Reconfigurable intelligent surface

Abstract

This invention provides a method of beam training in a wireless telecommunications network, and a device for implementing said method, the wireless telecommunications network comprising a transmitter, a Reconfigurable Intelligent Surface, RIS, and a receiver, the method comprising the steps of: iteratively, obtaining a plurality of reflecting beamforming configurations for the RIS, each reflecting beamforming configuration defining a multi-location reflection beam covering a set of locations, causing configuration of the RIS according to each reflecting beamforming configuration so as to reflect signals from the transmitter according to each multi-location reflection beam, obtaining data indicating a received power of each multi-location reflection beam at the receiver, and identifying a candidate multi-location reflection beam of the multi-location reflection beams based on the power of the multi-location reflection beam at the receiver, so as to identify a plurality of candidate multi-location reflection beams; identifying a location that is a member of each candidate multi-location reflection beam of the identified plurality of candidate multi-location reflection beams; generating a reflecting beamforming configuration for the RIS based on the identified location; and causing configuration of the RIS according to the generated reflecting beamforming configuration.

Description

[0001] RECONFIGURABLE INTELLIGENT SURFACE

[0002] Field of the Invention

[0003] The present invention relates to a reconfigurable intelligent surface.

[0004] Background

[0005] In wireless telecommunications, a wireless signal being transmitted between a transmitter and receiver generally degrades due to interference from other wireless signals and / or other physical phenomena (e.g. fading and blockage). This has generally been addressed by improving the transmission characteristics (e.g. higher power transmissions or repeaters) or transmission processing techniques (e.g. more robust modulation schemes). An emerging concept in wireless telecommunications is the concept of a reconfigurable propagation environment, or “smart radio environment”, which may improve the transmission quality. This may be achieved by use of a surface of electromagnetic material, often known as a Reconfigurable Intelligent Surface (RIS), which may be operated to apply a change to an incident wireless signal, such as a change in phase, amplitude, frequency and polarisation, so as to improve the transmission quality between the transmitter and the receiver. Alternative names for the RIS include intelligent reflective surface, large intelligent surface, large intelligent metasurface, programmable metasurface, reconfigurable metasurface, smart reflect- arrays, software-defined surface, and passive intelligent surface. The term “reconfigurable” is often used to indicate that the angle of reflection can be configured regardless of the angle of incidence.

[0006] A RIS may be a cost-effective solution to improving transmission quality by providing high reflecting beamforming gain compared to alternative solutions, such as by increasing access point density, as RISs are nearly passive and easy to deploy. Reliable RIS reflecting beamforming requires accurate Channel State Information (CSI). There are two categories of CSI acquisition in RIS-assisted wireless telecommunication systems - explicit CSI acquisition (i.e. channel estimation) or implicit CSI acquisition (i.e. beam training). In channel estimation, a transmitter transmits pilot signals to a receiver via the RIS, and the receiver directly estimates the channel based on the received pilot signals. Since each RIS element is passive, the cascaded channel (that is, the cascade of the channel from the transmitter to the RIS and the channel from the RIS to the receiver) can be estimated, such as by a least squares or minimum mean square error algorithm. In beam training, the CSI can be obtained by estimating the physical directions of channel paths instead of the entire channel. That is, the transmitter and receiver perform the training procedure through transmission of multiple directional beams (codewords) predefined in a codebook to identify the optimal directional beam. Following beam training, the physical directions of the channel paths can be effectively obtained. This concept can be extended to RIS-assisted systems by constructing a codebook of multiple RIS directional beams, and then performing the training procedure between the RIS and receiver to search for the optimal directional beam between the RIS and the receiver.

[0007] It is desirable to increase the number of reflecting elements of a RIS for enhanced beamforming. However, the electromagnetic field of the RIS-assisted system changes as the number of reflecting elements in a RIS increases. An electromagnetic field can be divided into a near-field region and a far-field region. These regions may be defined from the perspective of the transmitter or the receiver. If the receiver is in the near-field of the transmitter, the propagation distances are so short that there are non-negligible amplitude and phase variations over the receiver aperture. In contrast, both the amplitude and phase variations are negligible in the far-field region, in which the amplitude only depends on the propagation distance to the centre of the receiver and the phase variations only depend on the incident angle. The boundary at which the electromagnetic field can be considered near-field or far-field is defined by the Rayleigh distance, which is proportional to the square of the array aperture. Therefore, as the array aperture increases (e.g. due to an increased number of reflecting elements) then the Rayleigh distance increases and the channel model may be considered in the near- field. A category of RIS, extremely-large RIS, is defined in which the number of reflecting elements surpasses a threshold such that their propagation environment is modelled as a near-field channel model. The conventional RIS beam training techniques, which assume a far-field channel model, may not apply to a RIS-assisted system having a near- field channel model.

