WO2019125126A1

WO2019125126A1 - Multi-carrier communication system for doubly selective channels using frequency dispersion and non-linear interference cancellation

Info

Publication number: WO2019125126A1
Application number: PCT/MX2018/000151
Authority: WO
Inventors: Fernando PEÑA CAMPOS; Joaquín CORTEZ GONZÁLEZ; Ramón PARRA MICHEL; José Alberto DEL PUERTO FLORES
Original assignee: Centro De Investigación Y De Estudios Avanzados Del Instituto Politécnico Nacional
Priority date: 2017-12-20
Filing date: 2018-12-14
Publication date: 2019-06-27
Also published as: MX2017016960A

Abstract

The invention relates to a communication system and method for overcoming the distortions and effects introduced by the V2V communication channels. In contrast to any existing invention operating under the same conditions, the described apparatus uses a completely novel receiving technique based on the concept of non-linear detection of signals with frequency dispersion. The receiver of this apparatus can achieve a level of performance such that it efficiently exploits the diversity in frequency, together with the band structure of the equivalent channel matrix. The performance achieved in terms of noise immunity is much better than any technique that has been found up to now and also requires a much smaller number of calculations in the receiver.

Description

Multi-carrier communication system for doubly selective channels using frequency dispersion and non-linear interference cancellation.

Field of the invention.

The present invention is related to the field of telecommunications, specifically to the implementation of a muiti-carrier communication system using orthogonal frequency modulation (OFDM: orthogonal frequency multiplexing division) that allows establishing broadband wireless links in environments with high mobilities such as vehicle-to-vehicle connections (V2V: vehicle ío vehicie).

BACKGROUND OF THE INVENTION

The development of V2V wireless communications links has had a great boom in recent years, this due to its main applications in terms of control and road safety as it is: avoid congestion of main roads, crash prevention, autonomous vehicles, remote tracking of vehicles, etc. Different campaigns of measurements and sounding of the channel in V2V environments verify the existence of high frequencies of Doppier dispersion (greater than 600 Hz) and non-stationary statistics of the channel, which causes that the interference between subporíadoras (ÍCI: iníercarrier interference) is one of the main problems that affect the performance of the various stages in the receiver.

The 802.11p standard retains a frame structure equal to that of the 802.11a / g / n standard, that is, it was not specifically designed to support the V2V channel conditions, which causes the received signal to experience high levels of ICI that eventually impact significantly reducing the performance of the system.

The specific problem of estimating channel parameters in the receiver is complicated in V2V systems since the ICI also affects the pilot required to perform this task. Among the inventions related to attacking this problem are EP20100450186 and EP2363985A1, which propose a receiver with iterative channel estimator based on an expansion model with bases of! channel (BEM: basis expansion modei) in two dimensions. A problem presented by said invention is e! that can only reconstruct the variations of cane! to OFDM block resolution and uses a data detection scheme that does not include the ICI. In addition, it requires a large number of iterations to converge to its best performance in terms of BER.

The inventions in US20100098146 and US9621389 present systems with data estimators that include the iCI and offer better performance than conventional receivers. However, they strictly require the modification of! standard 802.1 1 p to be able to introduce extra training sequences, which decreases the spectral efficiency of! system and represents a compatibility problem. A) Yes same, they use cane estimators! with an observation window that covers a large number of GFDM symbols, which greatly impacted the memory required for its implementation, as well as the system's presence.

Given that the data detector is the most complex operation of the receiver in computational terms, the practical feasibility of any system that intends to operate in V2V channels depends mainly on the computational complexity of this stage. The maximum likelihood detector (MI) in the invention EP0521744A3 has the best performance, but its complexity increases abruptly with the constellation size and the number of subcarriers of data, so its implementation in a system that operates In real time it is not viable. It is also possible to adapt multi-input and multi-output systems (MIMO: multiple input multiple output) techniques such as cancellation of the decision ordered with feedback (DFE: decision feedback equalizer) in US20120121045 and EP2234361A1, to reduce the IC! in OFDM systems, but its performance has been poor. A series of detectors have been proposed, with less complexity than MI, but with a better performance than the DFE. Some of these algorithms were designed for V-BLAST (Vertical-Bel! Laboratories Layered Space-Tlme) EP1587223A1, EP1587223B1. Others are general algorithms close to ML, such as the spherical detector in the invention US8117522,

