CN103152312B - A kind of optical OFDM system clock synchronization system of power adjustable and method - Google Patents

Abstract

The invention claims to protect a time synchronization system and method of an optical OFDM system with an adjustable power partially superimposed training sequence, which relates to the technical field of optical communication. Aiming at the defects of low synchronization performance and large interference to transmission data in the current optical OFDM system, the present invention proposes a new superimposed training sequence time synchronization method, which makes full use of cyclic prefix information to obtain synchronization information. When the cyclic prefix length L is determined, the training sequence is linearly superimposed on the last L sampling points of the OFDM symbol and the cyclic prefix, which overcomes the interference to data when superimposing long sequences. The method has high accuracy rate of time synchronization, little interference to OFDM data, and more flexible selection of power allocation factors of superimposed training sequences.

Description

A power-adjustable optical OFDM system time synchronization system and method

technical field

The invention belongs to the technical field of optical communication, in particular to an optical OFDM system time synchronization method for superimposing training sequences.

technical background

In the past 20 years, OFDM technology has been widely studied in the field of radio frequency, and because of its advantages such as effective use of bandwidth, reduction of interference and resistance to multipath effects, it is widely used in wireless local area networks, digital audio broadcasting, digital video, etc. Broadcasting, fourth-generation mobile communication technology and other fields. The optical OFDM system is a new type of communication system, which combines the dual advantages of optical fiber communication technology and OFDM technology, and utilizes the good robustness of OFDM to optical fiber channel dispersion in optical communication, which has higher spectral efficiency and Technical features such as faster transmission rate.

A typical Intensity Modulation/Direct Detection (IM/DD) single-mode fiber OFDM system consists of a RF OFDM transmitter, an RF-to-optical up-converter, an optical-to-RF down-converter, and a RF OFDM receiver. The transmitter and RF-to-optical up-converter are located at the transmitting end, and the optical-to-RF down-converter and RF OFDM receiver are located at the receiving end.

The advantages of the optical OFDM system are outstanding, but the disadvantages are also obvious. It is quite sensitive to time synchronization. The optical OFDM system requires the receiving end to be able to perform accurate Fast Fourier Transform (FFT) window synchronization. If the start position of the FFT window cannot be accurately determined, the received signal will introduce InterSymbol Interference (ISI) in the FFT link. and intercarrier interference (InterCarrierInterference, ICI), so ensuring the time synchronization accuracy of the optical OFDM system is the basic premise of the reliable transmission of the entire communication system.

At present, people's research on time synchronization methods for OFDM systems mainly focuses on two directions: inserting training sequences and superimposing training sequences. In the article "Anoveltimingsynchronization method for ACO-OFDM-based optical wireless communications" [IEEE Transactions on Wireless Communications, 2008, 7(12): 4958-4967], ShuangTian et al proposed using anti-symmetry to construct training sequences based on the three classic algorithms of Schmidl, Minn and Park. And put forward three feasible time synchronization algorithms suitable for ACO-OFDM system standard, these three methods are time synchronization methods of inserting training sequence. In the article "Super-Imposed Training Scheme for Timing and Frequency Synchronization in OFDM Systems" [IEEETransactionsonBroadcasting, 2007, 53(2): 574-583], Chih-PengLi et al. used the training sequence information contained in the received signal on the basis of superimposing the training sequence, and based on the maximum logarithm A time synchronization method for likelihood function estimation, this method is a time synchronization method for superimposing training sequences.

Although the plug-in training sequence time synchronization algorithm can achieve better synchronization performance, but in this synchronization method, the training sequence monopolizes the transmitter power and frequency band resources, resulting in a decline in the transmitter power efficiency and spectrum utilization of the system, and affects The disadvantage of the transmission rate of the system. The superimposition training sequence time synchronization algorithm superimposes the training sequence required for timing on the data symbol as a whole, which does not monopolize the frequency band resources and does not affect the transmission rate of the system, but the training sequence occupies the power resources originally allocated to the data information, so it will It interferes with the data information and causes the increase of the bit error rate of the system.

How to reduce the interference of the training sequence to data information while making good use of the advantages of superimposing the training sequence without occupying power and frequency band resources is an urgent problem to be solved at present.

