any of claims 1 to 5, further comprising: defining a first and / or a second optimization problem, to be processed in the first or the second sub-maximization (212H, 212V), respectively, using a factorization of the received signal power into a horizontal contribution and a vertical contribution.

7. The method (200) of claim 6, wherein the first and / or second optimization problem depends parametrically on the channel estimate.

8. The method (200) of claim 6 or 7, further comprising processing the first optimization problem and second optimization problem using different optimization solvers.

9. The method (200) of any of claims 1 to 5, wherein: the first sub-maximization (212H) comprises selecting, based on measurements on reference symbols differing with respect to the horizontal beamforming setting, a horizontal beamforming setting to be applied, and / or the second sub-maximization (212V) comprises selecting, based on measurements on reference symbols differing with respect to the vertical beamforming setting, a vertical beamforming setting to be applied.

10. The method (200) of any of the preceding claims, wherein the horizontal beamforming setting is varied at a horizontal codebook resolution and the vertical beamforming setting is varied at a vertical codebook resolution, which is independent of the horizontal codebook resolution.11 . The method (200) of any of the preceding claims, further comprising: using a factorization of the received signal power into a horizontal contribution and a vertical contribution to monitor (216H) the horizontal contribution independently of the vertical contribution and / or monitor (216V) the vertical contribution independently of the horizontal contribution.

12. The method (200) of claim 11, further comprising: adapting (218H) a first modulation scheme, to be applied with respect to a horizontal spatial diversity direction, on the basis of the monitored horizontal contribution to the received signal power; and / or independently adapting (218V) a second modulation scheme, to be applied with respect to a vertical spatial diversity direction, on the basis of the monitored vertical contribution to the received signal power.

13. The method (200) of claim 12, wherein each of the modulation schemes includes a modulation order.

14. The method (200) of any of claims 11 to 13, further comprising: performing horizontal beam management (220H) or horizontal beam failure detection (222H) on the basis of the monitored horizontal contribution to the received signal power; and independently performing vertical beam management (220V) or vertical beam failure detection (222V) on the basis of the monitored vertical contribution to the received signal power.

15. The method (200) of claim 14, wherein the horizontal beam management (220H) and vertical beam management (220V) are performed concurrently using different reporting periodicities.

16. The method (200) of claim 15, further comprising: determining the reporting periodicities based on respective observed fluctuations of the horizontal contribution to the received signal power and the vertical contribution to the received signal power.

17. The method (200) of claim 14, further comprising: in response to detecting (222H) a horizontal beam failure, initiating a partial beam failure recovery procedure (224H) restricted to adapting the horizontal beamforming setting; and / or in response to detecting (222V) a vertical beam failure, initiating a partial beam failure recovery procedure (224V) restricted to adapting the vertical beamforming setting.

18. The method (200) of any of claims 11 to 13, further comprising: assigning (214H) a transmit power to a horizontal precoder on the basis of the monitored horizontal contribution to the received signal power; and independently assigning (214V) a transmit power to a vertical precoder on the basis of the monitored vertical contribution to the received signal power.

19. The method (200) of any of claims 11 to 18, wherein the horizontal and / or the vertical contribution to the received signal power is monitored (216H, 216V) as part of a signal-to-noise metric.

20. The method (200) of any of the preceding claims, wherein obtaining (210) the channel estimate includes obtaining a first channel estimate (Hy, Hz) for a channel between the network node (110) and the repeater (120) and a second channel estimate (Gy, Gz) for a channel between the repeater (120) and the wireless device (130).

21. The method (200) of any of the preceding claims, wherein the repeater (120) includes an intelligent reflecting surface, IRS, and the emitter elements (122) are reflecting elements configured to apply a controllable phase shift between an incident radio-frequency wave and a reflected radio-frequency wave.

22. The method (200) of any of the preceding claims, wherein the repeater (120) includes a network-controlled repeater, NCR, and the emitter elements (122) are antenna elements configured to apply a controllable phase shift between a radio-frequency input signal and a transmitted radio-frequency wave.

23. The method (200) of any of claims 1 to 22, wherein the data is transmitted (214) from the network node (110) to the wireless device (130) via the repeater (120).