Recently, tree search algorithms have been applied to detection in multi-carrier systems as reported in US8428159; these have allowed to have a performance close to the ML detection with a reduced complexity in comparison with the spherical detector. One can also find inventions such as W02008151337A1 and EP2QG3833A1 with detectors based on the M-aigorithm combined with the GR decomposition of the channel matrix that exploit »the band structure of the channel matrix.

An important factor that impacts the performance of multi-carrier systems is frequency selectivity. This characteristic of the broadband channel makes it difficult to recover some symbols even under conditions of high signal-to-noise ratio (SNR: signa! Io noise ratio). The aforementioned inventions attack this problem in the coding stages of cana! with the additional cost of loss of spectral efficiency. However, at the signal level, no schemes have been incorporated to exploit the frequency selectivity in the form of diversity to obtain better performance of the system without significantly impacting the spectral efficiency of the system.

Brief description of the figures.

Figure 1 shows a diagram generated! of a vehicle-to-vehicle communication link.

Figure 2 shows the structure of the transmitter! communication system proposed for V2V communication incorporating a frequency dispersion scheme. Figure 3 shows the structure of! receiver of the low complexity communications system proposed for V2V communication using frequency dispersion and non-linear interference cancellation.

Figure 4 shows a diagram exemplifying the search tree used by the algorithm M to perform the detection of the data.

Figure 5 shows the performance of a V2V link using the system proposed in this invention compared to the performance obtained using other reception techniques for existing V2V systems.

Detailed description of the invention,

The characteristic details of the muiti-carrier communication system for V2V communication links using frequency dispersion and non-linear interference cancellation are clearly shown in the following description and in the accompanying illustrative drawings, the same reference signs serving to indicate the same. parts.

Figure 1 shows the vehicle-to-vehicle transmitter (101) that incorporates data modulation along with digital baseband processing and analog conversion for transmitting the signal on the transmitting antenna (103), the signal The carrier travels through the V2V channel Wirelessly by electromagnetic waves to the receiving antenna (104). In the V2V receiver (102) the signal is demodulated and the analog / digital processing is performed to detect the sent data.

Be _a GFDM symbol sent by the V2V transmitter (101) according to the 802.1 1 p standard with length

samples, where N and ^ ^g are the number of subcarriers and the length of the cyclic prefix (CP: cyclic prefix), respectively. On the receiver side, (Rx) (102) assuming that the analog demodulation and subsequent conversion to low-pass complex representation has been performed, an OFDM block of received signal having omitted the CP is given by the expression:

where ^x '- ⁿ ^ is the transmitted OFDM block,

is the received signal,

is the answer of the cane! in the "-th", for an impulse as input into the ^ -th previous sample; ^ ' ^N denotes the operation of modulus N, it is the additive circular Gaussian white noise and symmetric, with zero mean and variance cr, -

circular convocation between the response to! channel impulse (R! G) and ^x M can be rewritten in terms of matrices and vectors such as:

and ^~ Hx + w

(II) where y = [tfoj [íj · · · [N -i] f

w = [w [0] w [l] ·· · w [N - l]] ^r ,

(V)

H is a dimension matrix

^x ^ composed of the elements of the RiC in the form:

Using the Fourier transform, the OFDM symbol received in the frequency domain (DF) is calculated as;

u = Gs -fz, (VI!) where is the discrete Fourier transform (TDF) of the noise vector. is the vector of elements transmitted in the composite frequency domain!

N data symbols specified by the vector ^¾, N '' ^p _t pilot symbols and N ^"G symbols guard ^G -.. ^FHF is the matrix channel response in the frequency domain (MFC) When channel response is invariant in the period of duration of an OFDM symbol, ^G takes the form of a diagonal matrix involving an ICI free system for which data defection is simple, however, in V2V environments due to the high mobility both of the transmitter and the receiver, the Doppler spread is significant, causing the matrix ^{G to} have energy in the components outside the main diagonal, causing ICI.