[0008] A beam training technique for a RIS-assisted system having a near-channel model has been proposed in, “Codebook design and beam training for extremely large-scale RIS: Far-field or near-field?” X. Wei, L. Dai, Y. Zhao, G. Yu, and X. Duan, China Commun., vol. 19, no. 6, pp. 193-204. In this paper, a near-field codebook is designed to match the near-field channel model, and a corresponding near-field beam training scheme is proposed. Specifically, by considering the near-field cascaded array steering vector of the RIS cascaded channel, the near-field codebook is designed, where each codeword is determined by a pair of sampled points in the x-y-z coordinate system. Then, the optimal codeword for the RIS is obtained by the exhaustive training procedure between the RIS and the receiver. To reduce the beam training overhead, it is further proposed to use a hierarchical near-field codebook and corresponding near-field beam training scheme. Compared with the near-field codebook, the hierarchical near-field codebook consists of several different levels of sub-codebooks, which are determined by different sampling ranges and sampling steps. During the beam training process, the sub- codebooks are searched in turn from the first level sub-codebook to the last level sub- codebook, where the sampling ranges and sampling steps gradually become smaller. Finally, the globally optimal codeword can be achieved in the last level sub-codebook associated with the minimum sampling ranges and sampling steps. The technique of this paper suffers from severe searching precision degradation and unnecessary training overhead. Furthermore, the codebook size will increase rapidly as transmission distance increases, which is unacceptable for a real-world application.

[0009] It is desirable to alleviate some or all of the above problems.

[0010] Summary of the Invention

[0011] According to a first aspect of the invention, there is provided a method of beam training in a wireless telecommunications network comprising a transmitter, a Reconfigurable Intelligent Surface, RIS, and a receiver, the method comprising the steps of: iteratively, obtaining a plurality of reflecting beamforming configurations for the RIS, each reflecting beamforming configuration defining a multi-location reflection beam covering a set of locations, causing configuration of the RIS according to each reflecting beamforming configuration so as to reflect signals from the transmitter according to each multi-location reflection beam, obtaining data indicating a received power of each multi-location reflection beam at the receiver, and identifying a candidate multi-location reflection beam of the multi-location reflection beams based on the power of the multi-location reflection beam at the receiver, so as to identify a plurality of candidate multi-location reflection beams; identifying a location that is a member of each candidate multi-location reflection beam of the identified plurality of candidate multi-location reflection beams; generating a reflecting beamforming configuration for the RIS based on the identified location; and causing configuration of the RIS according to the generated reflecting beamforming configuration.

[0012] Each location may be a member of a multi-location reflection beam in a current iteration having a first set of locations, and a member of a multi-location reflection beam in a previous iteration having a second set of locations, wherein the first and second sets of locations have a unique membership.

[0013] The step of obtaining the plurality of reflecting beamforming configurations for the RIS may comprise generating the plurality of reflecting beamforming configurations for the RIS.

[0014] A reflecting beamforming configuration of the plurality of reflecting beamforming configurations may be generated randomly such that the set of locations of the multi- location reflection beam defined by the reflecting beamforming configuration are randomly generated.

[0015] A reflecting beamforming configuration of the plurality of reflecting beamforming configurations of a current iteration may be generated based on the power of the multi- location reflection beam at the receiver in a previous iteration, such that the set of locations of the multi-location reflection beam defined by the reflecting beamforming configuration of the current iteration may be generated based on the power of the multi- location reflection beam at the receiver in the previous iteration.

[0016] The iterative steps may be repeated until a termination condition is met, the termination condition being at least one of a group comprising: a predetermined count of iterations; and one or more locations are identifiable as being members of each candidate multi- location reflection beam, wherein a count of the one or more locations is less than a cardinality of the set of locations.

[0017] According to a second aspect of the invention, there is provided a computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of any one of the method of the first aspect of the invention. The computer program may be stored on a computer readable carrier medium. According to a third aspect of the invention, there is provided a device in a wireless telecommunications network, the wireless telecommunications network comprising a transmitter, a Reconfigurable Intelligent Surface, RIS, and a receiver, the device comprising a processor configured to perform the steps of the method of the first aspect of the invention.

