Radio communications system and method with increased transmission capacity

Abstract

A radio communications method includes carrying out, by a transmitter: providing a digital time signal carrying digital symbols to be transmitted; and transmitting a radio frequency signal carrying the digital time signal. The method further includes carrying out, by a receiver: receiving the radio frequency signal transmitted by the transmitter; processing the received radio frequency signal to obtain a corresponding incoming digital signal; and extracting, from the incoming digital signal, the digital symbols carried by the incoming digital signal. The digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein the frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

Claims

1. A radio communications method comprising carrying out, by a transmitter, the following steps: a) providing a digital time signal carrying digital symbols to be transmitted; and b) transmitting a radio frequency signal carrying said digital time signal; the method further comprising carrying out, by a receiver, the following steps: c) receiving the radio frequency signal transmitted by the transmitter; d) processing the received radio frequency signal so as to obtain a corresponding incoming digital signal; and e) extracting, from the incoming digital signal, the digital symbols carried by said incoming digital signal; wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

2. The method of claim 1, wherein the digital time signal is time-limited, carries a limited sequence of digital symbols to be transmitted, and results from: main mode frequency samples carrying respective digital symbols of said limited sequence via a frequency main mode; and twisted mode frequency samples carrying the other digital symbols of said limited sequence via one or more frequency twisted modes, wherein each frequency twisted mode is a complex harmonic mode that is orthogonal to the frequency main mode and to any other frequency twisted mode used.

3. The method of claim 2, wherein the main mode frequency samples are at main mode frequencies spaced apart by a predetermined frequency spacing; and wherein the twisted mode frequency samples comprise, for a frequency twisted mode, respective twisted mode frequency samples at corresponding twisted mode frequencies that: are related to said frequency twisted mode; are spaced apart by said predetermined frequency spacing; and are different from the main mode frequencies.

4. The method of claim 3, wherein the one or more frequency twisted modes comprise 2N frequency twisted modes each identified by a respective integer index n that is comprised between N and +N and is different from zero, N denoting an integer higher than zero; wherein the limited sequence of digital symbols to be transmitted comprises S.sub.TOT digital symbols, S.sub.TOT being equal to 2.sup.N+21; wherein the frequency main mode carries M.sub.MFS of said S.sub.TOT digital symbols by means of M.sub.MFS main mode frequency samples at corresponding main mode frequencies, that are spaced apart by said predetermined frequency spacing and that range from B.sub.S to M.sub.MFS times B.sub.S, B.sub.S denoting said predetermined frequency spacing and M.sub.MFS being equal to 2.sup.N+1+1; wherein said 2N frequency twisted modes carry the S.sub.TOTM.sub.MFS digital symbols not carried by the frequency main mode; and wherein each frequency twisted mode n carries 2.sup.N|n| respective digital symbol(s) by means of 2.sup.N+1 respective twisted mode frequency samples at corresponding twisted mode frequencies, that are spaced apart by said predetermined frequency spacing and that are located, in frequency domain, at $B_S\left(\frac{2^{.Math.n.Math.}-1}{2^{.Math.n.Math.}}+k\right),$ where k denotes an integer ranging from zero to 2.sup.N+11, or from one to 2.sup.N+1.

5. The method of claim 4, wherein each of said S.sub.TOT digital symbols to be transmitted is represented by a respective symbol complex value; and wherein, for each frequency twisted mode n, the 2.sup.N+1 respective twisted mode frequency samples comprise, for each of the 2.sup.N|n| respective digital symbol(s), 2.sup.|n|+1 frequency samples, that: carry said digital symbol; are at frequencies that are located, in frequency domain, at $B_S\left[\frac{2^{.Math.n.Math.}-1}{2^{.Math.n.Math.}}+\left(k^{*}+i.Math.2^{.Math.n.Math.+1}\right)\right],$ where k* denotes an integer ranging from zero to 2.sup.|n|+11, or from one to 2.sup.|n|+1, and where i is an index that identifies said digital symbol and is comprised between zero and 2.sup.N|n|1; and have, each, a respective complex value obtained by multiplying the symbol complex value representing said digital symbol by a respective complex factor related to said frequency twisted mode n and to the frequency of said frequency sample.

6. The method of claim 5, wherein, for each frequency twisted mode n and for each of the 2.sup.N|n| respective digital symbol(s), the 2.sup.|n|+1 respective frequency samples carrying said digital symbol have, each, a respective complex value obtained by multiplying the symbol complex value representing said digital symbol by a respective complex factor which: if n is higher than zero, is equal to $\frac{e^{+jk.Math.\frac{}{2^{.Math.n.Math.}}}}{2^{\frac{.Math.n.Math.+1}{2}}}$ or, if n is lower than zero, is equal to $\frac{e^{-jk.Math.\frac{}{2^{.Math.n.Math.}}}}{2^{\frac{.Math.n.Math.+1}{2}}}$ where j denotes the imaginary unit.

7. The method according to claim 4, wherein said step a) includes providing the digital time signal by using a predefined transmission matrix that relates the S.sub.TOT digital symbols to be transmitted to time samples of the digital time signal through coefficients related to a transform from frequency domain to time domain of the main mode frequency samples and the twisted mode frequency samples; and wherein said step e) includes extracting the digital symbols carried by the incoming digital signal by using a reception matrix derived from the predefined transmission matrix.

8. The method of claim 7, wherein the predefined transmission matrix is such that the matrix resulting from the multiplication of the transpose of said predefined transmission matrix and said predefined transmission matrix has a determinant different from zero; and wherein the reception matrix is derived from the predefined transmission matrix through a pseudo-inverse technique.

9. The method of claim 8, wherein the reception matrix is computed on the basis of the following formula:
[[GFFT]]=([[GIFFT]].sup.T[[GIFFT]]).sup.1[[GIFFT]].sup.T, where [[GFFT]] denotes the reception matrix, [[GIFFT]] denotes the predefined transmission matrix, [[GIFFT]].sup.T denotes the transpose of the predefined transmission matrix, and ([[GIFFT]].sup.T [[GIFFT]]).sup.1 denotes the operation of inversion of the matrix resulting from the multiplication of the transpose of the predefined transmission matrix and the predefined transmission matrix.

10. The method according to claim 7, wherein the digital time signal comprises a number of time samples equal to $\left[1+\frac{N\left(N+1\right)}{2}\right]2^{N+1}+1;$ and wherein the predefined transmission matrix comprises M.sub.TSXS.sub.TOT coefficients, M.sub.TS denoting said number of time samples of the digital time signal.

11. The radio communications method according to claim 2, wherein the main mode frequency samples are frequency samples of Orthogonal Frequency-Division Multiplexing (OFDM) type, or of Orthogonal Frequency-Division Multiple Access (OFDMA) type.

12. The radio communications method according to claim 2, wherein said step a) includes: providing a first digital time signal resulting from the main mode frequency samples and the twisted mode frequency samples; and providing a second digital time signal which includes a cyclic prefix followed by the first digital time signal, wherein the cyclic prefix is a replica of an end portion of said first digital time signal; and wherein said step b) includes transmitting a radio frequency signal carrying the second digital time signal.

13. A radio communications system comprising a transmitter and a receiver; wherein the transmitter is configured to provide a digital time signal carrying digital symbols to be transmitted and to transmit a radio frequency signal carrying said digital time signal, wherein the receiver is configured to receive the radio frequency signal transmitted by the transmitter, to process the received radio frequency signal so as to obtain a corresponding incoming digital signal and to extract, from the incoming digital signal, the digital symbols carried by said incoming digital signal, and wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

14. A system for radio communications comprising a transmitter configured to provide a digital time signal carrying digital symbols to be transmitted and to transmit a radio frequency signal carrying said digital time signal, said radio frequency signal transmitted by the transmitter to be received by a receiver that is configured to process the received radio frequency signal so as to obtain a corresponding incoming digital signal and to extract, from the incoming digital signal, the digital symbols carried by said incoming digital signal, wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

15. A non-transitory hardware component comprising software code portions which are: executable by a processor of a device or system for radio communications; and such that to cause, when executed, said device or system to become configured to provide a digital time signal carrying digital symbols to be transmitted, and to transmit a radio frequency signal carrying said digital time signal, said transmitted radio frequency signal to be received by a receiver that is configured to process the received radio frequency signal so as to obtain a corresponding incoming digital signal and to extract, from the incoming digital signal, the digital symbols carried by said incoming digital signal, wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

16. A system for radio communications comprising a receiver configured to: radio communicate with a transmitter configured to provide a digital time signal carrying digital symbols to be transmitted, and to transmit a radio frequency signal carrying said digital time signal; and receive the radio frequency signal transmitted by the transmitter; process the received radio frequency signal so as to obtain a corresponding incoming digital signal; and extract, from the incoming digital signal, the digital symbols carried by said incoming digital signal; wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

17. A non-transitory hardware component comprising software code portions which are: executable by a processor of a first device or system designed to radio communicate with a second device or system configured to provide a digital time signal carrying digital symbols to be transmitted, and to transmit a radio frequency signal carrying said digital time signal; and such that to cause, when executed, said second device or system to become configured to receive the radio frequency signal transmitted by the first device or system; to process the received radio frequency signal so as to obtain a corresponding incoming digital signal; and to extract, from the incoming digital signal, the digital symbols carried by said incoming digital signal; wherein said digital time signal carrying the digital symbols to be transmitted results from an approximation of the Hilbert transform in frequency domain, which approximation is based on a frequency main mode and one or more frequency twisted modes, wherein said frequency main and twisted modes carry, each, respective digital symbols to be transmitted.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) For a better understanding of the present invention, preferred embodiments, which are intended purely by way of non-limiting example, will now be described with reference to the attached drawings (all not to scale), wherein:

(2) FIG. 1 schematically illustrates a transmitting system for transmitting radio vortices according to the background art;

(3) FIG. 2 schematically illustrates a device for generating OAM modes, which device is exploited by the transmitting system shown in FIG. 1;

(4) FIG. 3 schematically illustrates a receiving system for receiving radio vortices according to the background art;

(5) FIG. 4 schematically illustrates a device for demodulating OAM modes, which device is exploited by the receiving system shown in FIG. 2;

(6) FIG. 5 schematically illustrates time behavior of a complex frequency function constituted by a real cosinusoidal function and an imaginary sinusoidal function;

(7) FIGS. 6 and 7 schematically illustrate complex frequency samples of a signal limited in time and of its corresponding analytical signal, respectively (assuming that the signal can be approximated as a band-limited signal too);

(8) FIGS. 8 and 9 schematically illustrate time inverse Fourier transform of a complex frequency function and time inverse Fourier transform of its frequency Hilbert transform, respectively;

(9) FIG. 10 schematically illustrates a comparison between a process for generating twisted waves in time domain according to International application PCT/FR2013/052636, and a process for generating twisted waves in frequency domain according to the present invention;

(10) FIGS. 11 and 12 schematically illustrate a complex frequency function of an analytical signal and its Hilbert transform in frequency domain, respectively;

(11) FIG. 13 schematically illustrates frequency Hilbert transform and approximations of the latter obtained by using, respectively, one, two, and five orthogonal exponential modes;

(12) FIGS. 14, 15 and 16 schematically illustrate frequency behavior of three different orthogonal exponential modes;