The receiver described in the following section of this invention combines the time-variant channel estimator iteratively coupled to the non-linear detection of data and a precoding scheme that allows obtaining ICI-free pilots in a single iteration, exploitation of the channel diversity and rapid convergence in the data detection stage.

Proposed reception method,

Channel estimation method.

The small number of pilots in a single OFDM symbol and the fact that they experience ICI make it difficult to estimate C! R variant over time. To counteract this, the algorithm proposed in [F. Pena-Campos, R. Carrasco-Alvarez, O, Longoria-Gandara, and R. Parra-Michel, "Estimation of fast tlme-varying channels in OFDM Systems using two-dimensional prolate," IEEE Trans. Wireless Commun ,, vol. 12, no. 2, pp. 898-907, 2013.] where the receiver's observation model is extended to include adjacent OFDM symbols in e! Channel estimation process in the following way:

(VI! I) where the super index ^ refers to! OFDM symbol from which we want to estimate the response of the channel. To integrate the characteristics of the channel dispersion into the model and reduce the parameters in the estimator, the channel is decomposed using BEM in the form:

where are the coefficients of the

{W $ _r ^t \ Lΐ b \ Vr Mg0 - M _ i] n ⁱ _are _as functions expand the time domain and delay time respectively. The modeling error for this representation in subspaces is concentrated in the term ^e ^ ^. Given that in V2V scenarios, the Doppler and delay dispersion presents statistics within a very diverse set. The discrete spheroidal prolate sequences (DPSS: discrete prolate spheroidai sequences) are used as base functions since they optimally concentrate energy in a finite window of time and bandwidth are used as base functions.

To determine the number of functions required in each of the CIR domains, the approximation is used:

P M M

where the rounding operator denotes upwards, ^{J r} and ^{1 D} are the number of functions used

F

for the expansion of the time domain of the delay and the time domain respectively. is the width

F

band system. is the maximum dispersion of! delay time and ^D is the maximum frequency of Doppler scattering.

TO! replace e! channel by its BEM in (VIII) you get the expression:

The information of the BEM in the Doppler frequency and frequency domains is compactly found in the dually indexed matrix:

ü! l) with

and the coefficients of the BEM given by the expression

By annexing this representation of the channel in equation XII and considering only the positions

are the transmitted and received pilots (denoted by the set P) you get:

where:

The subscript ^p denotes the sub-sampling of the vectors and matrices in the rows and columns corresponding to the position of the pilots. For reasons of simplification in the notation we used the index variable i = q + M _r) (r-í \ 0 £ r £ M _T 0 £ q £ M _n - ~ la

¾ ^D! , where ^r and ^{¾ ΰ} ^ is the vector that concentrates noise contributions, modeling error and intersymbolic interference in order to simplify expressions.

Counting on information known a prior! For the receiver of the matrix L and the vector received ^, the calculation of the vector coefficients of the estimated BEM of the channel is made using the least squares algorithm by means of the expression:

p = (A ^ A) 'Au

(XXI)

Once these coefficients are obtained, any of the representations such as the impulse response and the channel transfer function can be calculated directly by performing the weighted sum of the functions of the base. In this way the estimated MFC can be calculated using the expression:

Dispersion DF T.

The selectivity of the cane! V2V renders detection errors susceptible to OFDM systems because the local power of some subcarriers may be low due to deep fading, which makes it impossible to detect the transmitted data. In order to counteract this problem, the frequency spreading pre-coding technique (DFTS: direct Foulrer transform spreading) is used in the present invention since it uniformly distributes the energy of each symbol throughout the bandwidth and can be calculated with low complexity by means of the fast Fourier transform (FFT; fast Fourier transform).

Said operation can be formally represented on the trasmlsor side in the form:

^S D ^{= F} D (XXIII)

That is, the elements transmitted in frequency in the positions of the data are constructed by applying to the data symbols in the vector ^ the Fourier matrix whose elements are determined by:

(I1UN

for d, ⁷ d '= {'^'UJ . In this way the expression for the signal in the receiver at the positions of the modifies as follows:

O z

where ^D and ^D are the vector of the received signal and the noise vector respectively, each in the

G

position of the data subcarriers. The matrix ^D is obtained by taking the columns and rows of t ^G _" at the positions of the data carriers The term ^ ¾ ^pSp represents the interference in the carriers with data from the carriers with pilots.