[0018] According to a fourth aspect of the invention, there is provided a device of the third aspect of the invention and a RIS.

[0019] Brief Description of the Figures

[0020] In order that the present invention may be better understood, embodiments thereof will now be described, by way of example only, with reference to the accompanying drawings in which:

[0021] Figure 1 is a schematic diagram of a cellular telecommunications network;

[0022] Figure 2 is a flow diagram illustrating steps of a method of beam training implemented by a base station;

[0023] Figure 3 is a schematic diagram illustrating training beams being transmitted by the base station in a first iteration of the method of beam training;

[0024] Figure 4 is a flow diagram illustrating steps of the method of beam training implemented by a Reconfigurable Intelligent Surface;

[0025] Figure 5 is a flow diagram illustrating steps of the method of beam training implemented by a User Equipment (UE);

[0026] Figure 6 is a graph illustrating performance of the method of beam training; and

[0027] Figure 7 is a schematic diagram illustrating training beams being transmitted by the base station.

[0028] Detailed Description

[0029] Figure 1 illustrates a cellular telecommunications network 100. The wireless telecommunications network 100 comprises a base station 110, a Reconfigurable Intelligent Surface (RIS) 120, and a User Equipment (UE) 130. The base station 110 includes an N-element antenna array and the UE 130 comprises a single antenna. The RIS 120 consists of N ( N1x N2) elements and is placed on an y-z plane, in which the centre of the RIS 120 is positioned at an origin of an x-y-z coordinate system. The array aperture of the RIS 120 is sufficiently large for the channel model between the RIS 120 and UE 130 to be considered a near-field channel model. This may be, for example, due to the number of reflecting elements of the RIS 120 being sufficiently high. The RIS 120 may therefore be considered an extremely large RIS.

[0030] A communications path between the base station 110 and UE 130 is a cascaded channel comprising a first channel between the base station 110 and RIS 120, and a second channel between the RIS 120 and UE 130, . In this example, the direct channel between the base station 110 and UE 130 is blocked and therefore not considered in the following analysis. Based on this channel model, the received signal, r, at the UE 130 can be expressed as: r = hrdiag(θ)Gvs + n (1)

[0031] In which:

[0032] • is the reflecting beamforming vector at the RIS 120 with θnrepresenting the reflecting coefficient at the n-th reflecting element of the RIS 120 (n = 1 ,... ,N);

[0033] • represents the beamforming vector at the base station 110;

[0034] • s represents the symbol transmitted by the base station 110; and

[0035] • represents the received noise at the UE 130 with σ2representing the noise power.

[0036] The following is noted to aid understanding of the method of beam training described below. Since the base station 110 and RIS 120 are generally deployed in fixed positions and the UE 130 is generally mobile, then the first channel, G, will likely have a much longer coherence time than the second channel, hr. Accordingly, for simplicity, it can be assumed that the beamforming vector, v, at the base station 110 is aligned with the main path of the first channel, G. It is therefore only necessary to perform beam training at the RIS 120.

[0037] An effective distance, D(n1,n2), of the near-field channel, , may be represented as:

[0038] (2) In which represents the distance between the base station 110 and the (n1, n2)-th reflecting element of the RIS 120 and Dr(n1,n2) represents the distance between the (n1, n2)-th reflecting element of the RIS 120 and the UE 130.

[0039] The distance between two adjacent reflecting elements of the RIS 120 may be defined as d, the coordinate of the scatter (that is, a part of a beam) corresponding to the main path between the base station 110 and the RIS 120 may be defined as , and the coordinate of the scatter between the RIS 120 and the UE 130 may be defined as (rr, Φr,ψr) The near-field channel from the RIS 120 to the UE 130, , may be represented as: (3)

[0040] In which αrrepresents the path gain.

[0041] The distance between the (n1, n2)-th reflecting element of the RIS 120 and the UE 130, Dr(n1,n2) , may be represented as: (4)

[0042] In which rrrepresents the distance from the centre of the RIS 120 to the UE 130. The near-field cascaded channel (that is, encompassing the first channel from the base station 110 to the RIS 120 and the second channel from the RIS 120 to the UE 130), , may be represented as: (5)

[0043] The near-field cascaded array steering vector may be represented by:

[0044] (6)

[0045] Therefore, the received signal at the UE 130, , may be represented as: (7)

[0046] The beam training procedure is therefore to select, for a user for a particular timeslot, the optimal beam and codeword based on the received power of training beams transmitted during the training procedure. A method of beam training will now be described.

