CN101707579B - continuous phase modulation frequency domain equalization modulation and demodulation method - Google Patents

Abstract

The invention discloses a continuous phase modulation frequency domain equalization modulation and demodulation method which can be used in a frequency selective fading channel. The generation of the continuous phase modulation signal comprises mapping, modulating and inserting a cyclic prefix; demodulation of continuous phase modulated signals involves receiving the signal under a frequency selective fading channel, removing the cyclic prefix, quadrature decomposition, channel estimation, frequency domain channel equalization, viterbi decoding after conversion back to the time domain. The invention eliminates the self-correlation of CPM by an orthogonal decomposition method, and then transforms the CPM to the frequency domain for channel estimation and equalization, thereby greatly reducing the complexity, especially under a frequency selective fading channel. The invention can effectively counteract the influence of the channel by a frequency domain balancing method, eliminates the intersymbol interference caused by the influence of the channel, and can work under a frequency selective fading channel.

Description

Continuous Phase Modulation Frequency Domain Equalization Modulation and Demodulation Method

technical field

The present invention relates to signal modulation and demodulation technology, in particular to a continuous phase modulation frequency domain equalization modulation and demodulation method.

Background technique

The CPM (Continuous Phase Modulation) signal has a good constant envelope characteristic, which makes it unaffected by the high-efficiency nonlinear power amplifier amplification, and the power efficiency is high. At the same time, the CPM modulated signal has the characteristics of a continuous phase, and there is no sudden phase change, so there is no high-frequency component. Therefore, the spectrum of the CPM signal is compact, and its frequency band utilization rate is high. Under the same symbol rate, the bandwidth of CPM signal is smaller than ordinary FSK and PSK modulation. The CPM modulated signal has been widely used because of its good constant envelope characteristics, high power efficiency, and high spectral efficiency.

With the development of communication technology, and the requirements of communication services become more diversified, the requirements for communication bandwidth are getting higher and higher. In wireless channel transmission, broadband signals experience much greater fading than narrowband signals, and are frequency selective channels. Therefore, the selection of anti-fading technology is very important. Under fading channel conditions, the best time-domain equalization technique for CPM is the Maximum Likelihood Sequence Estimation (MLSE) method using the Viterbi algorithm. MLSE has a very high complexity when the channel dispersion length becomes longer, even using The reduced complexity Delayed Decision Feedback Sequence Estimation method (DDFSE) still has a high complexity, so it is very meaningful to seek the frequency domain equalization of CPM.

Contents of the invention

The purpose of the present invention is to overcome the deficiencies of the prior art, and provide a continuous phase modulation frequency domain equalization modulation and demodulation method, which can be used in frequency selective fading channels.

Including the following steps:

1) Obtain the binary sequence from the data source, map the binary sequence into a modulation symbol sequence, form the modulation symbol or training symbol into a symbol block, and send it to the continuous phase modulator block by block according to the length of the symbol block;

2) After each symbol block completes the continuous phase modulation, insert the cyclic prefix waveform to obtain the continuous phase modulation signal;

3) Sending the continuous phase modulation signal on the communication channel, and the continuous phase modulation signal undergoes frequency selective fading;

4) Receive the fading continuous phase modulation signal, remove the cyclic prefix, and obtain the training symbol blocks and data symbol blocks that have experienced fading, and use the orthogonal basis stored locally to positively correct each training symbol block or data symbol block cross decomposition;

5) After the orthogonal decomposition, the training symbol block is transformed into the frequency domain for channel estimation, and the data symbol block is equalized in the frequency domain according to the channel estimation result, and then transformed back to the time domain after removing the influence of channel interference;

6) After equalization, the data symbol block converted back to the time domain is sent to the Viterbi decoder for demodulation to obtain the source sequence.