Method for data detection.

The main point of innovation in the present invention is the efficient integration of DFTS precoding into e! process of the detection of the data in the receiver. One of the problems of using DFTS in the receiver is reflected in the difficulty to apply non-linear detection algorithms while maintaining low computational complexity since the channel equivalent matrix

It does not conserve structure in band, A solution to this disadvantage is to find an operator such that, when applied to the received signal, it reestablishes the band structure of the equivalent channel matrix. In the case of the present invention the inverse Fourier transform ^D tai is used that the Cramer-Loéve operator in the channel matrix is completed. In mathematical terms, the vector received at the position of the data is obtained as:

V:

where ¾¾

and is the equivalent matrix of channel after applying inverse Fourier transform to the symbols received in the position of the data with linear precoding. For reasons of simplicity in the notation the noise vector ^D is maintained with the same nomenclature given that the transformations orthonormal do not affect its statistics. Note

G w that the correlation characteristics and quasi-band structure of matrix ^D imply that the K matrix also maintains structure in band which allows coupling with efficient execution detection algorithms.

Criteria for detection of maximum likelihood.

Starting from equation (XXVII), the detection of data using the maximum likelihood criterion can be obtained by finding the values that optimize the following expression;

where is the set with all the possible combinations of transmitted symbols and% is the vector with the estimated data symbols. This method of data detection is of great computational complexity due to the exhaustive calculation of all Euclidean distances. The following describes two different methods of data detection that exploit the particular structure of the K matrix in the form of a band to obtain the defection of data in a suboptimal way with reduced computational complexity.

Detection not line! of low complexity.

The non-line detection method! of low complexity proposed in the present invention consists of two main steps:

1. Apply the QR decomposition of the pseudoinverse matrix of K; This process can be performed under the criterion of least squares (LS: least squares) or the least squared error (MMSE, minimum squared error). 2. Perform the non-linear detection process: for this stage, three different methods are proposed in this invention: the QR-ML detection, the QR-ML subopic detection (NML: near ML) and the successive cancellation of ordered interference (QSIC: ordered successíve interference cancellation),

Decomposition QR ordered.

The QR decomposition is used to obtain the matrices Q and R from the channel matrix with precoding K such that the relation is fulfilled:

KP = QR (XXIX) where Q is an orthonormal matrix! that meets the property

= ⁵ and R is a superior triangular matrix. P is a permutation matrix with reordering as a function of the signal to interference ratio. This decomposition can be carried out using different methods, however, in this invention a variant of the Gram-Schmidt algorithm is used. At each step of the orthonormalization of Q, permutations are made in the position of the columns with the objective that the lines resulting in matrix R are arranged according to their signal-to-noise ratio. The quasibanda structure of the K-matrix is exploited to reduce the computational complexity to the maximum in the calculation of the QR decomposition.

To perform said decomposition according to the criterion of LS, we start from an extended matrix:

K = [K | v] _{/ YYY} .

To which a sequence of unit rotations, also called Givens rotations, are applied. In the case of using the MMSE criterion, the extended matrix is constructed in such a way that it includes noise statistics in the form:

rxxxn

The advantage of using unit rotations in the orthogonalization process for obtaining the QR decomposition is that it conserves the original energy of all the elements of the original matrix K maintaining the dynamic range of all the variables used in the process. This feature facilitates the implementation of this method in devices for real-time execution using fixed-point arithmetic.

The sequence process! in the execution of Givens rotations part of the matrix:

(XXXII) in iteration zero, the rotation applied in the ith iteration can be expressed in terms of a transformation matrix ^j which is calculated from ta! so that an element of the matrix is canceled, N "

calculated in the previous iteration. After ^ύ rotations you get that for the criterion case

LS:

In the case of using the MMSE criterion, the resulting matrix after

Ions can be expressed as:

The complete method is described in detail of the sequence of steps below:

1. Take the quasibanda K matrix, of which only the main diagonal and adjacent diagonals are conserved, the rest are truncated to zero value.