[0047] Figures 2, 4 and 5 illustrate a method of beam training, in which Figure 2 illustrates steps performed by the base station 110, Figure 4 illustrates steps performed by the RIS 120 and Figure 5 illustrates steps performed by the UE 130. In step S101 of Figure 2, the base station 110 obtains a plurality of input parameters including:

[0048] • A polar-domain transform matrix, where Q represents the number of columns in matrix W and is also equivalent to the number of sampled near-field steering vectors in the polar domain;

[0049] • A number, N, of reflecting elements of the RIS 120;

[0050] • A wavelength, λ, used in communications between the base station 110 and UE 130; and

[0051] • A number of iterations, L, which will be explained in more detail below.

[0052] These input parameters may be locally generated by the base station 110, or generated by an external controller and sent to the base station 100 via a suitable communications interface. In one example, the polar-domain transform matrix, W, can be generated using the method described in “Channel estimation for extremely large-scale Ml MO: Far- field or near-field?” M. Cui and L. Dai, vol. 70, no. 4, pp.2663-2677, Apr. 2022.

[0053] In step S103, the base station 110 generates a polar-domain representation of the near field channel . The near-field channel can be represented as: (8)

[0054] In which, as noted above, W is a sparse matrix being the polar-domain transform matrix. Equation (8) can therefore be rearranged to define the polar-domain representation of the near-field channel, as: (9)

[0055] The base station 110 then enters an iterative loop for L iterations. In step S105, the first step of the iterative loop, the base station 110 generates the reflecting beamforming vector, θ, at the RIS 120. The reflecting beamforming vector, θ, is generated as:

[0056] (10)

[0057] In which:

[0058] • is a sparse vector with a maximum of R non-zero elements,

[0059] • l = 1,2 ,...,L,

[0060] • b = 1,2, ... B, in which B represents the number of “multi-location beams” (described in more detail below) transmitted in each iteration, and

[0061] • q = 1,2, ... Q.

[0062] The reflecting beamforming vector can simultaneously form R beams, each directed to a specific location, resulting in a generation of a “multi-location beam”. This can be achieved through the design of the sparse vector in which the non-zero elements in are randomly generated for each iteration. In this example, the non-zero elements are randomly generated with an additional constraint that the combination of locations covered by all the multi-location beams of a particular iteration covers all locations.

[0063] An example of a first, second and third multi-location beam of a first iteration is illustrated in Figure 3. Figure 3 illustrates 9 locations, labelled 1 to 9. A first multi-location beam is generated by a first reflecting beamforming vector in the first iteration as so as to simultaneously cover locations 1 , 3 and 5. A second multi- location beam is generated by a second reflecting beamforming vector in the first iteration as so as to simultaneously cover locations 2, 6 and 8. A third multi-location beam is generated by a third reflecting beamforming vector in the first iteration as so as to simultaneously cover locations 4, 7 and 9. These multi-location beams therefore each cover 3 locations (i.e. B =3), and the combination of all three multi-location beams cover all 9 locations.

[0064] An example beamforming vector generation method in which the beam is a function of both angle and distance (rather than just angle) is detailed in “Channel estimation for extremely large-scale MIMO: Far-field or near-field?” M. Cui and L. Dai, vol. 70, no. 4, pp.2663-2677, Apr. 2022. This enables, for example, the first beamforming vector to cover location 5 without also covering location 2.

[0065] An indicator set, , is defined representing the non-zero elements in . Accordingly, in this example of the first iteration, a first indicator set, , is defined as {1 ,3,5}, a second indicator set, , is defined as {2,6,8}, and a third indicator set, , is defined as {4,7,9}.

[0066] In the following example, the UE 130 is positioned at location 3.

[0067] In step S107, the base station 110 sends a configuration message to the RIS 120 comprising the generated reflecting beamforming vectors, to . As shown in step S201 of Figure 4, the RIS 120 receives these reflecting beamforming vectors and the reflecting elements of the RIS 120 are configured so as to reflect a training beam according to these reflecting beamforming vectors.