(13) FIG. 17 schematically illustrates powers associated with orthogonal harmonic modes used to develop frequency Hilbert transform;

(14) FIGS. 18 and 19 schematically illustrate frequency twist complex functions related to two orthogonal harmonic modes;

(15) FIG. 20 schematically illustrates an OFDM super frame with additional Twisted frame Frequency Units (TFUs);

(16) FIG. 21 schematically illustrates an approximation of the impulse response of the frequency Hilbert transform using three complex harmonic modes;

(17) FIG. 22 schematically represents a radio communications method according to the present invention as a generalization of the traditional OFDM (or OFDMA) technique;

(18) FIGS. 23 and 24 schematically illustrate time behavior of a time twisted mode +1 complex signal and of a frequency twisted mode +1 complex signal, respectively;

(19) FIGS. 25 and 26 schematically illustrate symbol and clock time relationship according to an aspect of the present invention;

(20) FIG. 27 schematically illustrates a traditional scheme of cyclic prefix for OFDM-OFDMA;

(21) FIG. 28 schematically illustrates an example of two delay spread components with cyclic prefix;

(22) FIG. 29 schematically illustrates time frame duration increase due to cyclic prefix;

(23) FIGS. 30 and 31 schematically illustrate time behavior of main mode and twisted modes, respectively, when a cyclic prefix is used;

(24) FIG. 32 schematically illustrates noise bandwidths for time twisted waves;

(25) FIG. 33 schematically illustrates noise impact in the case of frequency twisted waves;

(26) FIG. 34 schematically illustrates spectral efficiency as a function of the energy per symbol over noise density for frequency twisted waves;

(27) FIG. 35 schematically illustrates a frequency structure of frequency twisted waves according to an illustrative embodiment of the present invention;

(28) FIG. 36 schematically illustrates a transmitting system according to an illustrative embodiment of the present invention;

(29) FIG. 37 schematically illustrates a receiving system according to an illustrative embodiment of the present invention;

(30) FIG. 38 schematically illustrates an example of square matrix resulting from the multiplication of the transpose of a transmission matrix according to an aspect of the present invention by said transmission matrix;

(31) FIG. 39 schematically illustrates a multilayer architecture wherein a frame structure of frequency twisted waves according to an aspect of the present invention is embedded in a traditional OFDM architecture;

(32) FIG. 40 schematically illustrates computational complexity of the present invention and frequency reuse according to the present invention as a function of the number of frequency twisted modes used; and

(33) FIG. 41 schematically illustrates flexibility in using OFDM modularity, complex equation number and implementation criticality of the present invention as a function of the number of frequency twisted modes used.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS OF THE INVENTION

(34) The following discussion is presented to enable a person skilled in the art to make and use the invention. Various modifications to the embodiments will be readily apparent to those skilled in the art, without departing from the scope of the present invention as claimed. Thus, the present invention is not intended to be limited to the embodiments shown and described, but is to be accorded the widest scope consistent with the principles and features disclosed herein and defined in the appended claims.

(35) The present invention relates to a practical, efficient mode, in general, for increasing transmission capacity and, in particular, for increasing RF spectrum reuse. In this connection, International Application No. PCT/FR2013/052636 (whose content is herewith enclosed by reference) filed on 5 Nov. 2013 in the name of EUTELSAT S. A. discloses the feasibility of increasing transmission capacity at Radio Frequency (RF) (including frequencies from a few KHz to hundreds of GHz) by exploiting a proper approximation in time domain of the Hilbert transform of digital analytical signals, wherein said approximation of the Hilbert transform is implemented by using twisted waves, specifically orthogonal harmonic modes.

(36) In particular, PCT/FR2013/052636 describes a radio communications system, which comprises a transmitter and a receiver, wherein the transmitter is configured to: generate or receive digital symbols having a given symbol rate associated with a corresponding symbol period; generate, every S digital symbols generated/received, a respective multi-mode digital signal, which has a predefined time length shorter than S times the symbol period, which is sampled with a predefined sampling rate higher than the symbol rate, and which carries said S digital symbols by means of a plurality of orthogonal harmonic modes comprising a main mode which is a real harmonic mode and carries P of said S digital symbols, and one or more secondary modes carrying the other S-P digital symbols, each secondary mode being a complex harmonic mode time-shifted by half the symbol period with respect to the main mode (wherein S is an integer higher than three and P is an integer lower than S); and transmit a radio frequency signal carrying a sequence of the generated multi-mode digital signals.

(37) Moreover, the receiver of the radio communications system according to PCT/FR2013/052636 is configured to: receive the radio frequency signal transmitted by the transmitter; process the received radio frequency signal so as to obtain a corresponding incoming digital signal; and extract, from successive, non-overlapped portions of the incoming digital signal sampled with the predefined sampling rate, the S digital symbols respectively carried by each incoming digital signal portion by means of the orthogonal harmonic modes; wherein each of the successive, non-overlapped portions of the incoming digital signal has the predefined time length.

(38) Preferably, the transmitter according to PCT/FR2013/052636 is configured to generate a multi-mode digital signal carrying S digital symbols by: allocating P of the S digital symbols to the main mode by providing, for each of said P digital symbols, a corresponding complex value which represents said digital symbol and is related to the main mode; allocating each of the other S-P digital symbols to a corresponding secondary mode by providing, for each of said S-P digital symbols, a corresponding complex value which represents said digital symbol and is related to the secondary mode to which said digital symbol is allocated; computing, by using a predefined transmission matrix, M multi-mode complex values related to M successive time instants which, within the predefined time length, are separated by half the symbol period, wherein M is an integer equal to or higher than S, and wherein the predefined transmission matrix relates the S complex values representing the S digital symbols and related to the harmonic modes to the M successive time instants through complex coefficients each of which is related to a respective harmonic mode and to a respective time instant; and generating a multi-mode digital signal having the predefined time length and sampled with the predefined sampling rate on the basis of the M multi-mode complex values computed.

(39) Again preferably, the receiver according to PCT/FR2013/052636 is configured to extract the S digital symbols carried by an incoming digital signal portion having the predefined time length and sampled with the predefined sampling rate by: extracting, from said incoming digital signal portion, M multi-mode complex values related to M successive time instants which are, within the predefined time length, separated by half the symbol period; computing, by using a reception matrix derived from the predefined transmission matrix, S complex values representing the S digital symbol carried by said incoming digital signal portion by means of the orthogonal harmonic modes, wherein said reception matrix relates the M extracted multi-mode complex values related to the M successive time instants to the S complex values to be computed through complex coefficients each of which is related to a respective harmonic mode and to a respective time instant; and determining the S digital symbols represented by the S complex values computed.

(40) Conveniently, the reception matrix used by the receiver according to PCT/FR2013/052636 is derived from the predefined transmission matrix through a generalized inversion technique.

(41) More conveniently, according to PCT/FR2013/052636, the predefined transmission matrix is such that the matrix resulting from the multiplication of the transpose of said predefined transmission matrix and said predefined transmission matrix has a determinant different from zero, and the reception matrix is derived from the predefined transmission matrix through a pseudo-inverse technique.

(42) More and more conveniently, according to PCT/FR2013/052636, the reception matrix is computed on the basis of the following formula:
[[GMF]]=([[A]].sup.T[[A]]).sup.1[[A]].sup.T,
where [[GMF]] denotes the reception matrix, [[A]] denotes the predefined transmission matrix, [[A]].sup.T denotes the transpose of the predefined transmission matrix, and ([[A]].sup.T [[A]]).sup.1 denotes the operation of inversion of the matrix resulting from the multiplication of the transpose of the predefined transmission matrix and the predefined transmission matrix.

(43) Preferably, according to PCT/FR2013/052636, the main mode comprises, within the predefined time length, P samples with sampling period equal to the symbol period, the secondary modes comprise, within the predefined time length, P1 samples with sampling period equal to the symbol period, each secondary mode is time-shifted by half the symbol period with respect to the main mode, and said M successive time instants, which, within the predefined time length, are separated by half the symbol period, are the sampling times of the main mode and of the secondary modes, thereby resulting that M=2P1.

(44) More preferably, according to PCT/FR2013/052636, the harmonic modes comprise 2N secondary complex harmonic modes each of which carries a respective Orbital Angular Momentum (OAM) mode and has a respective topological-charge-related index n comprised between N and +N, wherein N is an integer higher than one; moreover, the main mode carries P=2.sup.N+1+1 digital symbols and each secondary complex harmonic mode having topological-charge-related index n carries 2.sup.Nn+1 digital symbols, thereby resulting that M=2.sup.N+2+1 and S=2.sup.N+21.

(45) Conveniently, according to PCT/FR2013/052636, the predefined sampling rate depends at least on the predefined time length of each multi-mode digital signal and of each of the successive, non-overlapped portions of the incoming digital signal.

(46) More conveniently, according to PCT/FR2013/052636, the predefined time length is equal to P times the symbol period.

(47) More and more conveniently, according to PCT/FR2013/052636, the predefined sampling rate is determined on the basis of the following formula:

(48) $\mathrm{CR}=\frac{2P+u}{2P}.Math.\frac{1}{T_S},$
where CR denotes said predefined sampling rate, T.sub.S denotes the symbol period, and u denotes a digital-vestigial-component-related parameter whose value is an integer and depends at least on the predefined time length.

(49) Preferably, the transmitter according to PCT/FR2013/052636 is configured to generate a multi-frame digital signal comprising successive, non-overlapped time frames each of which has the predefined time length and is occupied by a respective multi-mode digital signal; moreover, the multi-frame digital signal carries frame synchronization data related to frame synchronization of its time frames; accordingly, the radio frequency signal transmitted by the transmitter carries the multi-frame digital signal.

(50) Additionally, the receiver according to PCT/FR2013/052636 is further configured to: extract the frame synchronization data from the incoming digital signal; detect, on the basis of the extracted frame synchronization data, successive, non-overlapped time frames of the incoming digital signal with the predefined time length; and, for each detected time frame of the incoming digital signal, extract, from the incoming digital signal portion within said time frame, the S digital symbols carried by said incoming digital signal portion by means of the orthogonal harmonic modes.

(51) More preferably, according to PCT/FR2013/052636, the multi-frame digital signal comprises a preamble followed by F successive, non-overlapped time frames occupied, each, by a respective multi-mode digital signal, F being an integer higher than one; in particular, the preamble carries frame synchronization data related to frame synchronization of the F following time frames.

(52) More and more preferably, according to PCT/FR2013/052636, the frame synchronization data indicate time frame beginning and/or the predefined time length of the time frames.

(53) In order to increase, in general, transmission capacity at Radio Frequency (RF) (including frequencies from a few KHz to hundreds of GHz) and, in particular, RF spectrum reuse, the present invention, by exploiting duality between time and frequency, teaches to use a twisted-wave-based approximation of the Hilbert transform in frequency domain.

(54) In particular, thanks to duality principle between time and frequency it is possible to exploit twisted wave functions also in frequency domain. The results are very interesting and promising with features which are, on the one hand, similar to time domain case, but, on the other hand, rather different, for practical applications, from time domain case.

(55) In detail, frequency twist can be seen as a generalisation of the well-known OFDM approach, introducing an absolute novelty in the analysis and design of OFDM signals.