24. The method (200) of any of claims 1 to 22, wherein the data is transmitted (214) from the wireless device (130) to the network node (110) via the repeater (120).

25. A network node (130) arranged to operate a repeater (120) for a cellular communications network (100), said repeater having a two-dimensional array (121) of emitter elements (122) with controllable phase shifts, wherein the network node comprises processing circuitry (113) configured to: obtain a channel estimate of a channel between the network node and a wireless device via the repeater; maximize, based on the channel estimate, a received signal power by varying a beamforming setting including the phase shifts of the repeater; and transmit data between the network node and the wireless device via the repeater while applying a result of said maximization, wherein the maximization of the received signal power comprises a first sub-maximization, in which a horizontal beamforming setting is varied, and an independent second sub-maximization, in which a vertical beamforming setting is varied, wherein the horizontal and the vertical beamforming setting include, respectively, a horizontal and a vertical evolution of the phase shifts of the repeater.

26. A computer program (115) comprising instructions which, when run on processing circuitry (113) of a network node (110), causes the network node to perform the method (200) of any of claims 1 to 24.

27. A computer program product comprising the computer program (115) of claim 26 and a computer-readable storage medium (114) on which the computer program is stored.

28. A method performed by a wireless device (130) for a cellular communications network (100), said wireless device having a two-dimensional array (131) of antenna elements with controllable antenna weights, the method comprising: transmitting (916) structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receiving (926) configuration information from the network, which configuration information indicates a first factor (ivy) and a second factor (ivz) of a combiner to be used by the wireless device; and sending data to the network or receiving data from the network in accordance with the configuration information.

29. The method of claim 28, further comprising: forming a combiner by evaluating a Kronecker product of the first factor (ivy) and the second factor (ivz) indicated by the configuration information.

30. The method of claim 28 or 29, further comprising: receiving (928) mode information indicating one of operation mode to be used by the wireless device, wherein the operation mode is selected from one or more of the following:I) independent data transmission with the same modulation order along horizontal and vertical domains; ii) independent data transmission with different modulation order along horizontal and vertical domains; ill) data block is transmitted without exploiting any horizontal and vertical structure.

31. A wireless device (130) for a cellular communications network (100), said wireless device having a two- dimensional array (131) of antenna elements with controllable antenna weights, wherein the wireless device comprises processing circuitry configured to: transmit structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receive configuration information from the network, which configuration information indicates a first factor (wy) and a second factor (ivz) of a combiner to be used by the wireless device; and send data to the network or receive data from the network in accordance with the configuration information.

32. A computer program comprising instructions which, when run on processing circuitry of a wireless device, causes the wireless device to perform the method of any of claims 28 to 31 .

33. A computer program product comprising the computer program of claim 32 and a computer-readable storage medium on which the computer program is stored.

34. A method performed by a repeater (120) for a cellular communications network (100), said repeater having a two-dimensional array (121) of emitter elements (122) with controllable phase shifts, the method comprising: transmitting (910) structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receiving (924) configuration information from the network, which configuration information indicates a first factor (0y) and a second factor (0Z) of phase shifts to be used by the repeater; and facilitating a data transmission in the network in accordance with the configuration information.

35. The method of claim 34, further comprising: forming phase shifts by evaluating a Kronecker product of the first factor (0y) and the second factor (0Z) indicated by the configuration information.

36. A repeater (120) for a cellular communications network (100), said repeater having a two-dimensional array (121) of emitter elements (122) with controllable phase shifts, wherein the repeater comprises processing circuitry configured to: transmit structure information to the network, which structure information indicates a horizontal and a vertical size of the two-dimensional array; receive configuration information from the network, which configuration information indicates a first factor (0y) and a second factor (0Z) of phase shifts to be used by the repeater; and facilitate a data transmission in the network in accordance with the configuration information.

37. A computer program comprising instructions which, when run on processing circuitry of a repeater, causes the repeater to perform the method of claim 34 or 35.

38. A computer program product comprising the computer program of claim 37 and a computer-readable storage medium on which the computer program is stored.

Images