[0068] A training beam communication is then performed such that the base station 110 transmits training beams, the RIS 120 reflects the training beams according to the reflecting beamforming vectors (so as to reflect the multi-location beams), and the UE 130 measures the multi-location beams. These steps are illustrated as step S109 of Figure 2, step S203 of Figure 4, and step S301 of Figure 5. In this example, the base station 110 and RIS 120 communicate these multi-location beams to the UE 130 in respective timeslots, such that the first multi-location beam is transmitted in a first timeslot, the second multi-location beam is transmitted in a second timeslot, and the third multi-location beam is transmitted in a third timeslot. Furthermore, each multi-location beam is transmitted with a corresponding identifier.

[0069] In more detail, step S301 of Figure 5 comprises the UE 130 measuring and recording the received power (that is, J) of each multi-location beam, such that the received power of the first multi-location beam, received power of the second multi-location beam and the received power of the third multi-location beam are recorded, alongside their respective identifiers. In this example, these received powers are recorded in a detection matrix, Yl. In an example of this first iteration, Y1= {8,0.03,0.01}, in which 8 represents the received power of the first multi-location beam (covering locations 1 , 3 and 5, and therefore covers the UE 130 at location 3), 0.03 represents the received power of the second multi-location beam (covering locations 2, 6 and 8, and therefore not covering the UE 130), and 0.01 represents the received power of the third multi-location beam (covering locations 4, 7 and 9, and therefore not covering the UE 130).

[0070] In step S303, the UE 130 reports the detection matrix to the base station 110.

[0071] In step S111 of Figure 2, the base station 110 analyses the received detection matrix, Yl, so as to populate a candidate location index set, with the indicator set, corresponding to the largest element of the detection matrix Yl. In this example of the first iteration, the largest element of the detection matrix Y1is for the first multi-location beam, corresponding with indicator set . Accordingly, the candidate location index set for the first iteration is populated as .

[0072] Following this first iteration, it may be determined that the UE 130 is positioned at one of locations 1, 3 and 5 of the candidate location index set, . That is, the location index, , of the UE 130 satisfies: (11)

[0073] However, on completion of this first iteration, it is not possible to determine the location index of the UE 130 as a single location. In step S113, it is determined whether the current iteration number equals the total number of iterations, L. If decision block S113 is answered in the affirmative, then the method proceeds to step S115 (described below). If decision block S113 is answered in the negative, then the method loops back to step S105 for a further iteration.

[0074] An example second iteration will now be described. In step S105 of this example second iteration, a new reflecting beamforming vector is generated as and , These locations covered by the first multi-location beam, second multi-location beam and third multi-location beam of this second iteration are also illustrated in Figure 3.

[0075] Steps S107 and S109 of Figure 2, S201 and S203 of Figure 4 and S301 and S303 of Figure 5 are then performed in the same manner as described above for the first iteration. In this example second iteration, the detection matrix Y2for the second iteration is Y2= {0.1,0.02,7}. Accordingly, in step S111 of this example second iteration, the candidate location index set for the second iteration is populated as .

[0076] Accordingly, following the first iteration, the location index of the UE 130 satisfies both: (12)

[0077] For the purposes of simplicity, step S115 will now be described following completion of the first and second iterations. In step S115, the location of the UE 130 is determined as an intersection of the candidate location index set for all iterations, i.e. (13)

[0078] In this example, the location of the UE 130, , is determined as location 3 as the intersection of the respective candidate location index set of the first and second iterations {1 ,3,5}, {3,6,9}.

[0079] In step S117, the base station 110 sends a configuration message to the RIS 120 so as to configure the RIS 120 for reflecting signals transmitted by the base station 110 to the position of the UE 130 at location 3. In step S205, the RIS 120 configures its reflecting elements according to the configuration message.

[0080] In step S119, the base station 110 transmits a data signal to the UE 130 via the RIS 120, as configured in step S205 so as to reflect the data signal to the UE 130 at location 3.