(56) Theory underlying the present invention will be presented in the following.

(57) As is known, a signal can be represented in time or frequency domain, time and frequency being conjugate variables.

(58) Considering a time-limited signal within a time window T (as usual technique in the case of OFDM-OFDMA signals), in frequency domain said signal can be represented by a series of sinc functions:

(59) $X\left(f\right)=\underset{k=-}{\overset{+}{.Math.}}{a}_k{e}^{j{}_k}\frac{\sin \left[T\left(f-\frac{k}{T}\right)\right]}{T\left(f-\frac{k}{T}\right)}.$

(60) In the case the signal X(f) can be approximated with a band-limited signal with bandwidth B, this implies that:
X(ff.sub.0)=X.sup.+(ff.sub.0)+X.sup.(ff.sub.0),
where + and denote positive and negative frequencies, respectively.

(61) Taking into consideration only the positive frequencies, it is possible to write:
X.sup.+(ff.sub.0)=X(ff.sub.0) for f0, and
X.sup.+(ff.sub.0)=0 for f<0,
and also

(62) $X^{+}\left(f\right)=\underset{k=-N}{\overset{+N}{.Math.}}{a}_k{e}^{j{}_k}\frac{\sin \left[T\left(f-f_0-\frac{k}{T}\right)\right]}{t\left(f-f_0-\frac{k}{T}\right)},$
where N=TB.

(63) Thence, each sample is constituted by a real part given by a.sub.k cos .sub.k, and an imaginary part given by a.sub.k sin .sub.k. The time representation of such a sample is given by one cosinusoidal function in the time window T, having an amplitude of a.sub.k cos .sub.k, and one sinusoidal function having an amplitude of a.sub.k sin .sub.k, as shown in FIG. 5 where time behavior of said complex frequency sample is illustrated.

(64) The frequency pattern is given by two couples of sinc functions, namely one for the real part and one for the imaginary part, as shown in FIGS. 6 and 7 which illustrate complex frequency samples of a signal limited in time and of its corresponding analytical signal, respectively (assuming that the signal can be approximated as a band-limited signal too).

(65) As far as analytical signals are concerned, the traditional Hilbert transform is applied in time, assuming that the total bandwidth of the signal can be considered limited and that the baseband signal has been shifted to a proper frequency such that to allow the full bandwidth to be on the positive frequency semiaxis (and, of course, replicated on the negative one). On the positive frequency semiaxis, with respect to the central frequency sample for k=0, frequency samples are complex and there results that a.sub.ke.sup.j.sup.ka.sub.ke.sup.j.sup.k.

(66) Taking into consideration a complex frequency sample related to a limited time window, it is possible to apply a second Hilbert transform to the function X.sup.+(ff.sub.0) in frequency domain (on the assumption that the signal is a limited time duration signal):

(67) 0$X_H^{+}\left(f-f_0\right)={}_{-}^{+}\frac{X\left(\right)}{\left(f-f_0-\right)}d,$
where the integral can be understood as the main Cauchy value.

(68) Thence, the time function results to be:
x.sub.H(t)=x(t)e.sup.j2f.sup.0.sup.t(u.sub.0(t)u.sub.0(t)),
where u.sub.0(t) and u.sub.0(t) are the step functions for t>0 and t<0, respectively.

(69) On the assumption that

(70) $X\left(f\right){|}_{f_k=\frac{k}{T}}=\frac{\sin \left[\left(\mathrm{fT}-k\right)\right]}{\left(\mathrm{fT}-k\right)},$
then the time function is given by:

(71) $\begin{array}{cc}x\left(t\right)=e^{j2k\frac{t}{T}},& \mathrm{for}.Math.t.Math.<\frac{T}{2},\mathrm{and}\\ x\left(t\right)=0,& \mathrm{for}.Math.t.Math.\frac{T}{2}.\end{array}$

(72) Thence, the time transform of the frequency Hilbert transform results to be:

(73) $\begin{array}{cc}x\left(t\right)\left(u_0\left(t\right)-u_0\left(-t\right)\right)=e^{j2k\frac{t}{T}}& \mathrm{for}0<t<\frac{T}{2},\\ x\left(t\right)\left(u_0\left(t\right)-u_0\left(-t\right)\right)=-e^{j2k\frac{t}{T}}& \mathrm{for}-\frac{T}{2}<t<0,\mathrm{and}\\ x\left(t\right)\left(u_0\left(t\right)-u_0\left(-t\right)\right)=0& \mathrm{for}.Math.t.Math.\frac{T}{2}.\end{array}$

(74) In this respect, FIGS. 8 and 9 show time inverse Fourier transform of a complex frequency function and time inverse Fourier transform of its frequency Hilbert transform, respectively.

(75) The analysis of properties of this signal family is thence based on a sequential application of one time Hilbert transform to get the analytical signal and one frequency Hilbert transform to get the twisted wave signals, which are orthogonal to the original samples.

(76) The process just described is similar but somewhat substantially different from time twist case. In fact, as described in PCT/FR2013/052636, in time twist the Hilbert transform is applied twice in time: the first time Hilbert transform is used to get the analytical signal, and the second time Hilbert transform is used to create the family of twisted waves orthogonal to the original samples.

(77) In this respect, FIG. 10 schematically shows a comparison of the processes for generating twisted waves in time domain according to PCT/FR2013/052636 and in frequency domain according to the present invention.

(78) In particular, as shown in FIG. 10, both the process according to PCT/FR2013/052636 (denoted as a whole by 300) and the process according to the present invention (denoted as a whole by 400) are applied to a limited-band signal x(t) and include: applying to the limited-band signal x(t) a frequency shift (block 301 and 401, respectively); and performing a time Hilbert transform of the frequency-shifted signal to get the analytical signal (block 302 and 402, respectively).

(79) Instead, the two processes are differentiated by the fact that: the process 300 according to PCT/FR2013/052636 exploits an approximation in time domain of the Hilbert transform of the analytical signal to create time twisted waves (block 303); and the process 400 according to the present invention exploits an approximation in frequency domain of the Hilbert transform of the analytical signal to create frequency twisted waves (block 403).

(80) In detail, as far as the process 400 according to the present invention is concerned, the Hilbert transform in frequency domain can be seen as an inverse Fourier transform of the analytical signal previously described. In this respect, FIGS. 11 and 12 show complex frequency function of the analytical signal and its Hilbert transform in frequency domain, respectively.

(81) The frequency Hilbert transform increases the bandwidth necessary to represent the signal, due to the presence of a discontinuity in the time function at the origin (i.e., taking into consideration the meaning of the main Cauchy value, the position of the symmetry/asymmetry axis of the integration). This aspect is similar to the situation of the Hilbert transform in time domain, and can be handled by considering a development into a series of orthogonal modes. In this respect, FIG. 13 shows frequency Hilbert transform and approximations of the latter obtained by using, respectively, one, two, and five orthogonal exponential modes.

(82) Each mode higher than mode 0 is represented by a couple of odd pulses in frequency domain, centered with respect to the frequency f.sub.0. In this respect, FIGS. 14, 15 and 16 show frequency behavior of the mode 0, 1 and 2, respectively. The modes are mutually orthogonal, but the orthogonality between each of them and the main mode frequency pulses is given by the symmetry property of the full bandwidth and this could be defined as a synoptic orthogonality.

(83) Similarly to the time domain case, the above property is similar to interferometry, which is a property depending on the space geometry and not directly on the signal.

(84) Therefore, frequency domain can be assimilated to a sort of space (specifically, a frequency space), similarly to the situation of the time twist where the time is considered a space (specifically, a time space), with additional degrees of freedom.

(85) It is important to note a basic difference between time and the frequency pulses: time pulses are real, while frequency pulses are in general complex.

(86) Therefore, the frequency twist shows a more robust capability to carry an additional information channel. In fact, while for time twist it is necessary to increase the nominal Nyquist bandwidth (approximately of 33%), the frequency twist can work without this limitation.

(87) The frequency Hilbert transform allows, theoretically, to maintain all the information content of the original signal. Therefore, also the orthogonal harmonic mode development up to infinity of the frequency Hilbert transform allows, theoretically, to maintain all the information content of the original signal. Each mode contributes to the information content proportionally to the respective power of the mode (assuming that the overall power of the signal is equal to one). In this respect, FIG. 17 schematically shows the respective power (i.e., information content) associated with each mode up to the eleventh mode. From FIG. 17 it can be noted that with the first two or three modes it is possible to maintain about 90% of the information content of the original signal with a potential frequency reuse of 1.9. The use of additional higher-order modes results in an increase in implementation complexity rather than in an effective improvement in the performances.

(88) Generation of frequency twist, in analogy with time twist generation, is organized by associating the complex symbol value a.sub.ke.sup.j.sup.k to a set of frequency pulses, properly shifted and phase-rotated.

(89) For the sake of simplicity, it is considered to operate in an OFDM signal structure, where the main signal is represented by the Inverse Fast Fourier Transform (IFFT) of the symbol time flow.

(90) In addition to this frequency symbol set, it is added, for each mode, a set of frequency samples.

(91) Modes 1 are generated repeating the same symbol at 4 different frequencies

(92) $\frac{1}{T}\left(\frac{1}{2}+k\right),$
changing each time their phases according to

(93) $e^{jk\frac{}{2}},$
with k=0, 1, 2, 3. This means that the associated IFFT is the sum of 4 decimated IFFT, having only 1 row for each sample and each one is weighted by

(94) $0.5e^{jk\frac{}{2}}.$

(95) Modes 2 are generated repeating the same symbol at 8 different frequencies

(96) $\frac{1}{T}\left(\frac{3}{4}+k\right),$
changing each time their phases according to

(97) $e^{jk\frac{}{4}},$
with k=0, 1, . . . , 7. This means that the associated IFFT is the sum of 8 decimated IFFT, having only 1 row for each sample and each one is weighted by

(98) $\frac{1}{\sqrt{8}}{e}^{jk\frac{}{4}}.$

(99) In general, modes N are generated repeating the same symbol at 2.sup.N+1 different frequencies

(100) 0$\frac{1}{T}\left(\frac{2^N-1}{2^N}+k\right),$
changing each time their phases according to

(101) $e^{jk\frac{}{2^N}},$
with k=0, 1, . . . , 2.sup.N+11. This means that the associated IFFT is the sum of 4N decimated IFFT, having only 1 row for each sample and each one is weighted as

(102) $\frac{1}{2^{\frac{N+1}{2}}}{e}^{jk\frac{}{2^N}}.$

(103) In practical terms, phases can be simplified (in terms of implementation) assuming the same value each /2, in this way the rotation can be represented by a smaller number of bits.

(104) In this respect, FIGS. 18 and 19 show frequency twist complex functions for mode 1 and 2, respectively. From FIGS. 18 and 19 it can be noted that the bandwidth increases each time that the mode grows. This implies a sort of increased rigidity of the traditional OFDM structure.

(105) Then, let us take into consideration an OFDM signal architecture, which can be considered a sequence of frequency pulses having the shape of a sinc. In the same frequency band frequency twisted waves are added and these additional elements in the following will be called Twisted Frequency frame Units (TFUs). In this respect, FIG. 20 shows an OFDM super frame with additional TFUs.