Contents of the invention

The technical problem to be solved by the present invention is: in the existing superimposed training sequence time synchronization method, if the training sequence is only superimposed on the OFDM cyclic prefix, because the length of the cyclic prefix is limited, the accuracy of time synchronization is not high; if training The sequence is superimposed on the entire symbol data of the OFDM symbol, because the training sequence occupies part of the energy originally allocated to the OFDM symbol data, so it will cause great interference to the data information. Aiming at the above two defects, the present invention proposes a new superimposed training sequence mode, which uses the correlation between the training sequence information contained in the received signal and the local training sequence, and combines the maximum likelihood estimation method to obtain time synchronization information. Under the premise of determining the cyclic prefix length L and the OFDM symbol length N, the new superposition training sequence mode only linearly superimposes the training sequence part on the last L points of OFDM symbol data to reduce the interference to data information and effectively reduce the system BER. The new synchronization algorithm makes full use of the cyclic prefix information and judges the synchronization by combining the maximum likelihood rule to obtain a high synchronization accuracy rate.

The technical solutions adopted by the present invention to solve the technical problems include: designing a power-adjustable partially superimposed training sequence optical OFDM system time synchronization system, adding a transmitting end superimposed training sequence generating module, receiving The local training sequence generation module at the end and the time synchronization operation module at the receiving end. The superimposed training sequence generation module at the transmitting end generates a superimposed training sequence that marks the time synchronization position; the local training sequence generation module at the receiving end generates the same local training sequence as the superimposing training sequence at the transmitting end for cross-correlation calculation; the time synchronization operation module at the receiving end A cross-correlation operation is performed on the superimposed training sequence and the local training sequence, and the location where the maximum occurs is searched to determine the time synchronization location.

The pseudo-random sequence (PN sequence) in the superposition training sequence generation module at the transmitter is sent to the Hermitian symmetric transformation sub-module for transformation after serial-to-parallel conversion and constellation mapping. The transformed sequence T(k) satisfies: T(k )=[a,T(1),…,T(N/2-1),b,T ^* (N/2-1),…,T ^* (1)], k=0,1,…N -1, and a, b are real numbers, ^* is a conjugate operator. T(k) is output by inverse fast Fourier transform (InverseFourierTransform, IFFT), and after parallel-to-serial transformation, a bipolar real number training sequence t(n) is obtained, and then part of it is linearly superimposed on the OFDM symbol s(n ), the training sequence is not superimposed on the first NL symbols, and then a cyclic prefix is added to form the final transmission signal, wherein the number of subcarriers in the OFDM system is N, and the length of the cyclic prefix is L. The OFDM symbol superimposed with the training sequence is expressed as: $x\left(\mathrm{no}\right)=\left\{\begin{array}{cc}\mathrm{the\; s}\left(\mathrm{no}\right),& \mathrm{no}=\mathrm{0,1},...,N-L-1\\ \sqrt{1-\mathrm{\β}}\mathrm{the\; s}\left(\mathrm{no}\right)+\sqrt{\mathrm{\β}}t(\mathrm{no}+L-N),& \mathrm{no}=N-L,...,N-1\end{array}\right.$ Wherein, β represents a power allocation factor, s(n) is an OFDM symbol before superimposition, and t(n+LN) is a real number training sequence with a length of L.

During the cross-correlation operation, the time synchronization is obtained through the logarithmic approximate function θ, according to the formula:

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; \&Lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}<span\; class="notranslate">\; \}</span>$

$=\arg \underset{\mathrm{\&theta;}}{\max }\left\{\frac{1}{P}\underset{k=\mathrm{\&theta;}}{\overset{\mathrm{\&theta;}+L-1}{\mathrm{\&Sigma;}}}\left(\right|t\left(k\right)\&CenterDot;(r(N+k)+r\left(k\right))\left|\right)\right\}$ to determine the time synchronization position. In the formula, represents the estimated value of θ, Represents the value of the parameter θ when the value of Λ _θ is maximized, P represents the power of one OFDM symbol, used for normalization processing, r(k), r(N+k) are two received signals separated by N points sequence, t(k) is the local training sequence.