[0081] The above beam training method offers improved performance in terms of, for example, achievable rate versus training overhead relative to alternative beam training methods. The above beam training method achieves these benefits by applying the concept of Sparse Fast Fourier Transform (SFFT) to near field beam training. That is, the concept of SFFT is that the signal being processed is a sparse signal, so by processing only a small part of the signal, the complexity of processing is greatly reduced. In this case, the sparsity of the polar-domain representation of the near-field channel allows such SFFT processing. These benefits are illustrated by a comparison of the performance of the beam training method of Figures 2, 4 and 5 to alternative beam training methods, as shown in Figure 6. In these simulations, the beam training overhead can be represented as: (14) In which:

[0082] • represents the complexity of the algorithm; and

[0083] • K represents the number of elements in a candidate index set for each round (that is, K is equivalent to B above). The performance of the beam training method of Figures 2, 4 and 5 was simulated with the following input parameters, N=512 and K=10. The simulation indicates that the beam training method of Figures 2, 4 and 5 achieves around 95% of the achievable rate performance of an exhaustive beam training method based on spherical coordinates. The overhead of the beam training method of Figures 2, 4 and 5 and alternative beam training methods is shown in the following table:

[0084] Table 1 : Comparison of beam training methods

[0085] The method of beam training described above may be summarised as the following algorithm:

[0086] In the above description, the multi-location beams transmitted by the base station were transmitted to specific locations. However, this is non-essential. In an alternative implementation, each beam of the multi-location beams may be transmitted in a continuous range from a particular angle. This is illustrated in Figure 7. Figure 7 illustrates the base station 110 being configured to transmit a multi-location beam covering 3 of 9 possible locations. The base station 110 is configured to transmit a multi- location beam (illustrated by solid line) to locations 1 , 3 and 5. The dotted lines illustrate the beams (and locations) that are not part of the multi-location beam.

[0087] The above beam training method defines an iterative loop which is performed for L iterations, at which point the candidate location index sets are analysed to identify the location of the UE. However, the skilled person will understand that the beam training method may be implemented in alternative ways, which may include an alternative or additional termination condition for the iterative steps. For example, the following alternative implementation includes an iterative loop which is performed for L iterations or until a location can be uniquely identified in the candidate location index sets of all previous iterations. This alternative implementation is summarised in the following algorithm:

[0088] Algorithm 2 will be described by the following example. The reflecting beamforming vectors for a first and second iteration are as described above (i.e. , and , The score array is initialised as b = [0,0,0,0,0,0,0,0,0], Following the first iteration, the detection matrix is Y1= {8,0.03,0.01} and so the candidate location index set for the first iteration is . The score array is therefore modified such that each location of the candidate location index set of the first iteration is increased by 1 , i.e. b = [1,0, 1,0, 1,0, 0,0,0], As I < L and sum(b = max(b)) = 3, then a second iteration is performed.

[0089] In the second iteration, the detection matrix is Y2= {0.1,0.02,7} and so the candidate location index set for the second iteration is . The score array is therefore modified such that each location of the candidate index set of the second iteration is increased by 1 , i.e. b = [1,0, 2, 0,1, 1,0, 0,1], As sum(b = max(b)) is not greater than 1 , then the while loop ends and the process proceeds to step 10 of algorithm 2. In step 10, it is determined that sum(b = max(b)), and so the index of the score matrix having the greatest value - location 3 - is determined as the location of the UE 130.

[0090] The termination condition of the while loop may alternatively be triggered when I = L, despite the score matrix not identifying a single location as the location of the UE 130 (i.e. sum(b = max(b)) > 1). In this scenario, the location of the UE 130 is determined according to step 13 of algorithm 2 in which the optimal location is determined as the indices of the score matrix having the greatest value. This therefore results in the location of the UE 130 being determined as one of a plurality of locations, but this is still an improvement so long as there are fewer locations covered relative to the multi-location beam used in the beam training method.

[0091] In the above beam training methods, a plurality of multi-location beams are transmitted covering the UE so as to identify a plurality of candidate location sets, wherein the actual location of the UE is a member of each candidate location set. The location of the UE may then be identified when a single location is identifiable as a member of all candidate location sets. This identification is enabled by generating the multi-location beams of each iteration until there are sufficient candidate location sets for a unique identification to be made. This is achieved, in the above embodiment, by randomly generating these multi-location beams and performing the method for a sufficiently large number of iterations, as described above (at least in relation to algorithm 1 , and potentially algorithm 2). However, this random generation of the reflecting beamforming vectors is non- essential. More generally, the multi-location beams are generated such that a particular location is a member of a first multi-location beam and a second multi-location beam, wherein the set of other locations in the first and second multi-location beams are different. In other words, if a particular location is a member of a first and second multi- location beam, then the membership of locations of the first multi-location beam is unique relative to the membership of locations of the second multi-location beam. Put yet another way, in the second and any subsequent iteration of the beam training method, a location is comprised in a unique membership of a set of locations covered by its multi- location beam of a current iteration relative to its membership of a set of locations covered by its multi-location beam in a previous iteration. It is also possible for the allocation of locations to a multi-location beam to be determined based on the detection matrix of one or more previous iterations. For example, following a first iteration in which the candidate location index set is identified as , then the multi-location beams of a second iteration may include a first multi-location beam covering location 1 (but not locations 3 and 5), a second multi-location beam covering location 3 (but not locations 1 and 5), and a third multi-location beam covering location 5 (but not locations 1 and 3). This approach may reduce the number of iterations required to uniquely identify the location of the UE (relative to the random generation of reflecting beamforming vectors).