2. The extended matrix is constructed according to the criterion to be used, either LS or MMSE.

3, The matrix is initialized

, in which the permutations used for ordering the

data elements. 4, The matrix is initialized

TO!

5. Start the vector ^ with the Euclidean norms of the ^JD columns of the matrix

6. For ^{J ~ 1} the column of the matrix '- which has the smallest norm is determined and its index is stored as ^{~~ l} .

7. The columns J and - ^{~ l} are exchanged in e! vector by the matrix P.

· M- _' V + / - 1 X.

8. Columns J and

in first ios ¹ ^"D J ^{J ~ l} lines for the case of the MMSE criterion or the first M = N ^D lines in the case of criterion LS.

0

9. The set of rotation matrices of Givens ^J tai that is fulfilled is calculated

matrices has been used to refer to the elements of X contained in the rows within the range ja, c] and the columns within the range [b, d], 10. It applies e! Set of rotation matrices of Givens ^J to the elements of the remaining columns

Neither

11. The last ones are updated ^{; DJ} vector values

12. J ^{^} J ⁺ is increased

, ^{1 D} the process is completed, otherwise it will be returned to Step 6.

13. The matrix R and the vector ^{v are} obtained from the matrix X ^l¾> , thus the matrix of permutations P is also returned.

Note that in the execution of this method, the band structure of the K matrix implies that one can do without the calculations relating to the large number of elements that are equal to zero. OSiC data detection.

As mentioned earlier in this document, the QR decomposition serves as the preprocessing stage in data detection. An effective method to subsequently perform the interference cancellation is the OSiC algorithm which, in combination with the QR decomposition described above, allows the suboptimal detection of data with very low computational complexity. The performance of these two techniques together in V2V systems with DFTS gives very low erroneous bit rates. The QR decomposition of - provides an upper triangular matrix R, an orthogonal matrix of unitary norm Q as well as a. permutation vector P, such that

. Ai replace this decomposition in equation (XXVII), you get:

First, to distinguish both sides of this equation by Q, we obtain a system of equations in the form:

where ^

is the vector with the elements of Information ordered according to the signal to noise / interference ratio estimated in the QR decomposition. The noise vector " ^D keeps its statistics,

Due to the triangular structure of the matrix R the elements of! vector ^v can be expressed individually as:

where! to notation

they indicate the ^ -th element of the vector and the b - th element of the ^a - th row of a matrix according to e! case. In this way the detection of each of the data can be done in an iterative manner using the following expression;

where the operator

is used to specify quantization towards the closest symbol of the constellation used by the transmitter W, assuming that in each iteration the previous decisions are correct, the interference of previously detected symbols can be subtracted prior to the detection of the symbol. As a last step in the detection process, the decoded symbols are reordered according to the permutation matrix ^R ,

Using e! preproeesamienío that provides the decomposition QR, the optimal algorithm of ML detection can be adapted to exploit the properties of the matrix R reformulating the equation (XXVIII) as;

The search for the ML solution based on (XL) can be reflected in the construction of a search tree b?] Í t

whose nodes in the * -thmost level correspond to the possible candidates ^{lA '} ' ^{Nb έ + i} 5, in order to calculate the minimum in (XL) the following branch metric is defined:

(XLI

(Í

where is the value of the branch metric of a node ^{LA: Í!} who has a

(e € 1 i + 1 k N

^ "" ^D as its ancestor nodes.The distance between each node of a ^ -th level and the root is defined as the accumulated metric value, which represents the addition of all the metrics of branches from the root to the indicated node. For a given level ⁿ , the accumulated metric is obtained from:

n

According to (XL), the optimal vector es is the one whose path minimizes (XLI!) When

QR-NML detection

The design of the NML detector for the V2V multi-carrier system is based on the incorporation of! algorithm

