CN111431607B

CN111431607B - Block matrix interference elimination method in WO-FTN transmission system

Info

Publication number: CN111431607B
Application number: CN202010290749.XA
Authority: CN
Inventors: 曹明华; 张伟; 王惠琴; 康中将; 吕佳芸; 张家玮; 吴照恒; 李文文; 夏皆平
Original assignee: Lanzhou University of Technology
Current assignee: Lanzhou University of Technology
Priority date: 2020-04-14
Filing date: 2020-04-14
Publication date: 2023-01-31
Anticipated expiration: 2040-04-14
Also published as: CN111431607A

Abstract

A method for eliminating interference of block matrix in WO-FTN transmission system includes blocking data signal at sending end, generating data in each sub-block into FTN signal by FTN shaping filter, calculating FTN transmission characteristic coefficient matrix according to shaping filter coefficient, multiplying inverse of said matrix with received data sub-block to reduce interference brought by FTN transmission and raise error code performance of system. Meanwhile, in the execution process of the method, because the forming filters are consistent and the inverses of the FTN transmission characteristic coefficient matrixes are the same, the inversion operation of the matrixes in the whole information transmission process only needs to be calculated once, and therefore the overall calculation complexity of the system can be reduced.

Description

Block matrix interference elimination method in WO-FTN transmission system

Technical Field

The invention relates to a block matrix interference elimination technology suitable for wireless optical super-Nyquist (WO-FTN) rate transmission, which is characterized in that a 4PAM signal is divided into a plurality of sub-blocks and then subjected to super-Nyquist transmission, and matrix operation is adopted in a receiver to reduce intersymbol interference (ISI) of a system, and belongs to the technical field of wireless optical communication.

Background

With the rapid development of the mobile communication industry, the demand of communication networks for high-capacity and high-rate transmission is increasing. Wireless optical communication has the advantages of unlimited frequency spectrum, flexible links, etc., and thus can be used as an alternative measure for solving the high bandwidth requirement. But the link state of the wireless optical communication system is susceptible to weather, micro-particles and turbulence, which causes the transmission rate of the system to be reduced. In order to solve the problem, researchers have proposed methods such as a high-order modulation technique, a wavelength division multiplexing technique, and a Faster-than-Nyquist (FTN) rate transmission technique to compensate for the defects of the link. The FTN technique is a novel non-orthogonal transmission technique, and can improve the transmission rate of the system without increasing the bandwidth of the original system.

In recent years, with the rapid development of digital signal processing technology and the pursuit of high-speed transmission, researchers have made a great deal of research into FTN technology and have achieved abundant results. For example, rusek provides a receiver based on continuous interference cancellation for transmission of two-dimensional FTN signals, which effectively improves the error code performance of the system. Prlja studies the effect of Tubor equalization on FTN signals in a receiver, and finds that an FTN system subjected to Turbo equalization has better error code performance. Jana et al propose to replace the complex equalization algorithm in the receiver with a pre-equalization technique in the FTN system that enables the system to achieve optimal error performance while providing high spectral efficiency. Meanwhile, researchers have made a lot of research on the problem of high complexity of FTN receivers. In order to overcome the problem of high complexity of an MMSE equalizer in a double-selection channel, a scholarly proposes a low-complexity receiver based on two variational methods, so that the system error performance is kept similar to that of the MMSE equalizer, and meanwhile, the low-complexity receiver has low computational complexity. 2016, researchers propose a low-complexity Turbo detection scheme based on an FG-SS-BP equalization method, which is close to an optimal detector under an ISI free condition and has very low complexity. As the wireless optical communication has the advantages of strong confidentiality, simple laying, low cost and the like, the Shanghai Compound denier university builds an indoor FTN wireless optical communication system based on CAP modulation, and proves that the FTN technology can be applied to optical communication.