[0092] The skilled person will also understand that the method of beam training may apply to any form of wireless telecommunications network that utilises beam forming and a RIS. The steps of the method performed by the base station may therefore more generally be performed by an access point or transmitter, and the steps performed by the UE may therefore more generally be performed by a receiver.

[0093] The skilled person will also understand that it is non-essential that the method is performed by the base station (or access point / transmitter), as it may otherwise be performed by another entity (e.g. a controller) in cooperation with the base station 110, RIS 120 and UE 130 to perform their respective transmitting, reflecting, measuring and reporting steps.

[0094] The skilled person will understand that recording the identities of the candidate multi- location beam having the greatest received power and identifying a single location that is a member of all multi-location beams does not require the use of sets and an intersection analysis. These descriptions of the data and analysis are provided as an example to improve clarity.

[0095] The benefits of using the above beam training method when a RIS-aided communication is based on a near-field channel is discussed above. However, the above beam training method is also applicable to far-field channels.

[0096] The skilled person will understand that any combination of features is possible, within the scope of the invention as claimed.

Claims

CLAIMS1. A method of beam training in a wireless telecommunications network comprising a transmitter, a Reconfigurable Intelligent Surface, RIS, and a receiver, the method comprising the steps of: iteratively, obtaining a plurality of reflecting beamforming configurations for the RIS, each reflecting beamforming configuration defining a multi-location reflection beam covering a set of locations, causing configuration of the RIS according to each reflecting beamforming configuration so as to reflect signals from the transmitter according to each multi-location reflection beam, obtaining data indicating a received power of each multi-location reflection beam at the receiver, and identifying a candidate multi-location reflection beam of the multi- location reflection beams based on the power of the multi-location reflection beam at the receiver, so as to identify a plurality of candidate multi-location reflection beams; identifying a location that is a member of each candidate multi-location reflection beam of the identified plurality of candidate multi-location reflection beams; generating a reflecting beamforming configuration for the RIS based on the identified location; and causing configuration of the RIS according to the generated reflecting beamforming configuration.

2. A method as claimed in Claim 1 , wherein each location is: a member of a multi-location reflection beam in a current iteration having a first set of locations, and a member of a multi-location reflection beam in a previous iteration having a second set of locations, wherein the first and second sets of locations have a unique membership.

3. A method as claimed in Claim 1 or Claim 2, wherein the step of obtaining the plurality of reflecting beamforming configurations for the RIS comprises generating the plurality of reflecting beamforming configurations for the RIS.

4. A method as claimed in Claim 3, wherein a reflecting beamforming configuration of the plurality of reflecting beamforming configurations is generated randomly such that the set of locations of the multi-location reflection beam defined by the reflecting beamforming configuration are randomly generated.

5. A method as claimed in Claim 3, wherein a reflecting beamforming configuration of the plurality of reflecting beamforming configurations of a current iteration is generated based on the power of the multi-location reflection beam at the receiver in a previous iteration, such that the set of locations of the multi-location reflection beam defined by the reflecting beamforming configuration of the current iteration is generated based on the power of the multi-location reflection beam at the receiver in the previous iteration.

6. A method as claimed in any one of the preceding claims, wherein the iterative steps are repeated until a termination condition is met, the termination condition being at least one of a group comprising: a predetermined count of iterations; and one or more locations are identifiable as being members of each candidate multi- location reflection beam, wherein a count of the one or more locations is less than a cardinality of the set of locations.

7. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of any one of Claims 1 to 6.

8. A computer readable carrier medium comprising the computer program of Claim 7.

9. A device in a wireless telecommunications network, the wireless telecommunications network comprising a transmitter, a Reconfigurable Intelligent Surface, RIS, and a receiver, the device comprising a processor configured to perform the steps of any one of Claims 1 to 6.

10. A system comprising: a device as claimed in Claim 9; anda RIS.

Images