M to ML detection with conventional QR decomposition. As illustrated in figure 4 the symbol

it is located in the root node of the tree, and the child nodes that emanate from it are a possible solution for UHAflJ ^Í ¾ r - ^U fLy / J] ^A ¾ - -2 '· · · ^> MLAJ2' ML ^Á J ^I ) / _CaC _^ _aa p¡i tio _n M the algorithm is selecting at each level of the tree up to M (for M <P ^\ candidates for the detection ^of? -th estimated symbol vector ¾, discarding P ^{~ ~} M remaining nodes of the current level, Each branch is assigned a distance metric defined by:

selecting the M nodes that maintain a smaller distance between each node of! ^ -th level and the root node, Therefore a! end the defection will have only M possible routes, each with a total distance! equal to:

The solution ^ is given by the path that meets the optimization criteria:

Small values in decrease the number of search nodes in the detector, resulting in a decrease in complexity, but at the cost of decrease in performance in terms of bit error rate, As the value of M is increased, the performance of! proposed algorithm approaches the performance of the optimal ML detector, Iterative method of channel estimation and cancellation of ICi.

Among the two main problems that coherent systems face are: the degradation of the information provided by the pilot symbols due to the selectivity of the channel and the distortion of ICi coming from the data subpopulations and the AWGN noise present in them causing a low performance in the cane estimation stages! and detection of data in the receiver.

In this invention, an iterative method of ICi cancellation is used; the main ideal is to use the ^G- channel matrix estimated in a first iteration to calculate the ICI that contaminates the pilot subcarriers and eliminate it. These pilots with lower ICI are used again to perform channel estimation and data detection. This process can be repeated iteratively for GFDM providing the same symbol in each of the iterations best performance ^'in estimating the cana! and the detection of the data.

Operation of the transmitter system.

The structure of the transmitter is shown in Figure 2, the bits (201) of data that are input to a convolutional encoder (202) in order to add redundancy to the data. Subsequently the encoded data (203) are scattered using an interleaving block (204), then the interleaved data is modulated in phase and quadrature by means of the modulator (206) which delivers symbols belonging to a certain constellation. The modulated symbols (207) are grouped in blocks by means of the serial to parallel converter (208) to construct a block or vector (209) that is processed by the DFTS precoder (210). The precoded data is input to a carrier dispatcher (212) which assigns the elements to data carriers or piiots as the case may be, likewise assigns the guardians in the corresponding indexes. The output of (212) is then modulated in a conventional orthogonal frequency multiplexing scheme (GFDM) with the help of the IFFT (213), after which the cyclic prefix in (215) is appended to it. Finally, the converter block parallel to series takes each of the samples of! GFDM block to be delivered as output from the baseband digital processing stage of the receiver.

Operation of the receiver system.

The structure of the receiver is shown in Figure 3, the received signal (301) is converted into blocks with the help of the serial to parallel converter (302), then the prefix cycle is eliminated in the block (303). Once the extraction of the cyclic prefix has been performed, the received signal is demodulated in a conventional orthogonal frequency multiplexing scheme (GFDM) with the help of the fast Fourier transform block (304). The GFDM block in the frequency domain is introduced to a demapping device (308) for extracting the pilot symbols therefrom. The pilots vector (307) is used to perform the estimation of the channel matrices in the domain of time (309) and in the frequency domain (310). The time channel matrix (309) is truncated into major diagonal bands in truncation block (311). As part of a preprocessing for the detection of data, the interference from the pilot carriers (312) to the OFDM vector in the frequency domain is subtracted, in order to subsequently perform the inverse operation of DFTS in the e! block called IDFT (314). The proposed GRM block (315) performs the GR decomposition of the temporal channel matrix according to the selected method: LS (ZF) or MMSE, then the detection of data symbols is executed in the proposed Near-lvILD detector (316). (317) by the proposed signal model. As final stages in the estimated data vector, deinterlacing (318) and decoding (319) are performed to obtain the vector with the received data.