For ISI problem in FTN systems, most researchers reduce ISI due to FTN transmission by designing an equalizer in a receiver or performing FTN precoding in a transmitter. Meanwhile, color noise occurs due to the effect of matched filtering in the FTN receiver. There are two ways to process color noise, one is to cascade a whitening filter after a matched filter, and the other is to redesign the matched filter by using the idea of orthogonal basis decomposition. The above approaches all add complexity and cost to the system. In addition, with the rapid development of mobile communication technology, the demand for spectrum resources is increasing. The wireless optical communication has the characteristics of unlimited frequency spectrum, flexible link and the like, so that the FTN technology is an effective measure for solving the current shortage of frequency spectrum resources when being applied to a wireless optical communication system. Based on the method, the FTN technology is introduced into a wireless optical communication system, and a low-complexity block-partitioning faster-than-Nyquist transmission method based on matrix operation is provided. The method has important research significance and application value for improving the error code performance of the wireless optical FTN communication system and reducing the complexity of the system.

Disclosure of Invention

The invention aims to provide a block matrix interference elimination method in a WO-FTN transmission system.

The invention relates to a method for eliminating blocking matrix interference in a WO-FTN transmission system.A transmitting terminal divides modulated information into a plurality of subblocks, and then calculates an FTN transmission characteristic coefficient matrix according to an acceleration factor and a forming filter coefficient; at a receiving end, multiplying the received data subblocks by the inverse of a transmission characteristic coefficient matrix to reduce the interference problem caused by FTN technology transmission; meanwhile, the order of a data transmission matrix is reduced, and the calculation complexity of the system can be reduced; the method comprises the following specific steps:

step 1: at a transmitting end, firstly mapping a binary bit stream into a 4PAM signal, and dividing the binary bit stream into n/k subblocks;

step 2: the ith sub-block P _i Inputting an FTN pulse shaping filter to form an FTN signal;

wherein n represents the total number of the 4PAM signals, and k represents the number of the 4PAM signals of each sub-block; j represents the jth symbol in the ith sub-block; τ = sps1/sps2 denotes an acceleration factor, sps1 is an upsampling factor, and sps2 is a downsampling factor; . q (t) represents a normalized raised cosine waveform;

and step 3: sampling the output signal of the photoelectric detector at a receiving end; assume that the sample value of the received signal is Y _i ＝η×G×P _i X H + Z; wherein eta represents the photoelectric conversion efficiency, and G is an FTN transmission characteristic coefficient matrix; h is a Gamma-Gamma channel fading coefficient matrix, and Z represents a Gaussian white noise matrix; left multiplying the ith data sub-block by G ^-1 Thus, the reduction of the intersymbol interference of the system is realized, namely:

wherein g is a filter coefficient of the current code element influenced by the previous code element and the next code element;

and 3, step 3: to Y' _i Carrying out maximum likelihood detection, and recovering the original information through demapping;

and 4, step 4: and (5) circulating the steps 2 and 3 until i = n/k.

The invention has the advantages that: the complexity of matrix operation is reduced by blocking the transmitted data, and a higher error code performance is obtained by combining a matrix elimination algorithm and a maximum likelihood detection algorithm in a receiver. Therefore, the method is particularly suitable for occasions with high operation complexity in large-capacity data transmission.

Drawings

Fig. 1 is a block diagram of an FTN wireless optical communication system under a turbulent flow channel, fig. 2 is a flowchart of an algorithm, fig. 3 shows a relationship curve between system error code performance and signal-to-noise ratio under different methods when a filter roll factor is 0.7 and an acceleration factor is 0.8, fig. 4 shows a relationship curve between system error code performance and signal-to-noise ratio under different methods when a filter roll factor is 0.5 and an acceleration factor is 0.8.

Detailed Description

The invention provides a block matrix interference elimination method in a wireless optical super-Nyquist (WO-FTN) rate transmission system, which can enable the FTN communication system to obtain higher error code performance and lower calculation complexity. The present invention will be described in detail below with reference to specific embodiments thereof.

