DEVICE AND METHOD FOR HYBRID RF-OPTICAL MULTI-BEAMFORMING USING CORRELATION-BASED EVM METRICS  The present patent application claims priority from the European patent application filed on 21 March 2022 and assigned application no. EP22305332, the contents of which is hereby incorporated by reference. Technical field [0001] The present disclosure relates generally to the field of RF and mmWave beamforming transmitters and receivers, and in particular to a device and method for hybrid RF-optical multi-beamforming using correlation-based EVM (Error Vector Magnitude) metrics. Background [0002] Beamforming is employed in order to achieve directional signal transmission and/or reception. In the state of the art, beamforming typically involves supplying signal components to multiple antennas extending across an antenna array, each component having a specific amplitude and phase. This involves the use of phase-shifters and power- combiners. [0003] A drawback of existing techniques is that they tend to consume relatively high amounts of energy, and multi-beam transmission is generally challenging. [0004] A further challenge in such beam-forming systems is that determining channel quality metrics, such as the error vector magnitude (EVM), noise power ratio (NPR) and the normalized mean square error (NMSE), is complex and time consuming. There is thus a need for an improved method and apparatus permitting the measurement and computation of such channel quality metrics. Summary of Invention [0005] According to one aspect, there is provided a multi- beam transmitter comprising: - an antenna array; - a plurality of beam supply paths each configured to supply a corresponding beam; and - - a selection matrix configured to simultaneously couple two or more of the beam supply paths to a plurality of antennas of the antenna array, wherein the selection matrix comprises at least one unit cell coupling first and second beam supply paths among the plurality of beam supply paths to first and second antennas of the antenna array, each unit cell comprising: a first switch coupling a first input node of the unit cell to a first output node of the unit cell; a second switch coupling a second input node of the unit cell to a second output node of the unit cell; a third switch coupling the first input node to the second output node; and a fourth switch coupling the second input node to the first output node of the unit cell. [0006] According to one embodiment, the first and second switches are configured to be controlled by an odd-state control signal and the third and fourth switches are configured to be controlled by an even-state control signal. [0007] According to one embodiment, the selection matrix further comprises a four-port cell comprising: - a first of said unit cells coupling a first pair of input nodes of the four-port cell to a first pair of output nodes of the four-port cell; - a second of said unit cells coupling a second pair of input nodes of the four-port cell to a second pair of output nodes of the four-port cell; - a third of said unit cells coupling the first pair of input nodes to the second pair of output nodes; and - a fourth of said unit cells coupling the second pair of input nodes to the first pair of output nodes. [0008] According to one embodiment, the phase antenna array comprises an integrated patterned lens. [0009] According to one embodiment, each of the unit cells is a differential switch capable of being controlled individually, or in one or more clusters. [0010] According to one embodiment, the multi-beam transmitter comprises a front-end module comprising GaN transistors configured to amplify the beams prior to transmission. [0011] According to one embodiment, the antenna array is implemented by an antenna-in-package solution having a machined base plate. [0012] A multi-beam receiver comprising: - an antenna array; - a plurality of beam reception paths each configured to process a corresponding beam; and - a selection matrix configured to simultaneously couple a plurality of antennas of the antenna array to two or more of the beam supply paths, wherein the selection matrix comprises at least one unit cell coupling first and second of the antennas of the antenna array to first and second beam supply paths among the plurality of beam supply paths, each unit cell comprising: - a first switch coupling a first input node of the unit cell to a first output node of the unit cell; - a second switch coupling a second input node of the unit cell to a second output node of the unit cell; - a third switch coupling the first input node to the second output node; and - a fourth switch coupling the second input node to the first output node of the unit cell. [0013] According to one embodiment, the multi-beam receiver comprises a front-end module comprising GaN transistors configured to amplify received beams. [0014] According to a further aspect, there is provided a method of determining a channel quality metric comprising: - receiving, by a receiver of a beamforming communication system, a first beam transmitted by a beamforming antenna, and generating a first vector by measuring a sequence of N symbols transmitted by the first beam; - calculating, by a processing device of the receiver, the cosine, sine or tangent of an angle between the first vector and a reference vector corresponding to ideal or estimated values of the sequence of N symbols; and - generating, by the processing device, the channel quality metric based on the cosine or tangent of the angle. [0015] According to one embodiment, calculating the cosine, sine or tangent of the angle comprises calculating the cosine of the angle based on the following equation: where is the cross-correlation between the first vector and the reference vector, is the auto- correlation of the first vector, and is the auto- correlation of the reference vector. [0016] According to one embodiment, the channel quality metric is the error vector magnitude of a modulated signal calculated based on the following equation: [0017] According to one embodiment, the channel quality metric is a noise power ratio (NPR) of a multicarrier signal calculated based on the following equation: [0018] According to one embodiment, the channel quality metric is a normalized mean standard error (NMSE) calculated based on the following equation: [0019] According to one embodiment, the method further comprises computing the cross-correlation of distorted and ideal symbols and the autocorrelations of distorted and ideal symbols by the processing device (1612) in the time domain through a time accumulation at the symbol rate, by computing, each time an ideal symbol xn and a distorted symbol yn are input to processing device (1612), the following products which are added to accumulators Cx, Cy and CC: Cx = Cx + xn . xn* Cy = Cy + yn . yn* CC = CC + yn . xn* wherein the EVM is computed from the square of the covariance: C2 = CC.CC*/sqrt(Cx.Cy) as: EVM = sqrt(1/C2-1) [0020] According to one embodiment, the method further comprises determining, by the processing device, the ideal or estimated values of the sequence of N symbols as a nearest symbol, among a constellation of ideal symbols, to each measured symbol. [0021] According to one embodiment, the channel comprises an over the air channel between the transmitted and the receiver. [0022] According to one embodiment, the transmitter is configured to transmit a plurality of sequences of symbols via a beamformer and active array antenna generating a plurality of beams including the first beam. [0023] According to one embodiment, the transmission between the transmitter and the receiver is a MIMO (multiple-input, multiple-output) transmission, both the transmitter and receiver comprising multiple antennas, wherein the processing device is configured to calculate a best quality signal among the received signals. [0024] According to a further aspect, there is provided a receiver in beamforming communication system configured to implement the above method, for example using an SoC, ASIC and/or an FPGA accelerator for implementing correlation-based computations having bounded operators, for example including a state variable bounded between -1 and +1. [0025] According to one embodiment, the multi-beam receiver further comprises a processing device configured to determine a channel quality metric according to the above method. [0026] According to a further aspect, there is provided a correlation-based secure quantum radar sensing solution comprising the above multi-beam transmitter and/or receiver, for example configured to provide correlation-aware wireless communications involving linear and non-linear signal processing operators. [0027] According to one embodiment, the multi-beam transmitter is configured to generate correlated multi-beams for wirelessly powering multiple devices. [0028] According to a further aspect, there is provided a nested X-topology multi-port system for correlated multi-beam wave-shaping comprising the above multi-beam transmitter and/or receiver. Brief description of drawings [0029] The foregoing features and advantages, as well as others, will be described in detail in the following description of specific embodiments given by way of illustration and not limitation with reference to the accompanying drawings, in which: [0030] Figure 1 schematically illustrates a sensing device configured for interferometric correlation-based energy sensing according to an example embodiment; [0031] Figure 2 schematically illustrates a mmWave multi-beam transmitter comprising a lattice-based selection matrix according to an example embodiment of the present disclosure; [0032] FIG.3 schematically illustrates two unit-cells of the lattice-based selection matrix of Figure 2 according to an example embodiment of the present disclosure; [0033] Figure 4 schematically illustrates the lattice-based selection matrix of Figure 2 in more detail according to an example of a 4-port selection matrix; [0034] Figure 5A schematically illustrates a switch-based X- topology architecture unit-cell; [0035] Figure 5B schematically illustrates the switch-based X-topology architecture unit-cell in an even (E) state; [0036] Figure 5C schematically illustrates the switch-based X-topology architecture unit-cell in an odd (O) state; [0037] Figure 6A schematically illustrates the X-topology architecture according to a further example embodiment; [0038] Figure 6B illustrates an example of a