The present invention also proposes a power-adjustable partially superimposed training sequence optical OFDM system time synchronization method, which is characterized in that: the superimposed training sequence generation module at the transmitting end generates a superimposed training sequence that marks the time synchronization position, and partially linearly superimposes it on the OFDM The cyclic prefix of the symbol and part of the data information; the local training sequence generation module at the receiving end generates the same local training sequence as the superimposing training sequence at the transmitting end, and is used for cross-correlation calculation; the time synchronization operation module performs superposition training sequence and local training sequence Perform a cross-correlation operation and search for the location where the maximum occurs to determine the time-synchronized location.

Because the intensity modulation direct detection (IM/DD) single-mode fiber OFDM system modulates and sends data through intensity modulation, and the channel can only transmit unipolar real signals, a PN sequence A with excellent autocorrelation is selected for this purpose. After serial-to-parallel conversion and constellation mapping, it is subjected to an inverse fast Fourier transform (IFFT) with the same length as the optical OFDM system, and then parallel-to-serial conversion to become the training sequence B. The length M of the training sequence B is required to be equal to the length L of the OFDM cyclic prefix, or nM=L, and n is an integer greater than or equal to 1.

Time synchronization operation module: by observing the sum of two consecutive OFDM symbol lengths and one OFDM cyclic prefix length of the received signal, that is, 2N+L sampling points. Let the locally generated training sequence and the received signal perform sliding correlation operations, calculate the sum of the correlation operations, and then search for the position where the maximum value is generated. The sliding position when the maximum value is obtained is the time synchronization position.

Beneficial effects of the present invention: the superimposed training sequence mode of the present invention only linearly superimposes the training sequence on the last L points of the OFDM symbol and the L points of the cyclic prefix on the premise of determining the length L of the cyclic prefix and the length N of the OFDM symbol . Because the training sequence information superimposed on the cyclic prefix can be eliminated by removing the cyclic prefix, it will not interfere with the data information. However, all the information of this part of the training sequence can still be used during the correlation operation. It is equivalent to using twice the useful synchronization information, only a single interference is introduced to the data symbols, and the interference is limited to the last L points of the OFDM data symbols, which has little influence on demodulation. This makes the selection of the power allocation factor of the superimposed training sequence very robust, and the system can more flexibly select the appropriate power allocation factor according to the needs. In actual use, it is possible to randomly adjust part of the superimposed training sequence within a certain range. power ratio.

Description of drawings

Figure 1 block diagram of a single-mode fiber OFDM system for direct detection of intensity modulation;

Figure 2 is a block diagram of an optical OFDM system time synchronization system with partially superimposed training sequences with adjustable power;

Fig. 3 is a schematic diagram comparing the partial superposition method of the training sequence with other time synchronization methods;

Figure 4 is a schematic diagram of superimposed power allocation factors and correlation calculation observation windows;

Figure 5 superimposes training sequence power allocation factors and BER performance simulation diagrams.

Detailed ways

Embodiments of the present invention will be specifically described below in conjunction with the accompanying drawings.

Figure 1 is a block diagram of a single-mode fiber OFDM system with intensity modulation direct detection (IM/DD). The IM/DD single-mode fiber OFDM system is mainly composed of four modules: RF OFDM transmitter, RF-to-optical up-converter, optical-to-RF down-converter and RF OFDM receiver. Because radio-frequency devices are more mature than corresponding optical devices, this system places most of the required signal processing in the radio-frequency domain.

Fig. 2 is a block diagram of the optical OFDM system after adding the time synchronization scheme of the present invention to the above system, that is, a block diagram of the time synchronization method of the optical OFDM system with adjustable power and partially superimposed training sequences. On the basis of Figure 1, a superposition training sequence generation module is added at the transmitting end, and a local training sequence generation module and a time synchronization operation module are added at the receiving end.

The superimposed training sequence generation module is used to generate the superimposed training sequence that marks the time synchronization position. This module is placed before the cyclic prefix is added at the transmitter. The partial linear superimposed sub-module combines the superimposed training and the data signal. The local training sequence generation module at the receiving end is used to generate the same local training sequence as the superimposed training sequence at the transmitting end, and is used for cross-correlation calculations. The output of this module is used as one of the two input signals of the time synchronization operation module and is used for Correlation operation is performed on the received signal. The time synchronization calculation module at the receiving end is used to perform correlation calculations and search for the position generated by the maximum value to determine the time synchronization position. The local training sequence and the received signal are the two input signals of this module, and the output of this module is fed back to the system The cyclic prefix removal sub-module of the receiving end is used for time synchronization.