In order to show the gains obtained with this apparatus operating in a typical V2V link, figure 5 is included with the comparison between the performance of the invention described here (NML tag) using only one iteration against the proposed system in EP20100450186 and ER2383985Ά1 (LMMSE tag) using several iterations. It also shows the maximum performance that a system can achieve in the exact knowledge of the response of! channel (ideal channel label). The evaluation metric is the erroneous bit rate (BER) while the evaluation parameter is the signal-to-noise ratio in the receiver. The proposed invention achieves better performance while requiring a lower number of iterations,

The methods and process diagrams presented in this invention are simply provided as illustrative examples and are not necessarily intended to require or imply that the steps of the various definitions must be performed in the order presented. As will be appreciated by a person skilled in the art, the steps of the various previous definitions can be carried out in any order. Words such as "then", "following", etc., are not intended to limit e! order of the steps; These words are used simply to guide the reader through the description of the methods.

The various illustrative logic blocks, modules, circuits, and logic stages described in connection with the definitions described herein may be supplemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the limitations of the application and / or the particular design imposed by a system in general. The experts can implement the described functionality in various ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

Claims

Claims 1 A multi-carrier communication system for V2V communication links comprising: a. A transmitter structure where the data is dispersed in frequency and then modulated using OFDM, to achieve the above, the following is required: i. The convolutional coding of the bits to be transmitted is carried out. ii. It is done e! interleaving of the coded bits, iii. The resulting bits are assigned to symbols in the complex plane according to the constellation to be used. g IV, Blocks of "D symbols are scattered in frequency using the Fourier transform or the fast Fourier transform in the form: v. The scattered symbols are modulated using QFDM, A receiver structure where the data are detected by applying nonlinear algorithms In the symbols with frequency dispersion to achieve the above, the following is required: i Samples of the received signal are taken at the exit of the anayogic-digitai storage, blocks of samples are formed which have been removed from the sample. part corresponding to the cyclic prefix ili Each block is converted to the frequency domain using the Fourier transform, iv, The carriers are taken with pilots of the current OFDM block, as well as those of the immediately preceding block and a subsequent block to perform the estimation of This process is carried out by means of the calculation of the channel coefficients: and the subsequent reconstruction of the channel matrix in i to frequency performing the operation: · - (A- ^ AV1 A

1. The coefficient estimation matrix of channel ^1} is obtained out of execution, that is, the process is carried out in the design stage and the Inverse matrix is simply stored in the receiver, in such a way that the receiver only requires the execution of a matrix-vector product to obtain estimates of the coefficients of! channel. v The estimated response of the channel is used to eliminate the interference in the data coming from the pilot carriers,

vi The inverse frequency dispersion operation is applied to the data carriers with which a model of the received signal is obtained in the form:

V = ¾ + F "G _P S _P + z _D , vil, The received data symbols are used v the matrix of the equivalent K channel to perform ordered QR decomposition This operation can be executed using any of the following criteria:

1. With the LS criterion the decomposition process on the augmented matrix starts

K = [K | v]

2 With the MMSE criterion the decomposition process on the augmented matrix begins

viii. You use sos received data symbols ^v and the matrix of the cane! K equivalent to perform the QR decomposition using any criteria

ix. The ordered decomposition process can be executed on truncated versions of the matrix

that is to say, matrices only some diagonals with significant energy of the signal have been conserved so that the amount of necessary operations is greatly reduced.

x. The detection of the data is performed using the matrices Q and F resulting from the ordered QR decomposition, as well as e! vector v = Q _"ara vp _to implement this process can be used any of the following variants:

1. Detection using the QSIC method which is based on the iterative calculation of the estimated data using the formula:

2 Detection using the sub-optimal NML method, which is executed using a search tree whose branches are characterized by the metrics

which are used to perform the detection of the data by optimizing the expression:

xi. The detection process can be performed iteratively by re-feeding the data to eliminate the ICI in the pilot carriers and repeat the steps of channel estimation and detection.

2. Other criteria than the examples shown in this document can be used for QR decomposition in the receptor of claim 1, these being considered as variants of the present invention.

3. Other algorithms than the examples shown in this document can be used to calculate the QR decomposition in the receiver of claim 1, these being considered variants of the present invention.

4. Other algorithms than the examples shown in this document for the cancellation of interference can be used starting from the observation model:

these being considered as variants of the present invention.