switch of the switch-based X-topology architecture of Figures 5A, 5B, 5C and 6A according to an example embodiment; [0039] Figure 7 schematically illustrates a switch-based X- topology architecture unit-cell driving a pair of antennas of a switched antenna array according to an example embodiment; [0040] Figure 8 schematically illustrates an extended four- port X-topology architecture according to an example embodiment; [0041] Figure 9 schematically illustrates a four-port X- topology architecture having parallel control branches according to an example embodiment; [0042] Figure 10 schematically illustrates lattice-based cell-to-cell couplings with parallel control branches according to an example embodiment of the present disclosure; [0043] Figure 11 schematically illustrates a bridge analogy of an X-topology representation according to an example embodiment of the present disclosure; [0044] Figure 12 schematically illustrates an additive orthogonal odd-mode configuration according to an example embodiment of the present disclosure; [0045] Figure 13 schematically illustrates an additive orthogonal even-mode configuration according to an example embodiment of the present disclosure; [0046] Figure 14 schematically illustrates a schematic representation of an X-topology architecture with local grounding in a non-symmetrical left-compensated configuration according to an example embodiment of the present disclosure; [0047] Figure 15 schematically illustrates a schematic representation of an X-topology architecture with local grounding in a non-symmetrical right-compensated configuration according to an example embodiment of the present disclosure; [0048] Figure 16 schematically illustrates a data aided EVM measurement system according to an example embodiment of the present disclosure; [0049] Figure 17 schematically illustrates a non-data aided EVM measurement system according to an example embodiment of the present disclosure; [0050] Figure 18 schematically illustrates a data aided EVM measurement system using a known pseudo-random sequence according to an example embodiment of the present disclosure; [0051] Figure 19 schematically illustrates a data aided EVM measurement system using error corrected symbols according to an example embodiment of the present disclosure; [0052] Figure 20 schematically illustrates a data aided EVM measurement system using a reference receiver according to an example embodiment of the present disclosure; [0053] Figure 21 schematically illustrates an over the air EVM measurement system according to an example embodiment of the present disclosure; [0054] Figure 22 schematically illustrates an over the air EVM measurement system of an active array antenna according to an example embodiment of the present disclosure; [0055] Figure 23 schematically illustrates an over the air EVM measurement system of a multiple-input, multiple-output (MIMO) transmission according to an example embodiment of the present disclosure; [0056] Figure 24 schematically illustrates a transmitter/receiver comprising front-end-modules including advanced auto and cross-correlation signal processing and Multi-Beam MIMO antennas including Correlation-Tuners according to an example embodiment of the present disclosure; [0057] Figure 25 schematically illustrates a transmitter/receiver having a dual-beam front-end-module in a pulsed-mode with moving target-object according to an example embodiment of the present disclosure; [0058] Figure 26 schematically illustrates a MIMO transmitter/receiver having a dual-beam front-end-module in a pulsed-mode with moved target-object according to an example embodiment of the present disclosure; [0059] Figure 27 schematically illustrates a transmitter/receiver having hybrid GaN-FDSOI front-end- modules combined with antenna-array module, co-integrated with patterned 3D-Lens and including correlation-based EVM signal processing for multi-beamforming systems; [0060] Figures 28 and 29 illustrates a CATR measurement setup for multi-beam testing; [0061] Figure 30 illustrates, in perspective view, a hybrid GaN-FDSOI front-end-modules; [0062] Figure 31 schematically illustrates a patterned lens; [0063] Figure 32 is a graph illustrating a measured 4-beam system at 26 GHz; [0064] Figure 33 illustrates beamformer modules combined with patterned lenses; [0065] Figure 34 schematically illustrates aggregated x8 BFN Modules to build a 64-channel beamformer; and [0066] Figure 35 schematically illustrates a system for thermal-EM harvesting. Description of embodiments [0067] Like features have been designated by like references in the various figures. In particular, the structural and/or functional features that are common among the various embodiments may have the same references and may dispose identical structural, dimensional and material properties. [0068] For the sake of clarity, only the operations and elements that are useful for an understanding of the embodiments described herein have been illustrated and described in detail. Unless indicated otherwise, when reference is made to two elements connected together, this signifies a direct connection without any intermediate elements other than conductors, and when reference is made to two elements coupled together, this signifies that these two elements can be connected or they can be coupled via one or more other elements. [0069] In the following disclosure, unless indicated otherwise, when reference is made to absolute positional qualifiers, such as the terms "front", "back", "top", "bottom", "left", "right", etc., or to relative positional qualifiers, such as the terms "above", "below", "higher", "lower", etc., or to qualifiers of orientation, such as "horizontal", "vertical", etc., reference is made to the orientation shown in the figures. [0070] Unless specified otherwise, the expressions "around", "approximately", “substantially” and "in the order of" signify within 10 %, and preferably within 5 %. [0071] It would be desirable to build Front-End-Modules (FEMs) combined with lens-based mmWave AiP (Antenna-in- Package) modules for energy-efficient low-complexity multi- beamforming systems. This will allow scanning directive multi-beam channels with sparsely distributed ports, i.e., with much fewer active electronics channels than in current state of the art phased-array solutions. As a consequence, a much sparser sampling of the “array” and hence much lower consumption: an improvement by a factor of 10 at least. This low-complexity approach will drive next generation of communication and sensing systems eliminating the concept of “elements” in arrays and replacing it by the vision of radiating current flows over a textured surface (Metasurface or Metavolumes) conformal to the patterned optical lens- module that is fed by a limited number of emitting/receiving points (or “ports”). It should be noted that this does not require the use of phase-shifters and power-combiners and that a pair-wise switch-matrix correlator using lattice-based differential switches is sufficient. The pair-wise switch- matrix correlator offers unified mmWave and Baseband correlation-based convolutional processing. [0072] At mmWave frequencies, auto-correlation and cross- correlation functions are used for processing stochastic signals: for stochastic signals, it is established that numerical values of noise amplitudes cannot be specified. Thus, modeling and measuring stochastic signals involves processing energy and power spectra through the extraction of correlation functions. Although Fourier transforms cannot be established rigorously for random processes (infinite energy), they can nevertheless be derived for the autocorrelation and cross-correlation functions which are non-periodic energy signals. The Fourier transforms of the correlation is called power spectrum or spectral density function (SDF). Stochastic noise-aware approaches (see references [1-8] cited below) create natural bridges between correlation formalisms and convolutional techniques. Linking correlation to convolution enables combining information signal theory (IT) and physical information theory (PT) into a unified approach (see references [9-10]). Such a unified approach is established in the general scope of the fluctuation–dissipation theorem (FDT) (see [5]) envisaged as a cornerstone for bridging noise mechanisms with the retrieval of Green’s functions through the formalism of auto and cross- correlation operators. [0073] At Base-Band frequencies, new DSP-based Convolutional-Accelerators are used for real-time EVM (see [11-12]) extraction of multi-beamformer signals in both connectorized and OTA configurations: the approach described here entails a calibration procedure that is independent of the modulation order (e.g., QAM 64, QAM 128 or QAM 1024 all lead to very similar test-time). It is important to notice that the DUT does not need to be connected to any device and that the proposed approach also works in connectorized configuration. [0074] ASIC-embedded Connectors are introduced for co-design and co-integration of adaptive Front-End-Modules including Correlation-Tuners (CT) with Antenna-in-Package (AiP) modules. The CT functionality is used for real-time interferences/couplings mitigation. ASIC-embedded connectors are used for building scalable and conformal Beamforming antenna-arrays for MIMO/Massive MIMO applications. [0075] Figure 1 schematically illustrates a sensing device 100 configured for interferometric correlation-based energy sensing according to an example embodiment. [0076] The device 100 comprises an antenna probe array 102 for interferometric energy sensing, two of the energy probes Epi and EPj being illustrated. The device 100 further comprises a front-end module (FEM) 104, for example adapted for mmWave sensing, the FEM 104 for example being a multi- beam correlator FEM. The device 100 further comprises a pre- FEM circuit 106 comprising for example a plurality of pre-FEM modules (PRE FEM) 108, 110, and a power management circuit 113. Signals from each of the pre-FEM 108, 110 are processed by corresponding RF up/down converters (RF UP/DN CONV) 112, 114, under control of a multiple-input, multiple-output (MIMO) correlation circuit (MIMO CORR) 116. Furthermore, the received signals from the up/down converters 112, 114, which for example correspond to a channel A (CH-A), are processed by a signal processing circuit 118, implemented for example by an ASIC (application specific integrated