The steps of obtaining the binary sequence from the data source, mapping the binary sequence into a modulation symbol sequence, forming the modulation symbols or training symbols into symbol blocks, and sending them to the continuous phase modulator block by block according to the length of the symbol block are:

1) Map the binary sequence a _n obtained from the data source to the data symbol sequence I _n , specifically map 1 to +1, and map 0 to -1;

2) every 252 data symbols I _n and 4 return-to-zero symbols form a data symbol block, and the length of the data symbol block is 256;

3) Insert a training character block every 20 data symbol blocks, and form a frame structure with the training character block as the starting point, and the training character block is still composed of a training character symbol with a length of 252 and 4 return-to-zero symbols;

$\left(t\right)={\∫}_{-\∞}^t\left(\mathrm{\τ}\right)\mathrm{d\τ}$ 为成形函数，I_n为数据符号块或训练序列符号块，

h＝0.5。4) Each symbol block enters the continuous phase modulator in turn, and the phase at the starting point of each symbol block is 0, and the phase at the 231st (256-25) symbol of each symbol block is also 0, and the continuous phase modulation signal waveform The expression is in

$\left(t\right)={\∫}_{-\∞}^t\left(\mathrm{\τ}\right)\mathrm{d\τ}$ is a shaping function, _In is a data symbol block or a training sequence symbol block,

h=0.5.

After each symbol block completes the continuous phase modulation, the steps of inserting the cyclic prefix waveform to obtain the continuous phase modulation signal are:

1) The size of the cyclic prefix length should be greater than the maximum dispersion length of the channel, and the maximum dispersion length of the frequency selective fading channel is set to 25 symbols, so the length of the cyclic prefix is set to 25 symbols;

2) Return the phase to zero at the 229th (256-25-2) symbol, which will use up 2 phase-zero symbols;

3) At the end of each symbol block, another 2 return-to-zero symbols need to be added to return the phase to 0 from any state;

4) For each symbol block, copy the CPM signal waveform of the last 25 symbols to the front of the symbol block to form a cyclic prefix.

The steps of sending the continuous phase modulation signal on the communication channel, and the continuous phase modulation signal undergoing frequency selective fading are as follows:

经历频率选择性衰落信号后表示为： $r\left(t\right)=\underset{l=0}{\overset{M_D-1}{\mathrm{\Σ}}}{h}_{l}s(t-\mathrm{lT})+z\left(t\right),$ 其中z(t)是零均值复高斯噪声，方差为N₀；M_D是信道的最大弥散长度，h_l为信道第L径的增益。1) Continuous phase modulation signal

After experiencing the frequency selective fading signal, it is expressed as: $r\left(t\right)=\underset{l=0}{\overset{m_{\mathrm{D.}}-1}{\mathrm{\Σ}}}{h}_l\mathrm{the\; s}(t-\mathrm{lT})+z\left(t\right),$ Where z(t) is zero-mean complex Gaussian noise with variance N ₀ ; M _D is the maximum dispersion length of the channel, and h _l is the gain of the L-th path of the channel.

The continuous phase modulation signal after receiving the fading, removes the cyclic prefix, and obtains the training symbol block and the data symbol block that have experienced fading, and uses the orthogonal base stored locally to do each training symbol block or data symbol block The steps of orthogonal decomposition are:

1) Receive the continuous phase modulation signal r(t) of the frequency selective fading channel, and remove the cyclic prefix for r(t);

使相位网格状态时不变，得到r′(t)；2) plus tilt phase

Make the state of the phase grid time-invariant, and get r'(t);

3) The orthogonal basis stored locally is f ₀ (t)=1, $f_1\left(t\right)=\frac{1}{\sqrt{1-\frac{4}{{\mathrm{\&pi;}}^2}}}(e^{\frac{\mathrm{j\&pi;t}}{T}}-\frac{j2}{\mathrm{\&pi;}}),$ 0≤t<T, where T is the symbol period; correlate the received signal r′(t) with the tilted phase with two orthogonal bases within the symbol period, and obtain two sets of vectors r ₀ (n) and r ₁ (n).