(106) The structure of a TFU is given by the superposition of the OFDM structure and of the structure of the twisted modes previously defined.

(107) The minimum length of a TFU bandwidth, where modes up to /N are used, is given by:

(108) $\frac{2^{N+1}+1}{T},$
where T is the time interval duration which is the inverse of the frequency pulse symbol bandwidth B.sub.S (i.e., T=1/B.sub.S).

(109) In this respect, FIG. 21 show an approximation of the impulse response of the frequency Hilbert transform, which approximation uses three modes.

(110) The mode structure in the TFU frame takes into account the length of each mode; therefore, using up to mode N, the number M.sub.MFS of the frequency samples of the main mode is:
M.sub.MFS=2.sup.N+1+1.

(111) It is worth recalling that, assuming B.sub.S=1/T, the frequency samples of generic frequency twisted mode N are at frequencies

(112) $\frac{1}{T}\left(\frac{2^N-1}{2^N}+k\right).$

(113) Moreover, the number of complex symbol values (or, at reception side, of complex unknowns) of the main mode n=0 is M.sub.MFS=2.sup.N+1+1, the number of complex symbol values (or, at reception side, of complex unknowns) of the modes +1 and 1 is 2.sup.N, the number of complex symbol values (or, at reception side, of complex unknowns) of the modes +2 and 2 is 2.sup.N1, the number of complex symbol values (or, at reception side, of complex unknowns) of the modes +i and i is 2.sup.Ni+1, and the number of complex symbol values (or, at reception side, of complex unknowns) of the modes +N and N is 2.sup.NN+1=2.

(114) Therefore, the overall number S.sub.TOT of complex symbol values (or of complex unknowns) is given by:

(115) $S_{\mathrm{TOT}}=\left(2^{N+1}+1\right)+\underset{i=1}{\overset{N}{.Math.}}{2}^{N-i+1},$
wherein the first addend represents the number M.sub.MFS of symbols (or, at reception side, of complex unknowns) of the main mode n=0, while the second addend (i.e., the summation) represents the number S.sub.TOT-M.sub.MFS of symbols (or, at reception side, of complex unknowns) of all the other modes with n0.

(116) The foregoing mathematical formula can be rewritten as:

(117) $S_{\mathrm{TOT}}=1+\underset{i=0}{\overset{N}{.Math.}}{2}^{N-i+1}=1+2^{N+1}.Math.\underset{i=0}{\overset{N}{.Math.}}{\left(\frac{1}{2}\right)}^i.$

(118) Thence, since it is known that

(119) $\underset{i=0}{\overset{N}{.Math.}}{x}^i=\frac{x^{N+1}-1}{x-1}\mathrm{if}x1,$
then it results that:
S.sub.TOT=2.sup.N+21.

(120) The overlapping of frequency pulses associated with different symbols creates a special form of orthogonality, which depends on the structure of the TFUs. In this sense the TFUs represent a frequency space and the different signals are orthogonal in this space according to symmetry and antisymmetry features of the signal structure. This property can be seen as equivalent to the interferometry in the traditional geometrical space.

(121) Anyway, even if the present invention deals with frequency space, the procedure for determining the transmitted signals is performed in the time domain and not in the frequency domain.

(122) In particular, as shown in FIG. 22 which schematically represents the radio communications method according to the present invention (denoted as a whole by 600) as a generalization of the traditional OFDM (or OFDMA) technique (denoted as a whole by 500), the radio communications method 600 according to the present invention can be considered similar to the traditional OFDM (or OFDMA) technique 500, which, as is broadly known, comprises: at the transmission side, the conversion of a symbol serial time stream into a parallel independent frequency stream via an Inverse Fast Fourier Transform (IFFT) (block 501); and, at the reception side, the back-transformation into the original symbol serial time stream via a Fast Fourier Transform (FFT) (block 502).

(123) Similarly, the radio communications method 600 according to the present invention exploits: at the transmission side, a Generalized Inverse Fast Fourier Transform (GIFFT) (block 601) which includes the implementation of the previously described frequency Hilbert transform approximation based on frequency twisted modes; and, at the reception side, a Generalized Fast Fourier Transform (GFFT) (block 602) which includes the extraction of the symbols carried by the frequency twisted modes.

(124) Let us now consider the structure of the twisted signals in time domain and in frequency domain (on the assumption that for both the domains the first mode 1 is used): a time twisted mode 1 signal can be expressed as
a)

(125) $x_{T1}\left(t\right)=a_k{e}^{j{}_k}\left\{\frac{1}{2}\left[\mathrm{rect}\left(t-\frac{T}{2}\right)-\mathrm{rect}\left(t-\frac{3T}{2}\right)\right]\frac{j}{2}\left[\mathrm{rect}\left(t-T\right)-\mathrm{rect}\left(t-\frac{5T}{2}\right)\right]\right\};$
and a frequency twisted mode 1 signal can be expressed as

(126) $X_{F1}\left(f\right)=a_k{e}^{j{}_k}\left\{\frac{1}{2}\left[\mathrm{rect}\left(f-\frac{B}{2}\right)-\mathrm{rect}\left(f-\frac{3B}{2}\right)\right]\frac{j}{2}\left[\mathrm{rect}\left(f-B\right)-\mathrm{rect}\left(f-\frac{5B}{2}\right)\right]\right\}.$

(127) The frequency twisted mode 1 signal is analyzed in time domain thereby resulting that:

(128) b)

(129) 0$x_{F1}\left(t\right)=\frac{a_k{e}^{j{}_k}}{2}\left\{\left[e^{-j2\left(f-\frac{B}{2}\right)t}-e^{-j2\left(f-\frac{3B}{2}\right)t}\right]j\left[e^{-j2\left(f-B\right)t}-e^{-j2\left(f-\frac{5B}{2}\right)t}\right]\right\}$$\mathrm{for}.Math.t.Math.<\frac{T}{2}.$

(130) From a comparison of the signals a) and b) it is evident that the frequency twist is more robust in keeping the independence of the signal equation system. These feature is evident also from the time behavior of the twisted signals, as shown in FIGS. 23 and 24 which illustrate the time behavior of a time twisted mode +1 signal and of a frequency twisted mode +1 signal, respectively.

(131) From an ideal point of view the frequency Hilbert transform is applicable to a time-limited signal. Therefore, in order for the frequency Hilbert transform to be applicable to a continuous time symbol flow, it is necessary to apply said transform to successive time windows of said continuous time symbol flow and to identify the beginning and the end of each time window. This implies that the time window length is increased of a proper portion so as to render each time window detectable. This feature is similar to the bandwidth increase necessary in the case of time twisted waves.

(132) Therefore, the frequency rotation requires a time interval slightly larger than the minimum one required by the sampling theorem. This condition is equivalent to consider a symbol duration T.sub.sym longer than the system clock duration T.sub.cl, as schematically illustrated in FIG. 25.

(133) The above condition implies that, for instance, every 18 frequency bands an additional one is necessary and that, as a consequence, the bandwidth efficiency is given by 18/190.95. In this respect, FIG. 26 show symbol and clock period relationship in the case of 18 signal samples and 19 filter samples per frame.

(134) An interesting aspect of this condition applied to the frequency twisted waves is that it can be interpreted as equivalent to the well-known cyclic prefix already used with the OFDM technique.

(135) For multiple path transmission the delay spread is generated by the set of different paths between the transmitter and receiver when those paths have different delays.

(136) As an example, a signal following a direct line-of-sight path would arrive before a different version of the same signal which is reflected by a distant building.

(137) Time domain receivers typically synchronize with each delay spread component and adjust their individual timings before combining the received signals.

(138) When using a rake receiver, each finger belonging to the rake receiver synchronizes itself with a specific delay spread component. The number of delay spread components which can be combined is, thence, limited to the number of rake fingers. Any delay spread component which is not combined appears as interference.

(139) LTE receivers do not need to synchronize themselves with individual delay spread components, i.e., it is not necessary to adjust the timing of delay spread components, nor it is necessary to do any combining of delay spread components. An LTE receiver can operate directly on the aggregate received signal without considering delay spread components.

(140) The cyclic prefix represents a guard period at the start of each OFDMA symbol which provides protection against multi-path delay spread. The cyclic prefix also represents an overhead which should be minimized.

(141) The duration of the cyclic prefix should be greater than the duration of the multi-path delay spread.

(142) LTE specifies both normal and extended cyclic prefix lengths. The normal cyclic prefix is intended to be sufficient for the majority of scenarios, while the extended cyclic prefix is intended for scenarios with particularly high delay spread. Durations for the normal and extended cyclic prefix vary from 7% in the standard case up to 25% in the extended case. The cyclic prefix is generated by copying the end of the main body of the OFDMA symbol at the beginning, as shown in FIG. 27 which illustrates the traditional scheme of cyclic prefix for OFDM-OFDMA.

(143) The signal is always continuous at the interface between the cyclic prefix and the main body of the symbol. This results from the main body of the symbol always including an integer number of subcarrier cycles.

(144) FIG. 28 shows an example of 2 delay spread components. The second delay spread component is received later than the first delay spread component. An FFT processing window is defined at the receiver: the processing window captures the main body of the OFDMA symbol belonging to the first delay spread component; the cyclic prefix belonging to the first delay spread component is discarded; the processing window captures part of the cyclic prefix and the majority of the main body of the OFDMA symbol belonging to the second delay spread component; sections of the cyclic prefix and main body of the OFDMA symbol which fall outside the processing window are discarded; and, in the extreme case, where the delay spread is equal to the duration of the cyclic prefix, the FFT processing window fully captures the cyclic prefix belonging to the delay spread component and discards a section of the main body of the ODFMA symbol which has a duration equal to the cyclic prefix.

(145) The time domain representation of each delay spread component within the processing window is different, however, the frequency domain representation of each delay spread component within the processing window is identical.

(146) Let us now come back to the description of the present invention and let us assume that modes up to N=2 are used, then the band occupied by the TFU configuration is given by (2.sup.2+1+1)=9 frequency slots. To this TFU corresponds a Twisted Time frame Unit (TTU), which is increased to avoid time duration ambiguities. If one half slot is considered, the TTU increases of 1/(2B.sub.sym) and the total length of the TTU is 9.5/B.sub.sym. In this respect, FIG. 29 shows time frame duration increase due to cyclic prefix (CP).

(147) This increase is much lower than the one required by OFDMA. This implies that in practical system there is no additional loss for including frequency twisted waves in the OFDM (or OFDMA) super frame.

(148) The increase of time interval duration creates automatically a replica of the signal at the beginning of the time interval itself, without any change in the occupied frequency bandwidth.

(149) This approach is, thence, much more interesting for the understanding of the physical meaning of the cyclic prefix, than the ordinary explanation about its use.

(150) Considering sampling in frequency domain at a symbol rate slightly smaller than the clock rate, the signal in time domain, on the assumption that only the main mode is used, has the time behavior shown in FIG. 30, where the sinusoid are not exactly a multiple of the symbol period: i.e., what the cyclic prefix is performing.

(151) Adding the FTUs, the twisted mode signals present the same behavior of the main mode signal, as shown in FIG. 31.

(152) Increasing the number of TTUs, the number of sinusoidal signal increases, but the ratio between T.sub.cl and T.sub.sym remains unchanged.