In the superposition training sequence generation module at the transmitting end, select a pseudo-random sequence (PN sequence) A with excellent autocorrelation, perform serial-to-parallel conversion, and then carry out Hermitian symmetric transformation through constellation mapping, and perform length and Inverse fast Fourier transform (IFFT) of the same number of points in the optical OFDM system, and then perform parallel-to-serial conversion to become a training sequence B. The length M of the training sequence B is required to be equal to the length L of an OFDM cyclic prefix, or nM=L, n is An integer greater than or equal to 1. The training sequence is superimposed on the L points of the OFDM cyclic prefix and the last L points of the OFDM data symbol, and the training sequence is not superimposed on the first N-L points of the data symbol, and is superimposed on the L points of the OFDM cyclic prefix and the OFDM data symbol. The training sequences on the last L points are exactly the same.

The sequence used to identify the synchronization position and the sequence used for cross-correlation operations need to be the same as the data to be transmitted. After serial-to-parallel conversion, constellation mapping is performed. The constellation mapping method depends on the needs of the system. BPSK, QPSK, 16QAM, 64QAM, etc. The data after constellation mapping should be sent to the Hermitian symmetric transformation sub-module for symmetrical transformation. The transformed sequence T(k) satisfies: T(k)=[a,T(1),…,T(N/2 -1),b,T ^* (N/2-1),…,T ^* (1)], k=0,1,…N-1, and a,b are real numbers, ^* is a conjugate operator. T(k) is output by IFFT, and after performing parallel-serial transformation, a bipolar real number training sequence t(n) is obtained, and then its part is linearly superimposed on the last L symbols of OFDM symbol s(n), and the first NL The training sequence is not superimposed on the symbols, and then a cyclic prefix is added to form the final transmitted signal. This superposition scheme will be described in detail later in conjunction with FIG. 3 . The OFDM symbol superimposed with the training sequence is expressed as $x\left(\mathrm{no}\right)=\left\{\begin{array}{cc}\mathrm{the\; s}\left(\mathrm{no}\right),& \mathrm{no}=\mathrm{0,1},...,N-L-1\\ \sqrt{1-\mathrm{\&beta;}}\mathrm{the\; s}\left(\mathrm{no}\right)+\sqrt{\mathrm{\&beta;}}t(\mathrm{no}+L-N),& \mathrm{no}=N-L,...,N-1\end{array}\right.,$ In the formula, β represents the power allocation factor, s(n) is the OFDM symbol when not superimposed, and t(n+LN) is a real number training sequence with a length of L. After adding the cyclic prefix CP (cyclic prefix) to the superimposed training sequence OFDM symbols, because the system uses light intensity modulation, clipping NegativeSignal processing must be performed to become a unipolar signal x′(n) suitable for optical fiber channel transmission, which can be expressed as for In the formula Represents the clipping operator, that is, the signal less than zero is set to zero, and the signal greater than or equal to zero remains unchanged; x(N-L+n) represents the OFDM symbol part of the non-superimposed training sequence; x(nL) represents the superimposed training sequence part of the OFDM symbol. Then the digital signal is converted into an analog signal through a digital-to-analog conversion module, and the interference signal is filtered out through a low-pass filter. The Mach-Zehnder modulator (MZM) modulates the signal in the radio frequency domain to the optical domain and sends it to the optical fiber link. .

At the receiving end, the photoelectric detection sub-module first converts the light intensity signal from the optical domain to the electrical domain. After the low-pass filter filters out the interference, the analog-to-digital conversion sub-module samples and processes the analog signal into a digital signal. At this time, the signal is divided into two channels, and one channel of received signal is copied, which is sent to the correlation calculation module and the training sequence generated locally at the receiving end for cross-correlation calculation, and the time synchronization position is determined by searching the position generated by the maximum value. After the time synchronization position is obtained, it is fed back to the cyclic prefix removal module. After the cyclic prefix is removed, the other received signal undergoes serial-to-parallel transformation, fast-forward Fourier transform (FFT), de-Hermitian symmetry, and constellation mapping. And restore the transmitted data after parallel-to-serial conversion.