circuit. Furthermore, energy sensing is for example performed by a sensing circuit 120 based on the signals from the up/down converters 112, 114 and/or from the circuit 118, the sensing circuit for example being configured to perform correlation- based multi-beam energy-sensing, based on multi-bit analog to digital conversion (ADC) adaptive sampling. [0077] An example of the energy computations, based on correlation functions for stochastic fields, performed by the circuit 120 will now be described in more detail. [0078] The cross-correlation function C _AB (τ) of stationary stochastic signals SA(t) and SB(t) is defined by the following equation, where the brackets denote the ensemble average:       [0079] The correlation matrix in the frequency domain can be expressed as a function of the time-windowed signal S _T (t): The superscript † refers to the Hermitian conjugate operation. [0080] For a given frame, the power spectra of the signals can be deduced from the correlation matrix C(t): [0081] Assuming signals and noise contributions are uncorrelated, by applying the Expectation operator E[.], the f ^{ollowing relations can be derived:} where and P _Noise are respectively the signal (channel A) and noise powers. [0082] In (4) S _A and S _B refer to the signals at access terminals (or channels) A and B, and N _^ and N _^ are the noise contributions on channels A and B. This equation clearly shows that uncorrelated noise contributions are totally eliminated. [0083] Thus, the uncorrelated noise power is removed based on the cross-correlation, however the signal power and the correlated noise power are not removed. As a result, the removal of the uncorrelated noise power improves the SNR, and therefore renders possible detecting signals with lower energy levels. [0084] In the time domain, the autocorrelation (AC) of signal (antenna i) and cross-correlation (CC) functions of signals and (antenna j) can be extracted using the following expressions: [0085] The energy density can be written as the sum of electric and magnetic energy densities: [0086] The correlation function of the electric or magnetic field is defined as: where refers to ensemble average (expectation) applied to stochastic variable X and * stands for complex conjugate. [0087] The correlation function of the electric and magnetic energies can be deduced as: [0088] Figure 2 schematically illustrates a mmWave multi-beam transmitter 200 comprising a lattice-based selection matrix according to an example embodiment of the present disclosure. The device 200, or a similar device, can also provide single- beam and/or multi-beam reception. [0089] Figure 2 illustrates in particular a front-end module of the transmitter 200, which for example comprises a patterned 3D lens 202, an optional integrated lens 204, and a antenna array 206, which is for example a phased antenna array, comprising switched antennas 208. Signals generated by the antenna array 206 are for example focused by the lens 204 and by the patterned lens 202 in order to form one or more beams, Figure 2 illustrating an example of the transmission of four beams Beam-1, Beam-2, Beam-3 and Beam-4. [0090] The FEM of Figure 2 is described in more detail in the publication by S.Wane et al. entitled “Energy-Efficient RF- Optics Multi-Beam Systems Using Correlation Technologies: Toward Hybrid GaN-FDSOI Front-End-Modules”, IEEE 2022, the contents of which is hereby incorporated by reference in its entirety. [0091] The patterned lens is for example implemented as described in the publication by X.Lleshi et al. entitled “Wideband Metal-Dielectric Multilayer Microwave Absorber based on a Single Step FDM Process”, 201949 ^th EuMC, pp. 678- 681, and/or as described in the publication by X.Lleshi et al. entitled “Design and Full Characterization of a 3-D- Printed Hyperbolic Pyramidal Wideband Microwave Absorber”, HAL open science, and/or as described in the publication by Sang-Hee Shin et al. entitled “Polymer-Based 3-D Printed 140- 220 GHz Low-Cost Quasi-Optical Components and Integrated Subsystem Assembly”, IEEE Access, Volume 9, 2021, the contents of these three publications being hereby incorporated by reference in their entirety. [0092] The transmitter 200 further comprises a selection matrix 210 configured to supply signals from multiple signal paths to the antenna array 206. For example, the selection matrix is a multi-beam lattice-based selection matrix comprising differential switches, as will be described in more detail below. In the example of Figure 2, the selection matrix receives signals on four paths Path 1, Path 2, Path 3 and Path 4, and directs these signals to four corresponding antennas of the antenna array, although in alternative embodiments, there could be a different number of signal paths, there for example being at least two signal paths. The signals on the signal paths are for example supplied by a signal generation circuit 210, which is for example a cognitive SDR (software-defined radio) baseband circuit. [0093] In some embodiments, an FPGA (field-programmable gate array) or the like is configured to receive signals from the signal generation circuit 210 and to generate control signals for controlling unit cells of the selection matrix 210, as will be described in more detail below. [0094] In some embodiments, the FEM of the transmitter 200 is further configured to perform heat sink power harvesting and/or correlation-based EVM signal processing. [0095] In operation, the selection matrix 210 is for example configured to propagate signals to two antennas 208, for example adjacent antennas, of the antenna array 206 at the same time, the presence of the patterned lens 202 permitting the formation of a beam that is for example relatively narrow, without the activation of additional antennas. Furthermore, in some embodiments, the selection matrix 210 is for example configured to propagate signals to multiple pairs of adjacent antennas 208 of the antenna array 206 at the same time, such that multiple beams are transmitted in unison. [0096] FIG.3 schematically illustrates two unit-cells CELLi and CELLj of the selection matrix 210 of Figure 2 according to an example embodiment of the present disclosure. In the example of Figure 3, the lattice is configured to propagate two channels CH-p and CH-q from the input side (left side of the figure) to the antenna side (right side of the figure) and comprises two-port switches 302 within each unit cell and for providing the interactions between the unit cells. [0097] The unit cells for example permit a scalable and balanced implementation of the selection matrix 210, allowing any number of channels to be propagated. [0098] In particular, an eigen-states approach using X- Topology differential architecture for describing radiation process of an n-port antenna as the combination of n orthogonal modes (states) of radiation. Inter-element coupling/interference can be seen as impedance mismatch in link to these modal states. [0099] Figure 4 schematically illustrates a lattice-based selection matrix 400 configured to implement the selection matrix 200 of Figure 2 in more detail according to an example of a 4-port selection matrix. For example, in Figure 4, four channels CH-p, CH-q, CH-r and CH-s are provided at four input nodes 402, 404, 406 and 408 respectively, and are propagated to four corresponding output nodes 410, 412, 414 and 416, with the particular output nodes being selectable. For example, input nodes 402 and 404 are coupled via a first two- port switch 302 to the output nodes 410 and 412, and to the output nodes 414, 416 via a second two-port switch 302. Similarly, input nodes 406 and 408 are coupled via a third two-port switch 302 to the output nodes 414 and 416, and to the output nodes 410, 412 via a fourth two-port switch 302. [0100] The matrix 400 is thus a four-port switch implemented by a specific arrangement of two-port switches, and an eight- port switch can be formed using the same arrangement of four- port switches, a sixteen-port switch can be formed using the same arrangement of eight-port switches, and so on and so forth, meaning that the solution is scalable to any number of input and output ports. In case of a different number of input and output ports, certain ports can be grounded or shorted together. [0101] Figure 5A schematically illustrates the two-port switch 302 in more detail according to an example embodiment, this switch for example implementing an X-topology architecture unit-cell. The two-port switch 302 for example comprises two input nodes or ports 502, 502 and two output nodes or ports 506, 508. The two-port switch 302 further comprises: a first switch 510 coupling the input node 502 to the output node 506; a second switch 512 coupling the input node 504 to the output node 508; a third switch 514 coupling the input node 502 to the output node 508; and a fourth switch 516 coupling the input node 504 to the output node 506. [0102] The switches 510, 512, 514 and 516 are for example respectively controlled by control signals SO1, SO2, SE1 and SE2. For example, each switch 510, 512, 514, 516 is implemented by one or more transistor switches controlled at its gate by the corresponding control signal, for example via a resistor R _O in the case of switches 510 and 512, or R _E in the case of switches 514 and 516. [0103] In some cases, the switch 302 is configured to be a two-state switch having an even state and an odd state, as shown in Figures 5B and 5C. [0104] Figure 5B schematically illustrates the switch-based X-topology architecture unit-cell 302 in an even (E) state in which the switches 510 and 512 are controlled by a same signal SO, which is low (OFF), thereby deactivating the switches 510, 512, and the switches 514 and 516 are controlled by a same signal SE, which is high (ON), thereby activating the switches 514, 516. [0105] Figure 5C schematically illustrates the switch-based X-topology architecture unit-cell in an odd (O) state in which the switches 510 and 512 are controlled by the same signal SO, which is high (ON), thereby activating the switches 510, 512, and the switches 514 and 516 are controlled by the same signal SE, which is low (OFF), thereby deactivating the switches 514, 516. [0106] For four port systems defined by a 4x4 impedance/admittance matrix (designated by X in the following equations) a nested X-Topology architecture can be extracted: given a block-partitioning XBij (1,j=1,2) of matrix X, the nested