(153) The OFDM-twisted frequency has two hierarchical levels: the former is related to the TTUs structure; and the latter is related to the assembly of the TFUs constituting the OFDM structure.

(154) Both the levels have the same time duration and the difference is given by the component frequency blocks: each TFU corresponds to a number of samples defined by the twisted frequency structure, which introduces additional frequency samples located between the main frequency samples; the super frame structure is a set of TTUs, centered at the proper frequency, and a set of traditional OFDM frequency samples, if wished; the standard frequency samples can simplify the process of synchronization and phasing.

(155) In order to consider the impact of thermal noise on the twisted waves, it is important to consider what happens on the time twisted waves, because there is a very interesting difference between the two families of twisted waves, which can have important applications in telecommunications, especially in the case of mobile LTE ones.

(156) The noise level for the time twisted waves can be represented as divided into two parts: a first part related to the symbol rate bandwidth; and a second part related to the difference between the symbol rate and the clock bandwidth.

(157) In this respect, FIG. 32 schematically illustrates noise bandwidths for time twisted waves: one defined according to minimum Nyquist bandwidth, the other related to the bandwidth increase for solving ambiguity aspects.

(158) The above noise structure can be written as:

(159) $n_t\left(t\right)e^{j2f_{0}t}=n_{\mathrm{int}B}\left(t\right)e^{j2f_{0}t}+\frac{1}{\sqrt{2}}{n}_{\mathrm{ext}}\left(t\right)\left[e^{j2\left(f_0+\frac{B+B/2}{2}\right)t}+e^{j2\left(f_0-\frac{B+B/2}{2}\right)t}\right],$
where n.sub.intB(t) denotes the noise part related to the symbol rate bandwidth, and n.sub.ext(t)=n.sub.extB(t)e.sup.j.sup.B.sup.(t) denotes the additional noise part due to the need of avoiding ambiguities on the symbol rate phasing.

(160) n.sub.intB(t), when sampled at the symbol rate, is an even function (no information on the odd sampling).

(161) Moreover,

(162) $\sqrt{2}{n}_{\mathrm{ext}B}\left(t\right)e^{j2f_{0}t}\cos \left[2\left(\frac{B+B/2}{2}\right)t+{}_B\left(t\right)\right]$
causes an additional contribution on the even and odd components.

(163) With reference to FIG. 32, the noise in the Nyquist band can be represented by a sinc time pulse, which is an even function, while the two sidebands can be represented by an even and an odd component.

(164) In terms of relationship between the noise components, it can be written:

(165) $\frac{{\mathrm{noise}}_{\mathrm{even}}}{{\mathrm{noise}}_{\mathrm{odd}}}\frac{B+B}{B}=1+\frac{B}{B}.$

(166) Considering the above for a simplified rect filter, the odd modes can be presented as:

(167) $\mathrm{POWER}{\mathrm{MODES}}_N=2\underset{k=1}{\overset{N}{.Math.}}{\left(\frac{2}{k}\right)}^2,$
for the first modes 1 there results

(168) $\frac{8}{{}^2}0.81;$

(169) ${\left(\frac{C}{N}\right)}_{\mathrm{odd}}=\frac{0.81B}{B+B}{}_t,$
for a bandwidth increase of 1 over 18 there results

(170) ${\left(\frac{C}{N}\right)}_{\mathrm{odd}}=-11.8+{}_{\mathrm{tdB}}\mathrm{dB}=-11.8+{\left(\frac{C}{N}\right)}_{\mathrm{even}}\mathrm{dB};$ the 11.8 dB term can be considered as the minimum gain against unwished interferences.

(171) Instead, as far as the case of frequency twisted waves is concerned, the noise spectrum occupies the bandwidth W and there is not any possibility of separating, in the time domain, its even and odd components for a single frequency pulse, as shown in FIG. 33 (which schematically illustrates noise impact in the case of frequency twisted waves).

(172) In fact, in the time domain the noise signal samples are not associated with the main signal samples, but they are distributed all along the time interval, which is utilized for reconstructing the frequency sampling value (FET). Therefore it is not possible to associate the main contribution of the noise to the main samples and there is no additional advantage for higher modes, as in the case of time rotation.

(173) The twisted waves add independent communication channels, one for each mode, and the information capacity increases with respect to the one of the single channel associated with the main mode.

(174) The above is valid for both frequency twist and time twist, but it is very interesting to analyze the similarities and the differences in order to optimize the use of the two twisting processes according to the overall system conditions.

(175) In general terms, it is possible to perform a system comparison on the basis of what presented in the foregoing and in PCT/FR2013/052636. In particular, the following TABLE I presents a comparison between time twisted waves and frequency twisted waves at system level, wherein supplementary references are provided for single carrier case and OFDM case.

(176) TABLE-US-00001 TABLE I Parameter assuming as reference Time Frequency OFDM S/N.sub.thermal = Twist Twist Single (parameters 13.9 dB QPSK (2 modes) (2 modes) carrier from LTE) Linearity 1.5 4.5 1.5 4.5 (HPA output back-off - dB) S/N with 10 12.7 10.5 12.7 interference (dB) Bandwidth 33% 0% 12.5% 0% increase (with respect to 1/T.sub.S, including roll-off) Time increase 0% 5% 0% 7-25% (with respect to 1/B.sub.S) Self-interference 28 28 35 35 (dB) Additional 0.5 0 0 0 thermal noise BER for main 6*10.sup.3 1.2*10.sup.4 3*10.sup.3 1.2*10.sup.4 mode (no code) BER for higher 8*10.sup.4 1.2*10.sup.4 NA NA modes (no code) Spectral efficiency 2.5 3.1-2.5 1.8 1.9-1.5 (bit/s/Hz) Shannon limit 3.5 5.4 3.6 5.4 (one channel - bit/s/Hz) Shannon limit 5.5 6.9 5.8 6.9 for twist

(177) In summary, time twist operates better in those cases in which the amplifiers work closer to the saturation, while frequency twist operates better when linearity can be preserved. That is a general condition for standard transmission too. In fact, it is well known that, in the case of LTE, on the Forward link (i.e., from the Base Station to mobile device) OFDM is used, while on the Return link (i.e., from the mobile device to the Base Station) Single Channel FDMA is preferred.

(178) FIG. 34 schematically shows (link) spectral efficiency (bit/s/Hz) as a function of the energy per symbol over noise density (E.sub.symbol/N.sub.0) for frequency twisted waves, on the assumption that there are present thermal noise and self-interference, which is the noise due to the interference between side-by-side frames.

(179) In particular, FIG. 34 shows that the behavior (in terms of bit/s/Hz) of frequency twisted wave modes presents values that are always lower than the main mode, due to the absence of the noise reduction for twisted modes.

(180) In the time domain the noise signal samples are not associated with the main signal samples, which are essentially complex values of frequency samples. They are distributed all along the time interval, which is utilized for reconstructing the frequency sampling value (FFT). Therefore, it is not possible to associate the main contribution of the noise with the main samples and there is no additional advantage for higher modes, as in the case of time rotation.

(181) As far as the transmitter according to the present invention is concerned, the generation of transmission signals is based on the transformation of a symbol serial time flow frames into a parallel flow for each frame, which is equivalent to the generation of a number of sinusoidal signal in the time window frame.

(182) This process is known and used for OFDM (or OFDMA) architecture; it is equivalent to an IFFT operation. In the case of OFDM, the frequency samples are spaced according to the sampling theorem applied to the frequency domain.

(183) When using frequency twisted waves, it is necessary to oversampling the overall frequency band via the introduction of additional frequency samples, spaced as previously defined for each mode.

(184) In this respect, it is worth recalling that generic modes N are generated repeating the same symbol at 2.sup.N+1 different frequencies

(185) $\frac{1}{T}\left(\frac{2^N-1}{2^N}+k\right),$
changing each time their phases according to

(186) $e^{jk\frac{}{2^N}},$
with k=0, 1, . . . , 2.sup.N+11. This means that the associated IFFT is the sum of 4N decimated IFFT, having only 1 row for each sample and each one is weighted as

(187) 0$\frac{1}{2^{\frac{N+1}{2}}}{e}^{jk\frac{}{2^N}}.$

(188) A preferred use of frequency twisted waves is inside an OFDM-OFDMA architecture. Taking into consideration that an OFDM structure includes a very large number of frequencies, a possible architecture is proposed here below.

(189) On the assumption that modes up to N=2 are used, the band occupied by this configuration is given by 2.sup.2+1+1=9 frequency slots. This section is called Twisted Frequency frame Unit (TFU) and to this TFU corresponds a Twisted Time frame Unit (TTU).

(190) The inclusion of the TFU cyclic prefix increases the time frame by

(191) $\frac{0.5}{B_S}$
and, thence, for the TTU there results

(192) $\frac{9.5}{B_S}.$

(193) The cyclic prefix is used for each TFU present in the full OFDM-OFDMA bandwidth (and, as previously explained, is physically the same one used for OFDM, but used for each TFU).

(194) As previously explained, the OFDM-twisted frequency has two hierarchical levels: the former is related to the TTUs structure, which depends on the number of modes chosen and the number of frequency slots adopted; and the latter is related to the assembly of the TFUs constituting the OFDM structure.

(195) Again as previously explained, both the levels have the same time duration and the difference is given by the component frequency blocks: each TFU corresponds to a number of samples defined by the twisted frequency structure, which introduces additional frequency samples located between the main frequency samples; the super frame structure is a set of TTUs, centered at the proper frequency, and a set of traditional OFDM frequency samples, if wished; the standard frequency samples can simplify the process of synchronization and phasing.

(196) The generation of the main mode signal and of the twisted mode signals via this process is called, as previously explained, Generalized Inverse Fast Fourier Transform (GIFFT).

(197) For the sake of simplicity, it is assumed to use twisted modes 1 and 2. This implies, as previously explained, the presence of 9 frequency slots related to the main mode; the frequency rotation requires a time interval slightly larger than the minimum one required by the sampling theorem in order to avoid ambiguities, due to the determination of the frame boundary; this implies that, for instance, every 18 frequency band one additional one is necessary; therefore, there are two time reference window: one defined by the clock T.sub.cl, and one defined by the symbol T.sub.sym. The relationship between T.sub.cl and T.sub.sym is given by:

(198) $T_{\mathrm{sym}}=\frac{2^{k+2}+3}{2^{k+2}+2}{T}_{\mathrm{cl}},\mathrm{where}$$\frac{1}{2^{k+2}+2}{T}_{\mathrm{cl}}$
is equivalent to the cyclic prefix.

(199) In the present case, T.sub.sym=19/18T.sub.cl, and 1/18T.sub.cl is the cyclic prefix.

(200) In order to create the correct reference between the real and the imaginary signals, it is important to avoid possible ambiguities on the zero of the reference system. In fact, this system shall be used as the reference system of the principal value of the Cauchy integral.

(201) Therefore, it is important to have a sampling rate slightly larger than the minimum possible for the symbol rate associated with the plane wave mode.

(202) The frequency representation is shown in FIG. 35, which schematically illustrates frequency slots with the additional one for ambiguity resolution (always on the assumption that twisted modes up to 2 are used). This additional frequency slot implies that the bandwidth of each symbol is smaller than the maximum one given by Nyquist criterion, and that the additional time duration creates the repetition of part of the sinusoidal at the beginning of the time slot (as in the case of traditional cyclic prefix).