FIG. 3 is a schematic diagram of a comparison between the training sequence partial superposition method and other time synchronization methods. There are 4 subgraphs in this figure. The dark part in the figure represents the training sequence, and the position of the arrow is the correct time synchronization position. (a) The sub-picture is a schematic diagram of the time synchronization method for inserting the training sequence. It can be seen from the figure that the training sequence enjoys all the transmitting power of the transmitter, and its time synchronization accuracy performance is undoubtedly good, but it occupies the time slot for sending data. Consumes frequency resources and affects data transmission rate. (b), (c) two sub-graphs are two common superimposed training sequence time synchronization methods, (b) sub-graph superimposes the training sequence on a complete OFDM data symbol, because it occupies a part that should be allocated to The power of the data symbol, so it will affect the demodulation accuracy of the data symbol of the superimposed training sequence and increase the bit error rate of the system; (c) the sub-graph only superimposes the training sequence on an OFDM cyclic prefix, because the length of the cyclic prefix is limited , so the time synchronization accuracy rate is not high. (d) The subgraph is the stacking method of the present invention. When the length of the OFDM cyclic prefix is L, the training sequence of length L is first superimposed on the OFDM cyclic prefix, and the training sequence is linearly superimposed on the last L of the data OFDM symbols at the sampling point. Through the comparison of the four pictures, it is easier to understand the partial superposition training sequence method of the present invention.

Fig. 4 is a schematic diagram of superposition power allocation factors and correlation calculation observation windows. There are 2 sub-graphs in this figure. (a) The training sequences of the two dark parts indicated by the dotted line in the sub-figure are exactly the same, which is a copy relationship. β represents the power allocation factor, which is the power ratio of the partially superimposed training sequence. $\mathrm{\&beta;}={\mathrm{\&sigma;}}_t^2/({\mathrm{\&sigma;}}_t^2+{\mathrm{\&sigma;}}_{\mathrm{the\; s}}^2),$ β ∈ [0,1], ${\mathrm{\&sigma;}}_t^2=\mathrm{E.}\left({\left|t\left(\mathrm{no}\right)\right|}^2\right)$ and ${\mathrm{\&sigma;}}_{\mathrm{the\; s}}^2=\mathrm{E.}\left({\left|\mathrm{the\; s}\left(\mathrm{no}\right)\right|}^2\right),$ Right now and Represent the power of the training sequence and data information respectively, t(n) is the real number training sequence, s(n) is the OFDM symbol when not superimposed, and E() is the mathematical expectation operator. The larger the value of β, the larger the transmit power occupied by the superimposed training sequence. During the superposition, it is required that the training sequence superimposed on the L points of the OFDM cyclic prefix and the last L sampling points of the data OFDM symbols are exactly the same. (b) The sub-figure is a schematic diagram of the correlation operation observation window. In the cross-correlation operation, by observing the continuous 2N+L sampling points of the received signal r(n), the reason for selecting 2N+L sampling points is that only when the observation window is greater than or equal to 2N+L, can it be guaranteed that the observation window contains A complete training sequence. Three examples of correct observation windows are marked in the figure, namely, window 1, window 2, and window 3. In the cross-correlation operation, the time synchronization is obtained by defining the logarithmic likelihood function (Log-LikelihoodFunction) θ, and its measurement function is

${\mathrm{<span\; class="notranslate">\; \&Lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; log</span>}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; log</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; |</span>\frac{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; log</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; |</span>\frac{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; )</span>$

In the formula, r=[r(0),r(1),…,r(L-1)] ^T , t=[t(0),t(1),…,t(L-1)] ^T , θ is the relative sliding position between the received signal and the local training sequence, L represents the length of the cyclic prefix, N represents the length of an OFDM symbol, r(k), r(N+k) are two received signals separated by N points sequence, t(k) is the local training sequence, and f() is the probability density operator. Combined with the maximum likelihood estimation method to

$\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\underset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\mathrm{<span\; class="notranslate">\; max</span>}}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; \&Lambda;</span>}}_{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}<span\; class="notranslate">\; \}</span>$

$=\arg \underset{\mathrm{\&theta;}}{\max }\left\{\frac{1}{P}\underset{k=\mathrm{\&theta;}}{\overset{\mathrm{\&theta;}+L-1}{\mathrm{\&Sigma;}}}\left(\right|t\left(k\right)\&CenterDot;(r(N+k)+r\left(k\right))\left|\right)\right\}$ to determine the time synchronization position. In the formula, represents the estimated value of θ, Represents the value of the parameter θ when the value of Λ _θ is maximized, P represents the power of one OFDM symbol, used for normalization processing, r(k), r(N+k) are two received signals separated by N points sequence, t(k) is the local training sequence.