X-Topology architecture can be derived using Kronecker product formalism: [0107] A recursive derivation of the two-port X-Topology architecture can be generalized as shown in the following equation: where denotes the Kronecker product. [0108] For symmetrical structures, the formalism of even (E) and odd (O) modal contributions can be adopted. Furthermore, in the context of netlist-oriented extractions, classical lumped elements equivalent circuit representations generally assume global grounding references which fail to reflect full- wave attributes. Major attributes of the X-Topology architectures include possibility, by construction, to account for floating ground references not possible to deal with using classical and T representations. The associated Y or Z matrix are linear combinations of the contributions of these two eigen-states (designated with the subscript Even/Odd for symmetrical two-port systems). [0109] In addition, the X-Topology architecture can be linked to the following analogies: - Hadamard71 transform: Beyond similarities with the FFT Butterfly algorithm, formal link can be established between Hadamard-Walsh transforms and the X-Topology architecture implementation. - Wave-Digital-Filter (WDF): in reference to classical filter networks, generally in lossless assumption, inserted between resistive terminations. The analogy between WDFs and their filters counterpart is based on wave quantities rather than on voltages and currents as signal parameters, hence the reference to Wave in the terminology of WDF. - Bridge-Balanced architecture classically used in electronic systems for stabilization or compensation purposes. The Lattice structure is also used in Filter design including use of piezoelectric resonators. [0110] The eigen-states contributions, for a two-port impedance/admittance matrix (symmetric) can be determined by a diagonalization procedure such as: where the matrix P gives the eigen-states excitation weight with the following structure:   [0111] The columns of P can be interpreted as excitation states to be impressed with proper magnetic/electric walls boundary condition at the symmetry planes. For two-port symmetrical structures the proposed synthesis methodology can be understood as a generalization of Bartlett’s theorem where open and short circuit conditions are referred to magnetic and electric walls conditions. For multi-port systems the eigen-states excitation weight can be rigorously determined from Gram-Schmidt orthogonalization process. [0112] Figure 6A schematically illustrates the X-topology architecture 302 according to a further example embodiment in which the switches are represented as branches. [0113] When a differential input signal is applied between nodes a ₁ and a ₂ , the differential output signal (between nodes b ₁ and b ₂ ) is perfectly equal to zero if the four branch impedances are equal. Let branches 1 and 2 be the direct branches (D) of admittance Y _D , and branches 3 and 4 be the crossed branches (X) of admittance Y _X . [0114] Figure 6B illustrates an example of a switch of the switch-based X-topology architecture of Figures 5A, 5B, 5C and 6A according to an example embodiment. Each branch impedance (Figure 6A) or switch is implemented, in the example of Figure 6B, using a series stack of three n-channel transistors 602, 604 and 606, for example NFET (n-channel field-effect transistors), for example having 20 nm gate length, and targeting above 20 dBm of power handling. The transistors are for example controlled at their gates by a voltage V _G , and at their back gates by a voltage V _BG . [0115] In the ON state, branches 1 and 2 are turned on by applying a front gate bias of V _G = 0.9 V and a back-gate bias of V _BG = 3 V while branches 3 and 4 are turned off with V _G = -0.9 V and V _BG = 0 V. In that case Y _D-ON is high and Y _X-OFF is low and Y _21-ON reduces to -Y _D-ON/2 . In the OFF state, all branches are turned off. Then, Y _21-OFF becomes (Y _X-OFF - Y _D-OFF )/2. If the branch impedances are equal, destructive interference occurs between the signal passed through the D branches and the anti- phase signal passed through the X branches, and Y ₂₁ = 0. [0116] Figure 7 schematically illustrates the switch-based X-topology architecture unit-cell 302 driving a pair of antennas 704 of a switched antenna array 702 according to an example embodiment. Figure 7 illustrates in particular a two- port dual-beam (beams Beam-i and Beam-j among 64 channels) driven by channels CH-i and CH-j using unitary antenna elements separated by a variable distance d. [0117] Figure 8 schematically illustrates an extended four- port X-topology architecture according to an example embodiment. In the example of Figure 8, a first input channel comprises an input signal IN-1 and a ground input GND-IN-1, a second input channel comprises an input signal IN-2 and a ground input GND-IN-2, a first output channel comprises an output signal OUT-1 and a ground output GND-OUT-1, a second output channel comprises an output signal OUT-2 and a ground output GND-OUT-2. In the example of Figure 8, the two-port unit cell linking the inputs IN-1 and GND-IN-1 to the outputs OUT-1 and GND-OUT-1 and the two-port unit cell linking the inputs IN-2 and GND-IN-2 to the outputs OUT-2 and GND-OUT-2, are configured to be in the odd state, while the two-port unit cell linking the inputs IN-1 and GND-IN-1 to the outputs OUT-2 and GND-OUT-2 and the two-port unit cell linking the inputs IN-2 and GND-IN-2 to the outputs OUT-1 and GND-OUT-1, are configured to be in the even state. [0118] Figure 9 schematically illustrates a four-port X- topology architecture 900 having parallel control branches according to an example embodiment. The architecture 900 is similar to that of Figure 4, and includes input ports/nodes 902, 904 coupled to output ports/nodes 906, 908 via a two- port unit cell 302, input ports/nodes 910, 912 coupled to output ports/nodes 914, 916 via another two-port unit cell 302, and two further unit cells 302 respectively coupling the input ports/nodes 902, 904 to the output ports/nodes 914, 916 and the input ports/nodes 910, 912 to the output ports/nodes 906, 908. Furthermore, an impedance circuit Y _F provides parallel branches between the input nodes 902 and 912, and between the input nodes 904 and 910, and a further impedance circuit Y _F provides parallel branches between the output nodes 906 and 916, and between the output nodes 908 and 914. [0119] Figure 10 schematically illustrates lattice-based cell-to-cell couplings 1000 of two four-port unit cells 900 of Figure 9 with parallel control branches according to an example embodiment of the present disclosure. [0120] For example, a circuit-based nodal approach shown in Figures 11 to 13 herein after creates a link to broadband SPICE representations suitable for simultaneous time and frequency domain representations. [0121] Figure 11 schematically illustrates a bridge analogy of an X-topology representation according to an example embodiment of the present disclosure. [0122] Figure 12 schematically illustrates an additive orthogonal odd-mode configuration according to an example embodiment of the present disclosure. [0123] Figure 13 schematically illustrates an additive orthogonal even-mode configuration according to an example embodiment of the present disclosure. [0124] To extend the eigen-state formalism to non-symmetrical multi-port systems, a modified X-topology topologies, shown in Figures 14 and 15, are proposed where a coupling branch is introduced. [0125] Figure 14 schematically illustrates a schematic representation of an X-topology architecture with local grounding in a non-symmetrical left-compensated configuration according to an example embodiment of the present disclosure. [0126] Figure 15 schematically illustrates a schematic representation of an X-topology architecture with local grounding in a non-symmetrical right-compensated configuration according to an example embodiment of the present disclosure. [0127] Nodal SPICE-based analysis is conducted to extract Z/Y/S representations of the coupled X-Topology Four-Port system based on the following equations (admittance description without loss of generality): [0128] The beam-forming transmitter and receiver described herein advantageously incorporates a processing device, for example implemented by the circuit 120, for measuring and computing one or more channel quality metrics, as will now be described in more detail below. These techniques could equally be applied in other single beam or multi-beam transmission systems. [0129] The following descriptions presents the correlation- based measurement of Error Vector Magnitude (EVM), as well as other channel quality metrics, in various set-ups including over the air (OTA) and for array antennas (with beamforming) and for multi-antenna (MIMO). C.1 Correlation based EVM laboratory measurement [0130] EVM is defined as the root mean square of the errors between distorted symbols and ideal symbols. Before measuring the errors, the ideal signal and the distorted signal must be synchronized, and the complex gain (amplitude and phase) must be optimized to minimize the errors (and the EVM) on the symbols. [0131] During laboratory measurement, it is supposed that there is access to the ideal symbols, and the so-called data aided EVM can be measured. If we have also access to the transmitted signal and to the local oscillator, the receiver is simplified as it is based on ideal and received signals cross-correlation after down-conversion. [0132] Figure 16 illustrates a first example of a transmission system 1600, and illustrates in particular a data aided EVM measurement system according to an example embodiment of the present disclosure. The system 1600 for example comprises a symbol sequence generator 1602 (SYMBOL SEQ GEN) configured to generate a symbol sequence, which is for example an ideal symbol sequence. For example, the generator 1602 is implemented by a processing device that forms part of the transmitter or of a separate computer. The system further comprises a transmitter 1604 (TRANS), implemented for example by the transmitter 200 of Figure 2, a channel 1606 (CH), and a receiver 1608 (REC), implemented for example by a circuit similar to the transmitter 200 of Figure 2. In the example of Figure 16, channel 1606 is for example implemented by a cable linking the transmitter 1604 to the receiver 1608. The receiver 1608 is for example configured