(203) The main mode has the same structure of the traditional IFFT, but the sampling is performed 19 times instead of 18.

(204) Higher-order modes are generated considering that each of them can be derived considering a set of frequency pulses properly shifted in frequency and phased (either 1;j;1;+j).

(205) The frequency pulses of each mode are properly positioned via a transformation algorithm, which is very similar to the IFFT, having in mind the fact that the starting frequency is properly positioned on the frequency axis and that more samples are associated with the same symbol, as explained in the foregoing.

(206) In order for the operation of the present invention to be better understood, reference is made to FIG. 36, which shows a functional block diagram of a transmitting system (denoted as whole by 7) according to an illustrative embodiment of the present invention.

(207) In particular, the transmitting system 7 shown in FIG. 36 is designed to generate frequency twisted waves up to modes 3, and comprises: a symbol generation section 70 configured to generate and output a digital symbol stream; a frequency twisted mode generation unit 700 based on GIFFT, said frequency twisted mode generation unit 700 being coupled with the symbol generation section 70 to receive the digital symbol stream outputted by the latter, and being configured to generate and output, for each sequence of S.sub.TOT digital symbols received from the symbol generation section 70 (in particular, in the example shown in FIG. 36, S.sub.TOT=31), a respective digital time signal obtained by transforming from frequency domain to time domain main mode frequency samples carrying M.sub.MFS of said S.sub.TOT received digital symbols (in particular, in the example shown in FIG. 36, M.sub.MFS=17) via a frequency main mode (preferably, as previously explained, the main mode frequency samples are OFDM/OFDMA-type frequency samples), and twisted mode frequency samples carrying the other S.sub.TOT-M.sub.MFS received digital symbols via frequency twisted modes, wherein, in the example shown in FIG. 36, the twisted mode frequency samples include frequency samples which are related to frequency twisted mode +1 and which carry four respective digital symbols via the frequency twisted mode +1, frequency samples which are related to frequency twisted mode 1 and which carry four respective digital symbols via the frequency twisted mode 1, frequency twisted mode which are related to frequency twisted mode +2 and which carry two respective digital symbols via the frequency twisted mode +2, frequency twisted mode which are related to frequency twisted mode 2 and which carry two respective digital symbols via the frequency twisted mode 2, frequency twisted mode which are related to frequency twisted mode +3 and which carry one respective digital symbol via the frequency twisted mode +3, and frequency twisted mode which are related to frequency twisted mode 3 and which carry one respective digital symbol via the frequency twisted mode 3; and an RF transmission section 7000 which is coupled with the frequency twisted mode generation unit 700 to receive the digital time signals outputted by the latter, and which is configured to transmit at predefined radio frequencies the received digital time signals by means of a single antenna or a plurality of antennas/antenna elements (not shown in FIG. 36 for the sake of illustration simplicity).

(208) Conveniently, the aforesaid predefined radio frequencies can range from a few KHz to hundreds of GHz depending on the specific application for which the transmitting system 7 is designed.

(209) Preferably, the transmitting system 7 is a device/system for wireless communications based on OFDM and/or OFDMA, or, more preferably, on LTE and/or WiMAX.

(210) Conveniently, the symbol generation section 70 is designed to generate the digital symbol stream by performing several operations, such as the following operations (not necessarily all performed and not necessarily performed in the following sequence): information encoding (conveniently by performing one or more signal modulations), one or more frequency shifting operations, one or more analog-to-digital conversion operations, and one or more filtering operations.

(211) Again conveniently, the RF transmission section 7000 can be designed to transmit at the predefined radio frequencies the digital time signals by performing several operations, such as the following operations (not necessarily all performed and not necessarily performed in the following sequence): frequency up-shifting (in particular. from Intermediate Frequency (IF) up to RF), one or more filtering operations, one or more digital-to-analog conversion operations, and power amplification.

(212) More in detail, as shown in FIG. 36, the frequency twisted mode generation unit 700 includes: a frequency main mode generation module 701, which, in use, determines, for each of said M.sub.MFS=17 digital symbols, a corresponding symbol complex value a.sub.pe.sup.j.sup.p (with p=1, 2, . . . , M.sub.MFS) which represents said digital symbol, allocates each of the M.sub.MFS symbol complex values to a respective frequency p/T, or (since T=1/B.sub.S) p.Math.B.sub.S (as in the case of traditional OFDM/OFDMA technique), thereby obtaining M.sub.MFS main mode frequency samples, and performs an IFFT of all the M.sub.MFS main mode frequency samples, thereby generating a digital time signal related to the frequency main mode; a frequency twisted mode +1 generation module 702, which, in use, determines, for the first of the four respective digital symbols, a corresponding symbol complex value a.sub.+1;1e.sup.j.sup.+1;1 which represents said digital symbol, allocates said symbol complex value a.sub.+1;1e.sup.j.sup.+1;1 to four respective frequencies B.sub.S(+k) (with k=0, 1, 2, 3) changing, for each frequency sample, the respective phase according to

(213) $e^{+jk\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.+1;1e.sup.j.sup.+1;1 by a respective complex coefficient

(214) $\frac{e^{+jk\frac{}{2}}}{2}),$
thereby obtaining four twisted mode frequency samples which are related to the frequency twisted mode +1 and which carry said first respective digital symbol via said frequency twisted mode +1, determines, for the second of the four respective digital symbols, a corresponding symbol complex value a.sub.+1;2e.sup.j.sup.+1;2 which represents said digital symbol, allocates said symbol complex value a.sub.+1;2e.sup.j.sup.+1;2 to four respective frequencies B.sub.S(+k) (with k=4, 5, 6, 7) changing, for each frequency sample, the respective phase according to

(215) $e^{+j\left(k-4\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.+1;2e.sup.j.sup.+1;2 by a respective complex coefficient

(216) $\frac{e^{+j\left(k-4\right)\frac{}{2}}}{2}),$
thereby obtaining four further twisted mode frequency samples which are related to the frequency twisted mode +1 and which carry said second respective digital symbol via said frequency twisted mode +1, determines, for the third of the four respective digital symbols, a corresponding symbol complex value a.sub.+1;3e.sup.j.sup.+1;3 which represents said digital symbol, allocates said symbol complex value a.sub.+1;3e.sup.j.sup.+1;3 to four respective frequencies B.sub.S(+k) (with k=8, 9, 10, 11) changing, for each frequency sample, the respective phase according to

(217) $e^{+j\left(k-8\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.+1;3e.sup.j.sup.+1;3 by a respective complex coefficient

(218) $\frac{e^{+j\left(k-8\right)\frac{}{2}}}{2}),$
thereby obtaining four further twisted mode frequency samples which are related to the frequency twisted mode +1 and which carry said third respective digital symbol via said frequency twisted mode +1, determines, for the fourth of the four respective digital symbols, a corresponding symbol complex value a.sub.+1;4e.sup.j.sup.+1;4 which represents said digital symbol, allocates said symbol complex value a.sub.+1;4e.sup.j.sup.+1;4 to four respective frequencies B.sub.S(+k) (with k=12, 13, 14, 15) changing, for each frequency sample, the respective phase according to

(219) 0$e^{+j\left(k-12\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.+1;4e.sup.j.sup.+1;4 by a respective complex coefficient

(220) $\frac{e^{+j\left(k-12\right)\frac{}{2}}}{2}),$
thereby obtaining four final twisted mode frequency samples which are related to the frequency twisted mode +1 and which carry said fourth respective digital symbol via said frequency twisted mode +1, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode +1, thereby generating a digital time signal related to the frequency twisted mode +1; a frequency twisted mode 1 generation module 703, which, in use, determines, for the first of the four respective digital symbols, a corresponding symbol complex value a.sub.1;1e.sup.j.sup.1;1 which represents said digital symbol, allocates said symbol complex value a.sub.1;1e.sup.j.sup.1;1 to four respective frequencies B.sub.S(+k) (with k=0, 1, 2, 3) changing, for each frequency sample, the respective phase according to

(221) $e^{-jk\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.1;1e.sup.j.sup.1;1 by a respective complex coefficient

(222) $\frac{e^{-jk\frac{}{2}}}{2}),$
thereby obtaining four twisted mode frequency samples which are related to the frequency twisted mode 1 and which carry said first respective digital symbol via said frequency twisted mode 1, determines, for the second of the four respective digital symbols, a corresponding symbol complex value a.sub.1;2e.sup.j.sup.1;2 which represents said digital symbol, allocates said symbol complex value a.sub.1;2e.sup.j.sup.1;2 to four respective frequencies B.sub.S(+k) (with k=4, 5, 6, 7) changing, for each frequency sample, the respective phase according to

(223) $e^{-j\left(k-4\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.1;2e.sup.j.sup.1;2 by a respective complex coefficient

(224) $\frac{e^{-j\left(k-4\right)\frac{}{2}}}{2}),$
thereby obtaining four further twisted mode frequency samples which are related to the frequency twisted mode 1 and which carry said second respective digital symbol via said frequency twisted mode 1, determines, for the third of the four respective digital symbols, a corresponding symbol complex value a.sub.1;3e.sup.j.sup.1;3 which represents said digital symbol, allocates said symbol complex value a.sub.1;3e.sup.j.sup.1;3 to four respective frequencies B.sub.S(+k) (with k=8, 9, 10, 11) changing, for each frequency sample, the respective phase according to

(225) $e^{-j\left(k-8\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.1;3e.sup.j.sup.1;3 by a respective complex coefficient

(226) $\frac{e^{-j\left(k-8\right)\frac{}{2}}}{2}),$
thereby obtaining four further twisted mode frequency samples which are related to the frequency twisted mode 1 and which carry said third respective digital symbol via said frequency twisted mode 1, determines, for the fourth of the four respective digital symbols, a corresponding symbol complex value a.sub.1;4e.sup.j.sup.1;4 which represents said digital symbol, allocates said symbol complex value a.sub.1;4e.sup.j.sup.1;4 to four respective frequencies B.sub.S(+k) (with k=12, 13, 14, 15) changing, for each frequency sample, the respective phase according to

(227) $e^{-j\left(k-12\right)\frac{}{2}}$
and weighting each frequency sample by (i.e., multiplying, for each of the four respective frequencies, the symbol complex value a.sub.1;4e.sup.j.sup.1;4 by a respective complex coefficient

(228) $\frac{e^{-j\left(k-12\right)\frac{}{2}}}{2}),$
thereby obtaining four final twisted mode frequency samples which are related to the frequency twisted mode 1 and which carry said fourth respective digital symbol via said frequency twisted mode 1, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode 1, thereby generating a digital time signal related to the frequency twisted mode +1; a frequency twisted mode +2 generation module 704, which, in use, determines, for the first of the two respective digital symbols, a corresponding symbol complex value a.sub.+2;1e.sup.j.sup.+2;1 which represents said digital symbol, allocates said symbol complex value a.sub.+2;1e.sup.j.sup.+2;1 to eight respective frequencies B.sub.S(+k) (with k=0, 1, . . . , 7) changing, for each frequency sample, the respective phase according to