The invention is achieved by the following technical measures:

1. the basic assumption is that:

the invention adopts 4PAM modulation, assumes the channel as Gamma-Gamma channel, and the channel state is known. Assuming that the background light has been filtered out by the filter, only additive white gaussian noise is considered. This assumption is typical of such systems and is not a particular requirement of the present invention.

2. The method comprises the following specific implementation steps:

the original user information is represented as follows after Gray coding and 4PAM mapping:

A＝[a ₁ a ₂ …a _n ] ^T (1)

a in formula (1) _n ∈A,A＝{a _r =2R-1-R, R =1,2, …, R } is a modulation symbol, and R is a modulation order. The modulated signal a is block processed to obtain the following form:

P＝{P ₁ …P _i …P _n/k },p _i ＝[a _i1 a _i2 …a _ik ] ^T (k＝1,2,…n/k) (2)

in the formula (2), the modulation symbol is divided into n ^ andk sub-blocks, each sub-block having k symbols. For sub-block P _i Performing FTN pulse forming to form a transmitting signal x _i (t), expressed as:

in equation (3), i represents the ith sub-block, and j represents the jth symbol in the ith sub-block. τ denotes an acceleration factor and q (t) denotes a normalized raised cosine waveform, i.e.

In theory the FTN shaping pulse is infinitely long and therefore the ISI introduced is infinitely long. The truncation of the raised cosine filter may be performed in practice. The invention assumes that the number of shaping points of the raised cosine filter is 61, and the expression form of the coefficient matrix is as follows:

Q＝[q ₁ q ₂ …q ₆₁ ] (4)

knowing q ₃₁ =1, according to equation (3), when τ =1, the value of the present symbol is the maximum value of the waveform. When τ < 1, interference occurs between symbols due to the closer superposition. It is assumed herein that ISI has a crosstalk length of 2, i.e. each sample point is affected by only the first and last symbols, where the first and last symbols are affected by only one symbol.

In the system, the FTN pulse forming comprises two processes of upsampling and filtering forming, wherein the upsampling factor is sps1, and the sampling factor of the filter is sps2, so that the downsampling factor is kept unchanged, and the transmission rate is changed by changing the size of the upsampling factor. I.e., τ = sps1/sps2. Depending on the characteristics of the shaping filter, the sample value of the current symbol can therefore be expressed as:

wherein G represents FTN transmission characteristic coefficient matrix, and G = q _sps2-sps1+1 The filter coefficients of the current symbol affected by the preceding and following symbols.

And the signal s (t) formed by the FTN is received and converted into an electric signal by a photoelectric detector after passing through an atmospheric channel, and the signal is sampled at the same time. Assume that the sampled signal is:

Y _i ＝η×G×P _i .×H+Z (6)

in the formula (6), H represents a Gamma-Gamma channel fading coefficient matrix.

It is obvious from equation (6) that the symbol is not only subject to fading caused by the atmospheric channel but also subject to interference from the preceding and following symbols after FTN transmission, so that the inter-symbol crosstalk caused by FTN transmission is reduced by performing matrix operation on the received signal at the receiving end. The process is as follows: left multiplication G at two ends of pair formula (6) ^-1 Obtaining:

G ^-1 Y _i ＝η×G ^-1 G×P _i .×H+G ^-1 ×Z (7)

that is, the signal subjected to matrix operation is represented as Y' _i ＝P _i .×H+G ^-1 Z, as can be seen from equation (7), the first term on the right side is the fading that the symbol is subjected to due to the atmospheric channel, and the second term is the noise sample. Since the channel state is known, it is Y' _i Maximum likelihood detection (MLSD) is carried out, and original information is restored through demapping. The maximum likelihood detection criteria are:

wherein, y _m Is Y' _i Of the value of (1) to be detected,

denotes a 2 norm, h _m Which is indicative of the fading coefficient of the channel,

representing the estimated modulated symbols.

And repeating the steps until the information transmission is finished when i = n/k.