to supply distorted signals received over the channel 1606 to an EVM computation circuit 1612, which is for example implemented by a processing circuit forming part of the receiver 1608, or by a separate computing device, such as an FPGA or ASIC. The circuit 1612 also for example receives the symbol sequence directly from the generator 1602, and is configured to compare the ideal and distorted symbols and to compute a data-aided EVM estimation, as will now be described in more detail. This estimation for example permits a characterization of distortion introduced in the transmission and/or reception chains, for example by the power amplifier that is present in the transmission chain of the transmitter 1604 and/or the low-noise amplifier (LNA) that is present in the transmission chain in the receiver 1608. [0133] On each symbol, the error is the difference between the measured symbol y _n and the reference symbol x _n multiplied by the optimum gain γ that minimizes the error: y _n =γ x _n +e _n T ^{he square of the error on each symbol is:} The root mean square of this error is: The optimum gain is given by the orthogonal projection of the vector y=[ y ₁ ,y ₂ ,..., y _N ] on the vector x=[ x ₁ ,x ₂ ,. _. ., x _N ] as: Replacing the gain γ by its optimum value in the rms error gives: [0134] The EVM is the ratio of σ to the quadratic average amplitude of the signal multiplied by the gain: In this expression of EVM, we recognize the inverse of the cosine of angle ^^ between vectors x=[ x ₁ ,x ₂ ,. _. .,,x _N ] and y=[ y ₁ ,y ₂ ,..., y _N ]: This cosine is the cross-correlation of both signals normalized by the product of the signals’ quadratic amplitudes. It is also called the correlation coefficient or Pearson coefficient. [0135] The EVM is the tangent of this angle: [0136] Advantageously, the circuit 1612 is configured to compute the EVM base on this equation. Indeed, by this shortcut, the EVM is obtained directly from ideal and measured symbols cross-correlation and from autocorrelations of ideal symbols and measured symbols. Tt is not necessary to compute the differences between ideal and measured symbols, which may increase numerical errors. [0137] The EVM maximum value is infinite for two uncorrelated signals or if one signal is only noise. [0138] It will be appreciated by those skilled in the art that the same formula can be applied to the spectra instead of the symbols. Furthermore, it can also be used to compute the noise power ratio (NPR) of a multicarrier signal instead of the EVM of a modulated signal. The matched filter may be different. [0139] Furthermore, the NMSE is for example obtained from the difference between power normalized ideal and measured sequences of symbols after computing the same optimum gain. For example, it is obtained as: [0140] A coarse value of the NMSE can be computed directly on the RF signal instead of the sequences of symbols. The same formulas are used but x _n and y _n are samples of the RF waveforms or of their spectra. [0141] The NMSE maximum value is 2 when the received signal contains only noise (or is uncorrelated to the reference signal). [0142] Furthermore, if the reference signal is projected on the received signal and the corresponding error is divided by the reference signal, the EVM is obtained as the sine of the same angle instead of the tangent. This EVM maximum value is 1 or 100% (for two uncorrelated signals or if one signal is only noise) instead of infinite. [0143] Furthermore, as the optimum gain is obtained by orthogonal projection of y on x, the vector of symbol errors is orthogonal to x. [0144] The computation, by the circuit 1612, of the optimal gain by orthogonal projection also defines the error as the part of the distorted signal y that is not correlated with the ideal signal x. [0145] Because of this orthogonality, it is possible to derive both of the following equalities: [0146] According to a further variant, the computation of the cross-correlation of distorted and ideal symbols and the autocorrelations of distorted and ideal symbols is performed by the circuit 1612 in the time domain through a time accumulation at the symbol rate. For example, each time an ideal symbol xn and a distorted symbol yn are input to the circuit 1612, for example implemented by FPGA, the the following products are for example computed and added to accumulators: - Cx = Cx + xn . xn* - Cy = Cy + yn . yn* - CC = CC + yn . xn* [0147] Periodically, such as at the end of each frame, the EVM is for example computed from the square of the covariance: C2 = CC.CC*/sqrt(Cx.Cy) as: EVM = sqrt(1/C2-1) [0148] The circuit 1612 is then for example configured to reset the three accumulators to 0 in order to obtain one value of EVM for each period or frame. Alternatively, rather than being reset, the three accumulators can be left running to average the EVM over a plurality of periods or frames. [0149] The main part of the EVM computation is thus reduced to three products and three additions at the symbol rate to be implemented for example by an FPGA or ASIC. C.2 Non-data aided EVM measurement [0150] If the testing setup does not permit a copy of the ideal symbols to be supplied to the circuit 1612, the signal receiver is for example configured to demodulate the signal, extract the distorted symbols, and compare them to the symbols of an ideal constellation (for the same modulation). The constellation is for example the same as the one used for transmission. This measurement will be called non-data aided EVM. [0151] Figure 17 schematically illustrates a non-data aided EVM measurement system 1700 according to an example embodiment of the present disclosure. The system 1700 comprises many of the same elements as Figure 16, and these elements are labelled with like reference numerals and will not be described again in detail. The system 1700 no longer comprises a link between the generator 1602 and the computation circuit 1612, and instead a symbol constellation 1702 (SYMBOL CONSTELLATION), for example stored in a memory device, which could form part of the computation circuit 1612, is supplied. [0152] In contrast to the data aided method, in the non-data aided method each distorted symbol is compared to the nearest ideal symbol, even if this symbol is not the correct one that was transmitted. However, whenever possible, it is more rigorous to have a copy of the ideal symbols and to perform a data aided measurement. This can be done in many ways, such as the 3 following cases: 1. By transmitting a known pseudo-random sequence that can also be generated in the receiver or the EVM measurement device. 2. By correcting the received bits in the receiver after error code correction (ECC) and using regenerated ideal symbols instead of using the received symbols. The error is much smaller but not zero. 3. By receiving the ideal sequence of symbols without errors through a reference receiver having a much better signal to noise ratio, such as one with a larger antenna and/or better positioned with respect to the transmitter. [0153] Figure 18 schematically illustrates a data aided EVM measurement system 1800 using a known pseudo-random sequence according to an example embodiment of the present disclosure, and corresponds to a solution according to case 1 above. In particular, with respect to the system 1700, the symbol constellation 1702 is replaced in Figure 18 by an identical system sequence generator 1802 (IDENTICAL SYMBOL SEQ GEN), which is a circuit like the circuit 1602 that is configured to generate the same symbol sequence as the circuit 1602. [0154] Figure 19 schematically illustrates a data aided EVM measurement system 1900 using error corrected symbols according to an example embodiment of the present disclosure and corresponds to a solution according to case 2 above. In particular, with respect to the system 1700, the symbol constellation 1702 is replaced in Figure 19 by an error correction circuit 1902 (ERROR CORRECTION) configured to receive the distorted symbols 1610, and symbol sequence regeneration circuit 1904 (REGEN SYMBOL SEQ) configured to compute and supply to the computation circuit 1612 a regenerated symbol sequence that can be used in place of the ideal symbol sequence. [0155] Figure 20 schematically illustrates a data aided EVM measurement system using a reference receiver according to an example embodiment of the present disclosure, and corresponds to a solution according to case 3 above. In particular, with respect to the system 1700, the symbol constellation 1702 is replaced in Figure 20 by a channel 2002 (BETTER CH), which is for example better than the channel 1606, and a reference receiver 2004 (REF REC) configured to compute and supply to the computation circuit 1612 the received symbol sequence (RECEIVED REF SYMBOLS) 2006 that can be used in place of the ideal symbol sequence. [0156] The receiver that is used for the measurement is for example configured to ensure that the correct demodulation of the signal, particularly the frequency down-conversion (even in case of Doppler in mobile cases) to complex baseband signal (IQ) and the clock symbol frequency and phase recovery for the best sampling of the signal to give the symbols. [0157] Generally, this is done by commercial EVM measuring devices. [0158] If the EVM is too high, the receiver may not synchronize at all. C.3 Over the air measurement of EVM [0159] The EVM is measured on a receiver using a true transmission channel with antennas and propagation. It can be done in any of the non-data aided (using ideal constellation only) or data aided (using known, error corrected or received sequence of ideal symbols) cases seen in the previous section. [0160] The EVM can be measured as a function of distance and orientation of both antennas. [0161] It also depends on reflections and losses due to environment. [0162] In this simple case, the transmitter and receiver pair EVM can be measured separately from the antennas gain, free space loss and environment multipaths. [0163] OTA measurement is not always necessary, it gives a global performance but depending on many environment parameters. [0164] Figure 21 schematically illustrates an over the air EVM measurement system 2100 according to an example embodiment of the present disclosure. The system 2100 is similar to the system 1700 of Figure 17, except that the channel 1606 is replaced by an antenna 2102, having an associated antenna diagram (ANTENNA DIAGRAM), coupled to or forming part of the transmitter 1604, and an antenna 2104, coupled to or forming part of the receiver 1608. C.4 Measurement of EVM on an antenna array [0165] In that case that only one beam is generated by the active or passive antenna array of the transmitter 1604, the measurement is for example as described above in relation with Figure 21. [0166] In normal operation, the array antenna will transmit many beams at the same time. Two phenomena will impact the EVM measured on each beam: 1. The beam-to-beam isolation is not perfect so that the receiver noise will increase due to the adjacent beams power, particularly if they use the same bandwidth or an adjacent bandwidth. 