(229) 0$e^{+jk\frac{}{4}}$
and weighting each frequency sample by 1/{square root over (8)} (i.e., multiplying, for each of the eight respective frequencies, the symbol complex value a.sub.+2;1e.sup.j.sup.+2;1 by a respective complex coefficient

(230) $\frac{e^{+jk\frac{}{4}}}{\sqrt{8}}),$
thereby obtaining eight twisted mode frequency samples which are related to the frequency twisted mode +2 and which carry said first respective digital symbol via said frequency twisted mode +2, determines, for the second of the two respective digital symbols, a corresponding symbol complex value a.sub.+2;2e.sup.j.sup.+2;2 which represents said digital symbol, allocates said symbol complex value a.sub.+2;2e.sup.j.sup.+2;2 to eight respective frequencies B.sub.S(+k) (with k=8, 9, . . . , 15) changing, for each frequency sample, the respective phase according to

(231) $e^{+j\left(k-8\right)\frac{}{4}}$
and weighting each frequency sample by 1/{square root over (8)} (i.e., multiplying, for each of the eight respective frequencies, the symbol complex value a.sub.+2;2e.sup.j.sup.+2;2 by a respective complex coefficient

(232) $\frac{e^{+j\left(k-8\right)\frac{}{4}}}{\sqrt{8}}),$
thereby obtaining eight further twisted mode frequency samples which are related to the frequency twisted mode +2 and which carry said second respective digital symbol via said frequency twisted mode +2, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode +2, thereby generating a digital time signal related to the frequency twisted mode +2; a frequency twisted mode 2 generation module 705, which, in use, determines, for the first of the two respective digital symbols, a corresponding symbol complex value a.sub.2;1e.sup.j.sup.2;1 which represents said digital symbol, allocates said symbol complex value a.sub.2;1e.sup.j.sup.2;1 to eight respective frequencies B.sub.S(+k) (with k=0, 1, . . . , 7) changing, for each frequency sample, the respective phase according to

(233) $e^{-jk\frac{}{4}}$
and weighting each frequency sample by 1/{square root over (8)} (i.e., multiplying, for each of the eight respective frequencies, the symbol complex value a.sub.2;1e.sup.j.sup.2;1 by a respective complex coefficient

(234) $\frac{e^{-jk\frac{}{4}}}{\sqrt{8}}),$
thereby obtaining eight twisted mode frequency samples which are related to the frequency twisted mode 2 and which carry said first respective digital symbol via said frequency twisted mode 2, determines, for the second of the two respective digital symbols, a corresponding symbol complex value a.sub.2;2e.sup.j.sup.2;2 which represents said digital symbol, allocates said symbol complex value a.sub.2;2e.sup.j.sup.2;2 to eight respective frequencies B.sub.S(+k) (with k=8, 9, . . . , 15) changing, for each frequency sample, the respective phase according to

(235) $e^{-j\left(k-8\right)\frac{}{4}}$
and weighting each frequency sample by 1/{square root over (8)} (i.e., multiplying, for each of the eight respective frequencies, the symbol complex value a.sub.2;2e.sup.j.sup.2;2 by a respective complex coefficient

(236) $\frac{e^{-j\left(k-8\right)\frac{}{4}}}{\sqrt{8}}),$
thereby obtaining eight further twisted mode frequency samples which are related to the frequency twisted mode 2 and which carry said second respective digital symbol via said frequency twisted mode 2, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode 2, thereby generating a digital time signal related to the frequency twisted mode 2; a frequency twisted mode +3 generation module 706, which, in use, determines, for the respective digital symbol, a corresponding symbol complex value a.sub.+3e.sup.j.sup.+3 which represents said digital symbol, allocates said symbol complex value a.sub.+3e.sup.j.sup.3 to sixteen respective frequencies B.sub.S(+k) (with k=0, 1, . . . , 15) changing, for each frequency sample, the respective phase according to

(237) $e^{+jk\frac{}{8}}$
and weighting each frequency sample by (i.e., multiplying, for each of the sixteen respective frequencies, the symbol complex value a.sub.+3e.sup.j.sup.+3 by a respective complex coefficient

(238) $\frac{e^{+jk\frac{}{8}}}{4}),$
thereby obtaining sixteen twisted mode frequency samples which are related to the frequency twisted mode +3 and which carry said respective digital symbol via said frequency twisted mode +3, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode +3, thereby generating a digital time signal related to the frequency twisted mode +3; a frequency twisted mode 3 generation module 707, which, in use, determines, for the respective digital symbol, a corresponding symbol complex value a.sub.3e.sup.j.sup.3 which represents said digital symbol, allocates said symbol complex value a.sub.3e.sup.j.sup.3 to sixteen respective frequencies B.sub.S(+k) (with k=0, 1, . . . , 15) changing, for each frequency sample, the respective phase according to

(239) 0$e^{-jk\frac{}{8}}$
and weighting each frequency sample by (i.e., multiplying, for each of the sixteen respective frequencies, the symbol complex value a.sub.3e.sup.j.sup.2 by a respective complex coefficient

(240) $\frac{e^{-jk\frac{}{8}}}{4}),$
thereby obtaining sixteen twisted mode frequency samples which are related to the frequency twisted mode 3 and which carry said respective digital symbol via said frequency twisted mode 3, and performs an IFFT of all the sixteen twisted mode frequency samples related to the frequency twisted mode 3, thereby generating a digital time signal related to the frequency twisted mode 3; and a combining module 708, which, in use, combines (namely, adds together) the digital time signals outputted by the frequency main mode generation module 701 (i.e., the digital time signal related to the frequency main mode), by the frequency twisted mode +1 generation module 702 (i.e., the digital time signal related to the frequency twisted mode +1), by the frequency twisted mode 1 generation module 703 (i.e., the digital time signal related to the frequency twisted mode 1), by the frequency twisted mode +2 generation module 704 (i.e., the digital time signal related to the frequency twisted mode +2), by the frequency twisted mode 2 generation module 705 (i.e., the digital time signal related to the frequency twisted mode 2), by the frequency twisted mode +3 generation module 706 (i.e., the digital time signal related to the frequency twisted mode +3), and by the frequency twisted mode 3 generation module 707 (i.e., the digital time signal related to the frequency twisted mode 3), thereby generating an overall digital time signal.

(241) Linearity of the frequency twisted mode generation unit 700 is important, due to the presence of a wide multicarrier architecture.

(242) Conveniently, the frequency twisted mode generation unit 700 carries out all the aforesaid operations of by using an overall complex transmission matrix [[GIFFT]] designed to implement, in a combined way and at one and the same time, all the aforesaid operations so that, when applied to a sequence of S.sub.TOT digital symbols received from the symbol generation section 70, time samples of the corresponding digital time signal are automatically computed by the frequency twisted mode generation unit 700.

(243) Preferably, for each digital time signal generated and outputted by the combining module 708, the frequency twisted mode generation unit 700 is further designed to insert, at the beginning of said digital time signal, a respective cyclic prefix which is a replica of an end portion of said digital time signal (in accordance with what was previously explained).

(244) Let us now consider the operation of the present invention at reception side, and, in this respect, reference is made to FIG. 37, which shows a functional block diagram of a receiving system (denoted as whole by 8) according an illustrative embodiment of the present invention.

(245) In particular, as shown in FIG. 37, the receiving system 8 comprises: an RF reception section 8000, which is designed to receive the RF signals transmitted at the predefined radio frequencies by the transmitting system 7 (in particular, by the RF transmission section 7000); said RF reception section 8000 being designed to receive the RF signals by means of a single antenna or a plurality of antennas/antenna elements (not shown in FIG. 37 for the sake of illustration simplicity), and to process the received RF signals so as to obtain, on the basis of said received RF signals, an incoming digital signal; a symbol extraction unit 800 based on GFFT, which is coupled with said RF reception section 8000 to receive the incoming digital signal therefrom, and which is designed to process said incoming digital signal so as to extract the digital symbols carried by said incoming digital signal, and output a stream of extracted digital symbols; and a symbol processing section 80, which is coupled with said symbol extraction unit 800 to receive the stream of extracted digital symbols outputted by the latter, and which is designed to process said stream of extracted digital symbols.

(246) The aforesaid predefined radio frequencies coincide with the radio frequencies used in transmission by the transmitting system 7, in particular by the RF transmission section 7000. Conveniently, as already said, the predefined radio frequencies can range from a few KHz to hundreds of GHz depending on the specific application which the overall radio communications system comprising the transmitting system 7 and the receiving system 8 is designed for.

(247) Preferably, the receiving system 8 is a device/system for wireless communications based on OFDM and/or OFDMA, or, more preferably, on LTE and/or WiMAX.

(248) Conveniently, the RF reception section 8000 is designed to obtain the incoming digital signal by performing several operations upon the received RF signals, such as the following operations (not necessarily all performed and not necessarily performed in the following sequence): low-noise amplification, one or more frequency down-shifting operations (in particular from RF down to IF), one or more filtering operations, and one or more analog-to-digital conversion operations.

(249) Again conveniently, the symbol processing section 80 is designed to process the stream of extracted digital symbols by performing several operations, such as the following operations (not necessarily all performed and not necessarily performed in the following sequence): one or more filtering operations, one or more digital-to-analog conversion operations, one or more frequency shifting operations, and information decoding (conveniently by performing one or more signal demodulations).

(250) At the reception side the parallel signal flow is to be considered, as in the case of OFDM (or OFDMA), and a reception matrix [[GFFT]] is used by the symbol extraction unit 800 to extract the digital symbols carried by the incoming digital signal.

(251) The main difference with respect to the standard OFDM signal is that OFDM exploits Hermitian matrices, while in the case of frequency twisted waves the transmission matrix [[GIFFT]] is rectangular and, thence, in order for the reception matrix [[GFFT]] to be obtained, pseudo-inverse approach is exploited. The use of such a procedure is called Generalized Fast Fourier Transform (GFFT) and is somewhat similar to the Generalized Matched Filter used for the time twisted waves and described in PCT/FR2013/052636.

(252) In use, the symbol extraction unit 800 processes the incoming digital signal by using a time window T.sub.Sym (including the cyclic prefix) so as to process successive, non-overlapped portions of the incoming digital signal each having a time duration equal to T.sub.Sym, and to extract the digital symbols respectively carried by each incoming digital signal portion.

(253) The input sequence in the time window T.sub.Sym is oversampled with the same law of the transmission flow; on the assumption that modes up to N are used, the size of the reception matrix is given by: the number of unknowns (i.e., symbol complex values) S.sub.TOT=2.sup.N+21 in a frequency frame of M.sub.MFS=2.sup.N+1+1 main mode frequency pulses; and the number of equations, which represents also the overall number M.sub.TS of the samples in time domain, which is given by

(254) $M_{\mathrm{TS}}=\left(1+\frac{N\left(N+1\right)}{2}\right)2^{N+1}+1.$

(255) More in detail, in order to solve the equation system at the reception side, a reception matrix [[GFFT]] is used by the symbol extraction unit 800, which reception matrix [[GFFT]] is derived from the transmission matrix [[GIFFT]] through a generalized inversion technique, such as the pseudo-inverse technique.

(256) In mathematical terms, given the transmission matrix [[GIFFT]] with M.sub.TSxS.sub.TOT complex coefficients, and given also the vector [S] of the S.sub.TOT symbol complex values to be transmitted, at transmission side there results that:
[[GIFFT]][S]=[TTU]
where [TTU] denote the vector of the M.sub.TS complex values of the time samples of a digital time signal outputted by the frequency twisted mode generation unit 700.