Fig. 2 is a flowchart of the method, in which the originating data processing includes gray coding and 4PAM mapping, then the 4PAM data information is divided into n sub-blocks, and then FTN forming and receiving end signal processing operations are performed on the sub-blocks, and the receiving end signal processing mainly includes four modules of sampling, matrix operation, maximum likelihood detection, and information storage. Finally, judging whether i is larger than n, if i is larger than n, indicating that all sub-blocks have been traversed, and ending the process; otherwise, the flow continues to work.

TABLE 1 computation complexity of blocking matrix interference cancellation method

Table 1 shows the computational complexity of the blocking matrix interference cancellation method. As can be seen from table 1, the blocking transmission method has lower computational complexity in terms of the number of addition operations, the number of multiplication operations, and the complexity of matrix inversion operations compared to the non-blocking transmission method. Therefore, the running time of the algorithm can be reduced, and the working efficiency is improved.

To further illustrate the correctness of the present invention and the influence of the atmospheric turbulence on the system error rate, a Monte Carlo (Monte Carlo) method is adopted to perform simulation verification on the system error rate. The simulation conditions are as follows, (1) the transmitted signal is modulated by 4 PAM; (2) Photoelectric conversion efficiency eta =0.5, wavelength 1.5 μm, and structure constant of refractive index of atmosphere

Is 1 × 10 ^-15 m ^-2/3 The transmission distance is 1km, the acceleration factor is 0.8, and the raised cosine filter roll factors are 0.7 and 0.5.

Fig. 3 and 4 show the system error performance curves of different methods at an acceleration factor of 0.8 under the conditions that the roll-off factor of the raised cosine filter is 0.7 and 0.5, respectively. It can be seen that: when the maximum likelihood detection is only carried out on the signals in the receiver, the error code performance of the system is poor. When the method provided by the invention is adopted under the same condition, the error code performance of the system is improved by about 3 orders of magnitude. Meanwhile, it can be seen that as the raised cosine filter rolls the reduction of the factor, the system error code performance is also reduced.

From the above description of the embodiments, it is clear for a person skilled in the art that the present invention can be implemented in software or by hardware. Based on the above, the contribution of the technical method of the present invention to the prior art can be partially performed by software or hardware.

Claims

1. A block matrix interference elimination method in a WO-FTN transmission system is characterized in that a transmitting terminal divides modulated information into a plurality of sub-blocks, and then an FTN transmission characteristic coefficient matrix is calculated according to an acceleration factor and a shaping filter coefficient; at a receiving end, multiplying the received data subblocks by the inverse of a transmission characteristic coefficient matrix to reduce the interference problem caused by FTN technology transmission; meanwhile, the order of a data transmission matrix is reduced, and the calculation complexity of the system can be reduced; the method comprises the following specific steps:

and 2, step: the ith sub-block P _i Inputting an FTN pulse forming filter to form an FTN signal;

wherein n represents the total number of 4PAM signals, and k represents the number of 4PAM signals of each sub-block; j represents the jth symbol in the ith sub-block; a is _ij Represents the jth modulation symbol in the ith sub-block; τ = sps1/sps2 denotes an acceleration factor, sps1 is an upsampling factor, and sps2 is a downsampling factor; q (t) represents a normalized raised cosine waveform; t represents a symbol period;

and step 3: sampling the output signal of the photoelectric detector at a receiving end; assume that the sample value of the received signal is Y _i ＝η×G×P _i H + Z; wherein eta represents the photoelectric conversion efficiency, and G is an FTN transmission characteristic coefficient matrix; h is a Gamma-Gamma channel fading coefficient matrix, and Z represents a Gaussian white noise matrix; to the ith data sub-block leftRiding G ^-1 Thus, the reduction of the intersymbol interference of the system is realized, namely:

wherein g is a filter coefficient of the current code element influenced by the previous code element and the next code element; for Y _i Carrying out maximum likelihood detection, and then recovering original information through demapping;

and 4, step 4: and (5) circulating the steps 2 and 3 until i = n/k.