2. In an active array antenna, all the signals will be mixed (with different phases) in all the amplifiers and intermodulation of all the signals may be transmitted in the measured beam bandwidth and direction. [0167] This type of measurement can only be done over the air. [0168] Figure 22 schematically illustrates an over the air EVM measurement system 2200 of a beamforming and active array antenna 2202 (BF + ACTIVE ARRAY ANTENNA) according to an example embodiment of the present disclosure. In particular, with respect to the system 2100 of Figure 21, the system 2200 comprises a plurality of symbol sequence generators 1602, 1602’, 1602” (SYMBOL SEQ GEN) and a plurality of corresponding transmitters 1604, 1604’, 1604” (TRANS), each supplying a corresponding signal for transmission as a beam to the active array antenna 2202, which for example comprises beamforming capabilities. For example, the active antenna array 2202 is implemented by the transmitter 200 of Figure 2. [0169] The active array antenna 2202 for example generates beams of which one corresponds to the measured beam diagram 2204, and two adjacent beams correspond to interfering beam diagrams 2206, 2208. There is equally an associated intermodulation beam diagram 2210. [0170] To correctly measure the effect of both of the above on the EVM, according to the embodiments described herein, the other beams are for example configured to transmit sequences of symbols that are un-correlated with the measured beam sequence of symbols and with the other interfering beams. As such, they are accounted as noise or error in the EVM computation code. [0171] For example, the transmitters 1604, 1604, 1604” are configured to load all of the interfering beams with uncorrelated pseudo-random filtered white Gaussian noise. The filter for example has the same bandwidth and shape as the shaping filter in the nominal modulation of the interfering beams. This is generally a worst case corresponding to high order and high PAPR (peak-to-average power ratio) modulations on all interfering beams. [0172] The computation, by the circuit 1612, of the cross- correlations between all sequences of symbols or noises in the measured and interfering beams and comparison with their autocorrelations will give an idea of the error made in the measurement. [0173] One possible problem is that, in normal operation, beams may be synchronized and transmit identical or nearly identical slot headers or frame headers at the same time. These are for example not used in the EVM measurement, which is normally defined on the payload of slots and frames. [0174] One EVM measurement is for example made on each beam, as beams are generally independent and sent to different users at different positions. [0175] The worst case can be when different beams are in the same frequency channel or when beams in different frequency channels go to users in the same direction from the antenna. C.5 Measurement of EVM on MIMO transmissions [0176] In MIMO transmissions, both the transmitter and the receiver use multiple antennas to improve the channel capacity in the presence of multipath. Each transmit antenna T _i sends a different signal x _i . [0177] Each receive antenna R _j receives all these signals with echoes, mixing and noise and produces a signal y _j . [0178] Figure 23 schematically illustrates an over the air EVM measurement system 2300 of a multiple-input, multiple- output (MIMO) transmission according to an example embodiment of the present disclosure. The system 2300 is similar to the system 2200 of Figure 22, except that the beamforming and active antenna array 2202 is replaced by multiple transmitters 2302, 2304, 2306 respectively coupled to the output signals x ₁ and x ₂ to x _I respectively of the transmitters 1604, 1604’, and 1604”, and there are corresponding antennas 2308, 2310 and 2312 coupled to the receiver 1608 receiving corresponding signals y ₁ and y ₂ to y _J . [0179] The total transmitted power is constrained to a maximum value: If we define x, the vector of transmitted signals and the covariance matrix of transmitted signals: Then the trace of this covariance matrix is constrained by the total power: [0180] The received signals result from the transmitted signals by a multiplication with a linear matrix H representing the environment. [0181] Generally, the matrix ^^ is supposed to be scalar (non- dispersive environment) but it could also be composed of functions of frequency ℎ _{j i} (f), one frequency response for each term of the matrix. In that case the matrix is applied to the spectra of transmitted signals. y=H x+n or  [0182] The matrix could also be composed of impulse responses and the result would be computed through convolutions with the signals in time domain. [0183] The matrix may also change slowly as a function of time. [0184] If transmitted signals x are uncorrelated or independent, each signal is amplified by one power amplifier for each transmit antenna. There are no inter-modulations between signals, only distortion (compression) of each transmitted signal. [0185] The receiver 1608 is configured to process all received signals and extract as much as possible each transmitted signal with as low as possible interference from other signals. For this, the receiver 1608 is for example configured to measure matrix H, which is done by comparing headers for each signal with a reference. [0186] The EVM measurement is for example made over the air and the transmitter and receiver for example apply the exact algorithms that will be applied in the system (or described in the standard). It may be possible to measure EVM separately for each transmitted signal, but an average on all signals sent to the same user is likely to be adequate. C.5.1 Space-time coding [0187] Space-time coding can be used to help the receiver in decoding received signals. This decreases the channel capacity but increases the robustness of the transmission by simplifying the real time computation of matrix H. The simplest example is Alamouti scheme with two transmit antennas and^^ is: and is: C.5.2 Precoding [0188] In case of precoding, transmitted signals x are computed from the wanted transmitted data s to decrease the interference between received signals. For this, the transmitter 1604 for example knows the response matrix H and computes a right singular matrix V that transforms H: x= Vs y=Hx+n= HVs +n and the receiver 1608 applies to the received signals a computed shaping matrix U ^H (transpose conjugate of U) that gives a diagonal matrix ∑. z= U ^H HVs+U ^H n=∑s+U ^H n [0189] This is exact only for linear distortion. It results in non-linear distortion that impacts the EVM measurement because each transmitted signal of x contains many components of ^^and transmitted signals x are no longer uncorrelated. Each received signal of y contains non-linear distortion that depend on the other signals of s. C.5.3 Multi-user MIMO [0190] In multi-user MIMO, signals may be sent by the same transmitter to more than one user. This is generally the case for telephony base stations. C.5.4 Combination of MIMO and active antenna [0191] In the case of multiuser MIMO, the antenna may be divided into four or more sub-arrays and each sub-array is used for one user only. [0192] The subarray antenna used for one user can be configured as a MIMO antenna or as an active antenna to generate a beam in the direction of the user or to decrease interference to other users. [0193] MIMO is used generally at lower frequency bands where multipath is large and beams cannot be directive and active antenna is used at higher frequency bands where multipath is lower and beams can be directive. C.6 EVM measurements - Conclusion [0194] The EVM measurement method using correlations of transmitted and received symbols can be adapted to over the air measurement by performing the measurement with the transmitter, receiver, antennas and as much as possible a realistic environment. The transmitter is for example loaded by independent (uncorrelated) data streams for all users. These transmissions will appear as additive noise in the measured channel. Common aspects [0195] Figure 24 schematically illustrates a transmitter/receiver 2400 comprising a front-end-module 2402 including advanced auto and cross-correlation signal processing and Multi-Beam MIMO antennas including Correlation-Tuners according to an example embodiment of the present disclosure. For example, the transmitter/receiver 2400 is capable of transmission and reception, and comprises elements in common with the transmitter 200 of Figure 2, including the patterned 3D lens 202 and the antenna array 206, which form part of the front-end-module 2402. The front- end-module 2402 forms part of a module 2404 further comprising the pre-FEM 106 of Figure 1, and the up/down converters 112, 114 and correlator (CORR) 116. While not illustrated in Figure 24, the module 2404 may additionally comprise the selection matrix 210 of Figure 2. The transmitter/receiver 2400 also for example comprises the signal processing circuit 118 and the sensing circuit 120 of Figure 1. The circuit 118 is coupled, for example via a high-speed input/output circuit 2406 to a digital/analog conversion stage 2408, comprising digital to analog converters (DACs) for converting digital signals from the circuit 118 into analog signals and analog to digital converters (ADCs) for converting analog signals from the up/down converters 112, 114 into digital signals. Furthermore, a mixing stage 2410 is provided between the digital/analog stage 2408 and the up/down conversion stage. For example, a mixer 2412 and local oscillator 2414 is provided per channel. As represented by a dashed arrow 2416, from the transmission and reception chains between the circuit 118 to the up/down converters 112, 