(257) Let us now consider the reception side, where it is useful to use a generalized inversion technique, such as the pseudo-inverse technique, to invert the foregoing matrix equation:
[[GIFFT]].sup.T[[GIFFT]][S]=[[GIFFT]].sup.T[TTU],
and thence
[S]=([[GIFFT]].sup.T[[GIFFT]]).sup.1[[GIFFT]].sup.T[TTU],(1)
where [[GIFFT]].sup.T denotes the transpose of the matrix [[GIFFT]], and ([[GIFFT]].sup.T [[GIFFT]]).sup.1 denotes the operation of inversion of the square matrix resulting from the multiplication [[GIFFT]].sup.T [[GIFFT]].

(258) In particular, at reception side [S] becomes the vector of the S.sub.TOT unknown symbol complex values to be determined by the symbol extraction unit 800, and [TTU] becomes the vector of the M.sub.TS complex values of the time samples determined by the symbol extraction unit 800 on the basis of an incoming digital signal portion.

(259) Condition for the existence of a set of solutions for the unknown vector [S] is that the square matrix resulting from the multiplication [[GIFFT]].sup.T [[GIFFT]] has a determinant different than zero, i.e., in mathematical terms,
det([[A]].sup.T[A])0.(2)

(260) Therefore, if the transmission matrix [[GIFFT]] is designed so as to satisfy the condition (2), then the S.sub.TOT unknown symbol complex values can be determined by the symbol extraction unit 800 by solving the equation system resulting from the matrix equation (1).

(261) Thence, the reception matrix [[GFFT]], which is a non-Hermitian matrix, can be defined as:
[[GFFT]]=([[GIFFT]].sup.T[[GIFFT]]).sup.1[[GIFFT]].sup.T.

(262) In this respect, FIG. 38 schematically shows an example of square matrix resulting from the multiplication [[GIFFT]].sup.T [[GIFFT]] on the assumption that modes up to 1 are used. In particular, the matrix shown in FIG. 38 includes cells which are blank or grey, wherein the grey cells represent the matrix cells actually occupied by coefficients, while the blank cells represent the matrix cells not occupied by any coefficient. This representation of the matrix stems from the fact that the FIG. 38 is mainly intended to show the matrix structure (and not the matrix coefficients).

(263) The condition (2) is satisfied more easily in the frequency twist case than in the time twist one, as it can inferred by looking at the shape of the time signals. The main reason for such a behaviour is based on the fact that a frequency function is intrinsically complex, while a time signal is real. In other words, the square matrix resulting from the multiplication [[GIFFT]].sup.T [[GIFFT]] is much more robust than the similar matrix obtained in the case of time twisted waves.

(264) It is important to note that the determinant is well sized and does not require an increase of the bandwidth as in the time twisted wave case. In fact, changing from 19 to 18 samples the determinant relative value changes from 1 to about 0.1, which are both values valid for the matrix inversion.

(265) Ideally, the use of the cyclic prefix allows interference level to be limited close to zero. This is true when a large number of side lobes are present outside the useful bandwidth. For the TFUs the bandwidth is limited to the first side-lobe of the frequency pattern, therefore the interference level is of the order of about 30/35 dB. This can be considered self-generated noise due to inter-frame interference.

(266) The presence of self-generated noise produces a limitation on the information transmission capacity when the E.sub.symbol/N.sub.0 is very high (namely, higher than 40 dB), as shown in FIG. 34. This limitation is present in all the cases where the physical resource is reused (e.g. orthogonal polarization limit 35-40 dB).

(267) Sizing and configuration of a transmitter and a receiver using frequency twisted waves according to the present invention can be considered an innovative updating of the OFDM, OFDMA and COFDM (i.e., Coded OFDM) architectures.

(268) In this respect, FIG. 39 schematically shows a multilayer architecture wherein the frame structure of the frequency twisted waves is embedded in a traditional OFDM architecture.

(269) The proposed architecture allows the OFDM basic structure to be used with the additional layer of frequency twisted waves (including the size and the references of the frequency twisted wave frame, i.e., the frequency slot positions, their phases and the association of these frequencies with the transmitted symbols, as described in detail in the foregoing).

(270) The advantage of frequency twisted waves with respect to time twisted waves is evident in this aspect; in fact, for frequency twisted waves there is no need to build up a dedicated space reference system as in the case of time twist and this is due to the important consideration that OFDM and similar transmission techniques are already block signal transmission techniques (i.e., implement simultaneous transmission of signals using IFFT).

(271) The size of transmission block is increased by a factor related to the number of frequency twisted modes used. In fact, as previously explained, each twisted frequency mode n exploits a sequence of 2.sup.N+1 additional frequency carriers (where 1V are the highest modes used) positioned at

(272) $B_S\left(\frac{2^n-1}{2^n}+k\right),$
where 0k2.sup.N+11.

(273) The receiver (in particular, the symbol extraction unit 800) handles a number of unknowns S.sub.TOT smaller than the number of equations M.sub.TS (as previously explained), this implies the use of the pseudo-inverse technique and an increase of the computational complexity with respect to the usual OFDM block signal computation, which, as is known, produces square matrixes L.sub.SFxL.sub.SF, where L.sub.SF is the length of the super frame.

(274) Instead, in the case of frequency twisted waves, assuming the same super frame length L.sub.SF and a frame length L.sub.TFU equal to 2.sup.N+1+1, the number of equation for said super frame is:

(275) $M_{\mathrm{TS}}\frac{L_{\mathrm{SF}}}{2^{N+1}+1}=\left[\left(1+\frac{N\left(N+1\right)}{2}\right)2^{N+1}+1\right]\frac{L_{\mathrm{SF}}}{2^{N+1}+1},$
where, if N=2 and L.sub.SF=2016, the number of equations is 7392 and the number of the unknowns is given by

(276) $\frac{2^{N+2}+1}{2^{N+1}+1}{L}_{\mathrm{SF}}=3808.$

(277) The increased complexity factor of the matrix operations is given by the ratio between the number of operations of twisted waves and the number of OFDM operations:

(278) $\frac{\left(\frac{2^{N+2}+1}{2^{N+1}+1}{L}_{\mathrm{SF}}\right)\left[\left(1+\frac{N\left(N+1\right)}{2}\right)2^{N+1}+1\right]\frac{L_{\mathrm{SF}}}{2^{N+1}+1}}{{\left(L_{\mathrm{SF}}\right)}^2}=6.1.$

(279) In this respect, FIG. 40 shows computational complexity of the present invention and frequency reuse according to the latter as a function of N (where, as previously said, N are the highest twisted frequency modes used). From FIG. 40 it is evident that a reasonable compromise is obtained with N=2, with a very good frequency reuse and a limited increase in computational complexity.

(280) The introduction of an additional layer in the super frame organization is somewhat limiting the possibility of OFDM adapting itself to the channel characteristics; i.e., in a traditional OFDM frame the adaptability is given by the single carrier bandwidth B.sub.S, while for the frequency twisted wave case the adaptability is given by (2.sup.N+1+1) B.sub.S. This is a limitation on the system adaptability and has the same trend of the computational complexity. In this respect, FIG. 41 schematically illustrates flexibility in using OFDM modularity, complex equation number and implementation criticality of the present invention as a function of N (where, as previously said, N are the highest twisted frequency modes used).

(281) Some features of the present invention are briefly summarized here below: due the structure of the OFDM signal there is no additional noise due to the introduction of frequency twisted waves; the OFDM cyclic prefix includes the equivalent cyclic prefix necessary for frequency twisted waves; anyway, it is clear that, if the cyclic prefix is fully used for OFDM, it should be accordingly increased; and there is no practical advantage of performance in using frequency twisted mode higher than 3, but, on the contrary, computational complexity grows quite rapidly.

(282) As explained in the foregoing, the implementation of the frequency twisted waves according to the present invention can be regarded as an approximation of the frequency Hilbert transform. This fact implies, on one side, a bandwidth increase, and, on the other side, the presence of an absolute limitation on the increase in frequency reuse, which is lower than two. In this respect, the following TABLE II lists some features related to the use of frequency twisted waves according to the present invention.

(283) TABLE-US-00002 TABLE II Parameter value Parameter (considering using approximate value Parameter up to modes N) for N = 2 Frequency reuse $\frac{2^{N+2}-1}{2^{N+1}+1}$ 1.67 Vestigial time interval increase $\frac{2^{N+2}+3}{2^{N+2}+2}$ 1.056 Total frame bandwidth/symbol bandwidth $\frac{B_F}{B_S}=\frac{2^{N+2}+3}{2}$ 9.5 Super Frame loss <1% 0.99 Additional bandwidth noise (dB) 0$10\log \left(\frac{2^{N+3}+3}{2^{N+3}+2}\right)\mathrm{dB}$ 0.25 dB Digitalization noise < 30 dB phase error $\frac{N}{2^N-1}$ 9 bits Maximum inter- <19-3N dB <25 dB frame interference (dB)

(284) For N=3, the frame length is smaller than 32 symbols, the necessary number of bits is about 10, the increase of the thermal noise is close to 0 dB, and the frequency reuse close to 1.7.

(285) As far as practical implementation of the present invention is concerned, the frequency twisted mode generation unit 700 based on GIFFT and the symbol extraction unit 800 based on GFFT are preferably implemented by means of Field-Programmable Gate Array (FPGA), Application-Specific Integrated Circuit (ASIC), and Software Defined Radio (SDR) technologies.

(286) From the foregoing, it may be immediately appreciated that the present invention allows to increase frequency reuse and transmission capacity by exploiting an original application of the Hilbert transform in frequency domain.

(287) The present invention can be considered very interesting and almost revolutionary to develop a new theory for digital communications beyond the classical approach based on analytical signals.

(288) In particular, as previously explained in detail, according to the present invention radio vorticity is considered as a way to approximate the Hilbert transform and is applied in frequency domain so as to generate independent radio channels within one and the same bandwidth. These channels have an available bandwidth decreasing with the radio vorticity mode number and the total bandwidth advantage is growing as .sup.N, limited by 2, which represents the maximum possible use of the imaginary channel of the Hilbert transform.

(289) From a mathematical (and physical) perspective, this Hilbert-transform-based approach is very similar to an interferometry measurement performed in frequency instead of in geometrical space.

(290) The present invention can be advantageously exploited, in general, in all kinds of radio communications, and, in particular, in radio communications based, in general, on OFDM and/or OFDMA, and, specifically, on LTE and/or WiMAX.

(291) Finally, it is worth noting that a combined use of frequency twist according to the present invention and time twist according to PCT/FR2013/052636 is particularly advantageous in asymmetrical radio communications systems, such as mobile radio communications systems, for example based on LTE and/or WiMAX. In fact, in such a scenario, frequency twist according to the present invention can be advantageously applied to the Forward channel from a Base Station to a mobile device, while time twist according to PCT/FR2013/052636 can be advantageously applied to the Return channel from a mobile device to a Base Station.

(292) In conclusion, it is clear that numerous modifications and variants can be made to the present invention, all falling within the scope of the invention, as defined in the appended claims.