In order to illustrate how the present invention achieves the power adjustable effect, the following will be combined with a simulation description. The specific parameters of the simulation: the transmission data bit stream rate is 10Gbit/s, the modulation method is 16-QAM, the number of system subcarriers N=128, the optical fiber link adopts 1550nm standard single-mode optical fiber, the transmission distance is 10km, and the attenuation constant α=0.2 dB/km, considering the dispersion effect, the value is 16.75ps/nm/km, the sensitivity of the PIN photodetector is set to 1A/W, the dark current is 10nA, and the cyclic prefix is selected to be one-eighth of the OFDM symbol length. The setting of simulation parameters does not affect the generality of the present invention. Fig. 5 is the performance simulation diagram of superposition training sequence power allocation factor and bit error rate (BER). The power allocation factors were chosen to be β=0,0.2,0.4,0.6 for simulation. It can be seen from Figure 5 that as the value of β increases, the bit error rate performance of the system will deteriorate. But it can also be seen from the figure that when the signal-to-noise ratio SNR∈[0,10], the bit error rate performance of β=0.2, 0.4, 0.6 is equivalent to the performance of β=0 without superimposing the training sequence, and the simulation curves almost overlap . Only when SNR=11, 12, the BER performance of β=0.2, 0.4, 0.6 drops slightly, but not obviously. The simulation shows that the value of β of the method of the present invention is very robust. In practical applications, the value of the power allocation factor β can be adjusted according to the synchronization performance requirements of the system time synchronization, and the increase of the value of β can bring about an increase in synchronization performance , but has little impact on system BER performance.

The method for time synchronization of an optical OFDM system with adjustable power partially superimposed training sequences described in the present invention has been described in detail above, and the above specific implementation descriptions can be used to help understand the core idea of the present invention. The present invention makes full use of the CP information and the joint maximum likelihood estimation criterion to perform time synchronization estimation, and the impact of the superimposed training sequence on OFDM data is small, so the selection of the power allocation factor of the superimposed training sequence in the present invention has better robustness, The system can more flexibly select an appropriate power allocation factor according to needs.

Claims (6)