114 comprise analog-digital mixed-signals. [0196] The pre-FEM 106 for example comprises power amplifiers (PA) for amplifying signals to be transmitted, and low noise amplifiers (LNA) for amplifying received signals. Furthermore, the pre-FEM 106 for example comprises a TX-RX tuner (TX-RX TUNER), one per channel. [0197] Figure 25 schematically illustrates a transmitter/receiver 2500 having a dual-beam front-end-module in a pulsed-mode with moving target-object according to an example embodiment of the present disclosure. In particular, the transmitter/receiver 2500 comprises the module 2404 of Figure 24, coupled to a pulsed mode TX-X based correlated beams processor 2502 (TX-RX PROCESSOR). The transmitter/receiver 2500 is for example capable of operating in a pulsed mode for communicating with a moving target 2504 (TARGET) using dual-beam correlations. [0198] Figure 26 schematically illustrates a MIMO transmitter/receiver system 2600 comprising having a dual- beam front-end-module in a pulsed-mode with moved target- object according to an example embodiment of the present disclosure. In particular, the system 2600 for example comprises two of the transmitter/receivers 2500 of Figure 25 (one labelled 2500 and the other 2500’) configured to communicate with each other via dual-beam correlations. [0199] Figure 27 schematically illustrates a transmitter/receiver 2700 having hybrid GaN-FDSOI front-end- modules combined with antenna-array module, co-integrated with patterned 3D-Lens and including correlation-based EVM signal processing for multi-beamforming systems. For example, the solution of Figure 27 comprises the front-end module 2402 of Figure 24, the front-end module 104 of Figure 1 comprising a multi-beam correlator, and the pre-FEM 106 of Figure 1, in which the front-end modules comprise GaN-based FEMs 2702 and 2704 (GaN-BASED FEM). For example, the front-end modules 2702 and 2704 each comprise one or more GaN transistors configured to amplify the respective beams prior to transmission. [0200] Figures 28 and 29 illustrates a CATR (compact antenna test range) measurement setup 2800 for multi-beam testing. Figure 28 illustrates in particular a reflector 2802, a mmWave correlator module 2804 with an antenna array 2808 and a 3D lens 2806, and an antenna under test (AUT). Figure 29 illustrates a measurement setup 2900 comprising a source 2902, a reflector 2904 and a DUT. For example, the DUT is coupled to measurement equipment 2906, which is for example an oscilloscope and/or VNA (vector network analyzer), which receives a signal B-RX corresponding to a beam received by the DUT, and to a computer 2908 (COMPUTER). The measurement equipment 2906 is also for example coupled to the source 2902, an in particular two signals P1-TX-Polar+θ° and P2-TX-Polar+θ° are supplied in order to generate a beam. The computer 2908 is for example coupled to the measurement equipment 2906. The setups of Figures 28 and 29 can for example be used for the EMV measurement described herein. [0201] Figure 30 illustrates, in perspective view, a hybrid GaN-FDSOI front-end-module with antennas in X and Y topologies. The antennas are for example spaced by 15 mm in one example. [0202] Figure 31 schematically illustrates the patterned lens 202 described herein in more detail in side view, the lens 202 providing for example a high gain beam 3102 and a low gain beam 3104. [0203] Figure 32 is a graph illustrating a measured 4-beam system at 26 GHz generated for example using the transmitter 200 of Figure 2. Each curve 3201 to 3208 in Figure 32 corresponds to a different port among eight ports, and it can be seen that beams are generated at an angle of between +10° and -10° and with an angular granularity of only between 2° and 3°. [0204] Figure 33 illustrates a transmitter 3300 formed of a stack 3302 of aggregated beamformer modules 3304, each of the beamformer unitary modules 3304 having an antenna array or metasurface 3206, which is for example a linear array, and combined with the patterned lens 202. Thus, Figure 33 represents a manner in which a transmitter having a 2D antenna array can be constructed from a stack of modules. [0205] Figure 34 schematically illustrates aggregated x8 BFN Modules to build a 64-channel beamformer. In particular, Figure 34 illustrates the aggregated beamformer modules 3304 in more detail, each module 3304 for example comprising a beamformer network 3402 (BFN) receiving an input signal from a splitter 3404, which receives a common input signal IN. Each module 3302 further comprises a DC-DC and ABB supply circuit 3406 and a LVDS (low voltage differential signaling) converter and control circuit 3408, each module and the circuits 3406 and 3408 for example being controlled and powered by an external control and power circuit 3410 (CTRL + POWER). Each beamformer network is for example coupled to a corresponding linear array of antennas, for example 8 antennas, and thus by stacking eight modules, a 64-channel beamformer can be realized. [0206] Figure 35 schematically illustrates a system 3500 for thermal-EM harvesting, comprising wafer-level chip scale packaging integrated circuits (WLCSP IC) mounted on a multilayer PCB 3502 comprising the stack 3302 of beamforming modules 3304 and hosting an antenna array system 3504 on its opposite side. Heat pipes 3504 for example travers a substrate 3506, which in some embodiments, as represented on the right of the figure, may be curved, and on which the WLCSP ICs are for example fixed with thermal glue 3508. [0207] Various embodiments and variants have been described. Those skilled in the art will understand that certain features of these embodiments can be combined and other variants will readily occur to those skilled in the art. [0208] Finally, the practical implementation of the embodiments and variants described herein is within the capabilities of those skilled in the art based on the functional description provided hereinabove. [0209] The following paragraphs describe further example embodiments of the present disclosure. [0210] Example 1: Front-End-Modules (FEMs) combined with patterned (including inhomogeneous material properties and shaping) lens-based mmWave AiP (Antenna-in-Package) modules for energy-efficient low-complexity multi-beamforming systems exploiting sparse Eigen-State formulation, in which: - Nested broadband X-Topology Differential Switches are for example introduced for optimal selection and control of beamforming states; - 180 degrees phase shifting through both the unitary X- Topology cell and its nested multi-port extension is for example natively provided. Native 180 degrees phase shifting can be used in combination with broadband lumped RLC (Resistance, Inductance, Capacitance) for relaxing constraints and requirements on the range of phase-shifting devices; - Broadband mutual-inductance free RLC-based X-Topology equivalent circuit canonical representation of multi-port radiators is for example applied; and - Broadband RLC-based X-Topology equivalent circuit representation is for example used for building correlation tuners. [0211] Example 2: According to some embodiments, the lens- based mmWave multi-beam beamformers, in offering sparse properties in both time and angular dimensions, provide significantly reduced signal processing which result in a small set of correlation-based RF measurements for channel estimation of MIMO systems: - Mosaic partitioning strategies exploiting the sparsity of MIMO correlation matrix, open new possibilities for combining multiple arrays into a full array state (FAS) to form one single beam, or for using them to form separate beams in the sub-array state (SAS); - partitioning strategies for Lattice-based differential switches with channels either considered individually, all combined, or partially combined (i.e. grouped or clustered) in accordance with how the sub-arrays may be merged to form simultaneous beams; and - Correlation-based EVM extraction of Single-Beam and Multi- Beam systems using ASIC or FPGA accelerators. The EVM of one or more signals is measured in the presence of interference and intermodulation from other signals and beams: The new EVM hardware implementation in SoC/ASIC or FPGA for example uses correlation-based accelerators handling bounded operators (with state variable between -1 and +1 providing robust numerical computation); [0212] Example 3: Co-integration of Beamformers and AiP (Antenna-in-Package) with Correlation-Tuners (CT) for optimal Multi-Beamforming EVM. [0213] Example 4: Hybrid GaN/SiGe or GaN/FDSOI for energy- efficient high TX/RX dynamic range for multi-beamforming with digital control and embedded EVM metrics evaluation in real- time. [0214] Example 5: Hybrid GaN/SiGe or GaN/FDSOI with Thermal- EM Harvesting. [0215] Example 6: The AiP module is reinforced with a base plate machined with a mechanism to allow the extremely high engage and disengage force to be overcome for MIMO/Massive- MIMO systems. [0216] Example 7: the example 1, 3 or 6, combined with interferometric synchronization using an RSRP (Reference Signal Receive Power – both in UL (uplink) and DL(Down-Link) (or SS-RSRP: Sync Signals RSRP) parameter in UE OTA Testing for assessing signal quality. Based on reference signal received power (RSRP) measurements, beamformed downlink (DL) reference signals (RSs) are transmitted by multiple sources or base stations (BSs) and measured by user equipment (UE) employing receive beamforming. The so-obtained beam-RSRP (BRSRP) measurements are fed back to the sources or BSs where the corresponding directions of departure (DoDs)are deduced for correlated multiple beams. [0217] Example 8: Example 7, combined with channel state information including RRC (Radio Resource Control) protocol measurement accounting for determining correlation and channel calibration matrix. [0218] Example 9: Example 7 or 8, with EVM testing of Beamformers accounting for Beam-to-Beam correlations. [0219] Example 10: Example 1, 2, 3 or 4, with classical beamformers for hybrid analog-digital multi-beamformers. [0220] Example 11: Example 1, 2, 3 or 4 for correlation-based secure quantum radar [15–23] sensing-related applications enabling correlation-aware wireless communication systems backed-up by linear and non-linear signal processing operators of technical references [24-38]. [0221] Example 12: Use of Correlated Multi-Beams for Wireless-Powering of multiple devices. [0222] Example 13: Use of Nested X-Topology Multi-Port system for Correlated Multi-Beam Wave-Shaping. [0223] Example 14: Use of Multi-Beam Correlations for building Invisible-Electronics [39] (transparency in the sense of not-detectable). [0224] Example 15: A mmWave multi-beam transmitter comprising: - an antenna array; - a plurality of beam supply paths (Path 1, Path 2, Path 3, Path 4) each configured to supply a corresponding beam; - a multi-beam lattice-based selection matrix configured to simultaneously couple two or more of the beam supply paths to a plurality of antennas of the antenna array. [0225] Example 16: The multi-beam transmitter of example 15, wherein the antenna array comprises an integrated lens or reflector. [0226] Example 17: The multi-beam transmitter of example 16, wherein the phase antenna array comprises an integrated patterned lens. [0227] Example 18: The multi-beam transmitter of any of examples 15 to 17, wherein the multi-beam lattice-based selection matrix comprises lattice-based differential switches capable of being controlled individually, or in one or more clusters. [0228] Example 19: The multi-beam transmitter of any of example 15 to 18, comprising MIMO correlators. [0229] Example 20: The multi-beam transmitter of any of examples 15 to 19, comprising a front-end module comprising GaN transistors configured to amplify the beams prior to transmission. [0230] Example 21: The multi-beam transmitter of any of examples 15 to 20, wherein the antenna array is implemented by an antenna-in-package (AiP) solution having a machined base plate. [0231] Example 22: A multi-beam receiver comprising: - an antenna array; - a plurality of beam reception paths (Path 1, Path 2, Path 3, Path 4) each configured to process a corresponding beam; - a multi-beam lattice-based selection matrix configured to simultaneously couple a plurality of antennas of the antenna array to two or more of the beam supply paths. [0232] Example 23: The above multi-beam receiver of example 22, comprising a front-end module comprising GaN transistors configured to amplify received beams. [0233] Example 24: A multi-beam transmitter comprising: - a metasurface; - a plurality of beam supply paths each configured to supply a corresponding beam; and - - a selection matrix configured to simultaneously couple two or more of the beam supply paths to a plurality of transmission points on the metasurface, wherein the selection matrix comprises at least one unit cell coupling first and second beam supply paths among the plurality of beam supply paths to first and second transmission points of the metasurface, each unit cell comprising: a first switch coupling a first input node of the unit cell to a first output node of the unit cell; a second switch coupling a second input node of the unit cell to a second output node of the unit cell; a third switch coupling the first input node to the second output node; and a fourth switch coupling the second input node to the first output node of the unit cell. [0234] Example 25: A multi-beam receiver comprising: - A metasurface; - a plurality of beam reception paths each configured to process a corresponding beam; and - a selection matrix configured to simultaneously couple a plurality of reception points of the metasurface to two or more of the beam supply paths, wherein the selection matrix comprises at least one unit cell coupling first and second of the reception points of the metasurface to first and second beam supply paths among the plurality of beam supply paths, each unit cell comprising: - a first switch coupling a first input node of the unit cell to a first output node of the unit cell; - a second switch coupling a second input node of the unit cell to a second output node of the unit cell; - a third switch coupling the first input node to the second output node; and - a fourth switch coupling the second input node to the first output node of the unit cell. - Technical References [0235] The contents of each of the following references is hereby incorporated by reference in its entirety: [01] Urkowitz, H. Energy detection of unknown deterministic signals. Proc. IEEE 1967, 55, 523–531. [02] Wane, S.; Patton, R.; Gross, N. Unification of instrumentation and EDA tooling platforms for enabling smart chip-package-PCB-probe arrays co-design solutions using advanced RFIC technologies. In Proceedings of the 2018 IEEE Conference on Antenna Measurements & Applications (CAMA), Västerås, Sweden, 3–6 September 2018, pp. 1–4. [03] Rowe, H. Amplitude modulation with a noise carrier. Proc. IEEE 1964, 52, 389–395. [04] Rowe, H. Frequency or phase modulation with a noise carrier. Proc. IEEE 1964, 52, 396–408. [05] Kubo, R. The fluctuation-dissipation theorem. Rep. Prog. Phys. 1966, 29, 255–284. [06] Nyquist, H. Thermal Agitation of Electric Charge in Conductors. Phys. Rev. 1928, 32, 110. [07] Callen, H.B.; Welton, T.A. Irreversibility and Generalized Noise. Phys. Rev. 1951, 83, 34–40. [08] Onsager, L. Reciprocal Relations in Irreversible Processes. I. Phys. Rev. 1931, 37, 405–426. [09] Lindgren, K. Microscopic and macroscopic entropy. Phys. Rev. E 1988, 38, 4794–4798. [10] Lesne, A. Shannon entropy: A rigorous notion at the crossroads between probability, information theory, dynamical systems and statistical physics. Math. Struct. Comput. Sci. 2014. [11] J. Sombrin, “On the formal identity of EVM and NPR measurement methods: Conditions for identity of error vector magnitude and noise power ratio,” in EuMC Proc., Manchester, UK, 2011, pp. 337-340. [12] Standard entitled “IEEE Recommended Practice for Estimating the Uncertainty in Error Vector Magnitude of Measured Digitally Modulated Signals for Wireless Communications”. [13] 3GPP Technical Specification (TS) 38.521-4 Version 16.6.0 NR; User Equipment (UE) Conformance Specification; Radio Transmission and Reception 3rd Generation Partnership Project (3GPP), 3rd Generation Partnership Project (3GPP), 2020. [14] 3GPP Technical Specification (TS) 38.509 Version 15.9.0 5GS; Special Conformance Testing Functions for User Equipment (UE) 3rd Generation Partnership Project (3GPP), 3rd Generation Partnership Project (3GPP), 2020. [15] Lloyd, S. Enhanced sensitivity of photodetection via quantum illumination. Science 2008, 321, 1463–1465. [16] Tan, S.-H.; Erkmen, B.I.; Giovannetti, V.; Guha, S.; Lloyd, S.; Maccone, L.; Pirandola, S.; Shapiro, J.H. Quantum Illumination with Gaussian States. Phys. Rev. Lett. 2008, 101, 253601. [17] Zhang, Z.; Tengner, M.; Zhong, T.; Wong, F.N.C.; Shapiro, J.H. Entanglement’s Benefit Survives an Entanglement- Breaking Channel. Phys. Rev. Lett. 2013, 111, 010501. [18] ELopaeva, E.D.; Ruo-Berchera, I.; Degiovanni, I.P.; Olivares, S.; Brida, G.; Genovese, M. Experimental Realization of Quantum Illumination. Phys. Rev. Lett. 2013, 110, 153603. [19] Zhang, Z.; Mouradian, S.; Wong, F.N.C.; Shapiro, J.H. Entanglement-Enhanced Sensing in a Lossy and Noisy Environment. Phys. Rev. Lett. 2015, 114, 110506. [20] Barzanjeh, S.; Guha, S.; Weedbrook, C.; Vitali, D.; Shapiro, J.H.; Pirandola, S. Microwave Quantum Illumination. Phys. Rev. Lett. 2015, 114, 080503. [21] Gu, X.; Kockum, A.F.; Miranowicz, A.; Liu, Y.-X.; Nori, F. Microwave photonics with superconducting quantum circuits. Phys. Rep. 2017, 718, 1–102. [22] Sanz, M.; las Heras, U.; Garcia-Ripoll, J.J.; Solano, E.; Di Candia, R. Quantum estimation methods for quantum illumination. Phys. Rev. Lett. 2017, 118, 070803. [23] Zhuang, Q.; Zhang, Z.; Shapiro, J.H. Optimum Mixed-State Discrimination for Noisy Entanglement-Enhanced Sensing. Phys. Rev. Lett. 2017, 118, 040801. [24] M. Bahri, E. S. M. Hitzer, A. Hayashi, and R. Ashino, An uncertainty principle for quaternion Fourier transform, Comps. Maths with Appl., 56(9) 2398–2410(2008). [25] M. Bahri, Correlation theorem for Wigner-Ville distribution, Far East J. Math. Sci. 80(1) 123– 133(2013). [26] R. F. Bai, B. Z. Li, and Q. Y. Cheng, Wigner-Ville distribution associated with the linear canonical transform, J. Appl. Maths, 2012, Article ID 740161. [27] L. Debnath, B. V. Shankara, and N. Rao, On new two- dimensional Wigner-Ville nonlinear integral transforms and their basic properties, Int. Trans. Sp. Funct., 21(3) 165–174(2010). [28] W. B. Gao and B. Z. Li, Convolution and correlation theorems for the windowed offset linear canonical transform arxiv: 1905.01835v2 [math.GM] (2019) 7. X. Guanlei, W. Xiaotong, X. Xiaogang, Uncertainty inequalities for linear canonical transform. IET Signal Process. 3(5) 392–402 (2009) [29] Y. El Haoui S. and S. Fahlaoui, Generalized Uncertainty Principles associated with the Quaternionic Offset Linear Canonical Transform, https://arxiv.org/abs/1807.04068v1. [30] E. M. S. Hitzer, Quaternion Fourier transform on quaternion fields and generalizations, Adv. Appl. Clifford Algs , 17(3) 497–517(2007). [31] M. Bahri, R. Ashino, and R. Vaillancourt, Convolution theorems for quaternion Fourier transform: properties and applications, Abst. Appl Anal. 2013, Article ID 162769. 2. [32] Y.E. Song, X.Y. Zhang, C.H. Shang, H.X. Bu, X.Y. Wang, The Wigner-Ville distribution based on the linear canonical transform and its applications for QFM signal parameters estimation, J. App. Maths (2014), 8 pages. [33] D. Urynbassarova, B. Z. Li, and R. Tao, The Wigner-Ville distribution in the linear canonical transform domain, IAENG Int. J. Appl. Maths. 46 (4) 559-563(2016). [34] D. Urynbassarova, B.Zhao, R.Tao, Convolution and Correlation Theorems for Wigner-Ville Distribution Associated with the Offset Linear Canonical Transform. Int. J. Light Elect. Optics http://dx.doi.org/10.1016/j.ijleo.2017.08.099. [35] D. Wei, Q. Ran, and Y. Li, A convolution and correlation theorem for the linear canonical transform and its application, Circuits Syst. Signal Process., 31(1) 301– 312(2012). [36] K. I. Kou, J. Morais and Y. Zhang, Generalized prolate spheroidal wave functions for offset linear canonical transform in clifford analysis, Mathematical Methods in the Applied Sciences 36 (9) 1028-1041(2013). [37] Y. G. Li, B. Z Li and H. F. Sun, Uncertainty principle for Wigner-Ville distribution associated with the linear canonical transform, Abstr. Appl. Anal., 2014, Article ID 470459. [38] H. Y. Huo, W. C. Sun, L. Xiao, Uncertainty principles associated with the offset linear canonical transform Mathl. Methods Appl. Scis. 42(2) 466-474(2019). [39] Wane, S. Power Integrity, Signal Integrity, EMI & EMC in Integrated Circuits and Systems: Towards Multi-Physics Energy-Oriented Approaches. Habilitation à Diriger des Recherches: 2013.