1. a kind of power-adjustable partial superposition training sequence optical OFDM system time synchronization system is characterized in that: increase superposition training sequence generation module at single-mode optical fiber OFDM system transmitting end, increase local training sequence generation module at receiving end, and time Synchronous operation module; the pseudo-random sequence PN in the superimposed training sequence generation module at the transmitting end undergoes serial-to-parallel conversion, constellation mapping, Hermitian symmetric transformation, and the transformed sequence T(k) is output by IFFT, and after parallel-to-serial transformation, Obtain a bipolar real number training sequence t(n), and then linearly superimpose the t(n) part on the last L symbols of the OFDM symbol s(n), and do not superimpose the training sequence on the first NL symbols, and then add a cyclic prefix , to generate a superimposed training sequence that marks the time synchronization position; the local training sequence generation module at the receiving end generates the same local training sequence as the superimposing training sequence at the transmitting end, and is used to generate the local training sequence for cross-correlation operations; the time synchronization operation module at the receiving end performs the superposition training sequence and The local training sequence performs cross-correlation operation, and searches for the position where the maximum value is generated to determine the time synchronization position. During the cross-correlation operation, the time synchronization position is obtained through the logarithmic approximate function θ, according to the formula: $\begin{array}{c}\widehat{\mathrm{\&theta;}}=\arg \underset{\mathrm{\&theta;}}{max}\left\{{\mathrm{\&Lambda;}}_{\mathrm{\&theta;}}\right\}\\ =\arg \underset{\mathrm{\&theta;}}{max}\left\{\frac{1}{P}\underset{k=\mathrm{\&theta;}}{\overset{\mathrm{\&theta;}+L-1}{\&Sigma;}}\left(\right|t\left(k\right)\&CenterDot;\left(r\left(N+k\right)+r\left(k\right)\right)\left|\right)\right\}\end{array}$ Get an estimate of θ that maximizes the value of the time metric function Λ _θ To determine the time synchronization position, where the number of subcarriers in the OFDM system is N, the length of the cyclic prefix is L, P represents the power of one OFDM symbol, r(k), r(N+k) are two receiving points separated by N points Signal sequence, t(k) is the local training sequence. 2. The time synchronization system according to claim 1, wherein the transformed sequence T(k) satisfies: T(k)=[a, T(1),..., T(N/2-1) ,b,T ^* (N/2-1),...,T ^* (1)], where the number of subcarriers in the OFDM system is N, the length of the cyclic prefix is L, k=0,1,...N-1, a, b are real numbers, ^* is a conjugate operator. 3. time synchronization system according to claim 1, is characterized in that: the OFDM symbol behind the superposition real number training sequence is expressed as: $x\left(\mathrm{no}\right)=\left\{\begin{array}{cc}\mathrm{the\; s}\left(\mathrm{no}\right),& \mathrm{no}=0,1,...,N-L-1\\ \sqrt{1-\mathrm{\&beta;}}\mathrm{the\; s}\left(\mathrm{no}\right)+\sqrt{\mathrm{\&beta;}}t(\mathrm{no}+L-N),& \mathrm{no}=N-L,...,N-1\end{array}\right.$ Wherein, β represents a power allocation factor, s(n) is an OFDM symbol without superimposition, and t(n+LN) is a real number training sequence with a length of L. 4. A power-adjustable partial superimposed training sequence optical OFDM system time synchronization method, characterized in that: the pseudo-random sequence PN in the transmitting end superimposed training sequence generation module undergoes serial-to-parallel conversion, constellation mapping, Hermite symmetric transformation, Obtain the transformed sequence T(k) and output it through IFFT, and perform parallel-to-serial transformation to obtain a bipolar real number training sequence t(n), and then partially linearly superimpose it on the last L of OFDM symbols s(n) On the symbol, the training sequence is not superimposed on the first N-L symbols, and then a cyclic prefix is added to generate a superimposed training sequence that marks the time synchronization position, and it is linearly superimposed on the cyclic prefix of the OFDM symbol and part of the data information; the receiver trains locally The sequence generation module generates the same local training sequence as the superposition training sequence at the transmitting end for cross-correlation calculation; the time synchronization operation module performs cross-correlation calculation on the superposition training sequence and the local training sequence, and searches for the position where the maximum value is generated to determine the time Synchronization position, during cross-correlation operation, the time synchronization position is obtained through the logarithmic approximate function θ, according to the formula: $\begin{array}{c}\widehat{\mathrm{\&theta;}}=\arg \underset{\mathrm{\&theta;}}{max}\left\{{\mathrm{\&Lambda;}}_{\mathrm{\&theta;}}\right\}\\ =\arg \underset{\mathrm{\&theta;}}{max}\left\{\frac{1}{P}\underset{k=\mathrm{\&theta;}}{\overset{\mathrm{\&theta;}+L-1}{\&Sigma;}}\left(\right|t\left(k\right)\&Center\; Dot;\left(r\left(N+k\right)+r\left(k\right)\right)\left|\right)\right\}\end{array}$ Get an estimate of θ that maximizes the value of the time metric function Λ _θ To determine the time synchronization position, where the number of subcarriers in the OFDM system is N, the length of the cyclic prefix is L, P represents the power of one OFDM symbol, r(k), r(N+k) are two receiving points separated by N points Signal sequence, t(k) is the local training sequence. 5. time synchronization method according to claim 4, it is characterized in that: described linear superposition is specifically, training sequence is superimposed on L points of OFDM cyclic prefix and rear L points of OFDM symbol, and the front of OFDM symbol No training sequence is superimposed on N-L points. 6. The time synchronization method according to claim 4, the OFDM symbol after superimposing the training sequence is expressed as: $x\left(\mathrm{no}\right)=\left\{\begin{array}{cc}\mathrm{the\; s}\left(\mathrm{no}\right),& \mathrm{no}=0,1,...,N-L-1\\ \sqrt{1-\mathrm{\&beta;}}\mathrm{the\; s}\left(\mathrm{no}\right)+\sqrt{\mathrm{\&beta;}}t(\mathrm{no}+L-N),& \mathrm{no}=N-L,...,N-1\end{array}\right.$ Wherein, β represents a power allocation factor, s(n) is an OFDM symbol without superimposition, and t(n+LN) is a real number training sequence with a length of L.