After the orthogonal decomposition, transform the training symbol block into the frequency domain for channel estimation, and perform equalization in the frequency domain on the data symbol block according to the channel estimation result, and then transform back to the time domain after removing the influence of channel interference for:

1) Orthogonal decomposition decomposes the training character block and the data symbol block into two vectors at the same time; extract the vector segments r ₀ ^T (n) and r ₁ ^T (n) of the training character block from the two vectors respectively;

2) Transform r ₀ ^T (n), r ₁ ^T (n) and the locally stored training symbol block vectors s ₀ ^T (n), s ₁ ^T (n) into the frequency domain to obtain R ₀ ^T (k ), R ₁ ^T (k) and S ₀ ^T (k), S ₁ ^T (k);

3) According to the minimum mean square error criterion, the estimated value of the channel frequency domain after orthogonal decomposition is:

${\tilde{h}}_0=\frac{h_0^{*}}{{\left|h_0\right|}^2+N_0},$ ${\tilde{h}}_1=\frac{h_1^{*}}{{\left|h_1\right|}^2+N_0}$ ( $h_0=R_0^T/S_0^T,$ $h_1=R_1^T/S_1^T$ );

$\widetilde{S_0}=\widetilde{H_0}*R_0,$ $\widetilde{S_1}=\widetilde{H_1}*R_1;$ 4) Transform the vectors r ₀ (n) and r ₁ (n) obtained after the orthogonal decomposition of the data symbol block into the frequency domain to obtain R ₀ , R ₁ , and remove the influence of the channel in the frequency domain according to the channel estimation results to obtain Two balanced outputs

$\widetilde{S_0}=\widetilde{h_0}*R_0,$ $\stackrel{~}{{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">S</figure-callout>}}_1}=\stackrel{~}{{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">h</figure-callout>}}_1}*{\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">R</figure-callout>}}_1;$

变换到时域，送进解调器解调。5) Will

Transform to the time domain and send it to the demodulator for demodulation.

The present invention eliminates the correlation of the CPM itself through the method of orthogonal decomposition, and then transforms it into the frequency domain for channel estimation and equalization, which greatly reduces the complexity, especially in the frequency selective fading channel. The invention can effectively counteract the influence of the channel through the frequency domain equalization method, and eliminate the intersymbol interference caused by the channel influence, so it can work in the frequency selective fading channel.

Description of drawings

FIG. 1 is a block diagram of an exemplary communication system according to an embodiment of the present invention;

Fig. 2 is a symbol block structure diagram according to an embodiment of the present invention;

Fig. 3 is a schematic diagram of adding a cyclic prefix according to an embodiment of the present invention;

Fig. 4 is a block diagram according to the CPM signal orthogonal decomposition in the embodiment of the present invention;

Fig. 5 is a block diagram of CPM frequency domain equalization according to an embodiment of the present invention.

Detailed ways

The invention realizes the modulation and demodulation of the CPM signal under the frequency selective fading channel by utilizing frequency domain equalization technology.

Figure 1 shows an exemplary communication system of an embodiment of the present invention. The sending part is connected by data source 101, mapper 102, framing 103, CPM modulator 104, and adding cyclic prefix 105 modules to complete the generation process of CPM signal; The modulation process consists of cyclic prefix removal 107, orthogonal decomposition 108, channel estimation 110, frequency domain equalization 109 and demodulation 111 modules, and finally demodulates the binary stream and restores the data source.

The continuous phase modulation frequency domain equalization modulation and demodulation method includes the following steps:

1) Obtain the binary sequence from the data source, map the binary sequence into a modulation symbol sequence, form the modulation symbol or training symbol into a symbol block, and send it to the continuous phase modulator block by block according to the length of the symbol block;

2) After each symbol block completes the continuous phase modulation, insert the cyclic prefix waveform to obtain the continuous phase modulation signal;

3) Sending the continuous phase modulation signal on the communication channel, and the continuous phase modulation signal undergoes frequency selective fading;

4) Receive the fading continuous phase modulation signal, remove the cyclic prefix, and obtain the training symbol blocks and data symbol blocks that have experienced fading, and use the orthogonal basis stored locally to positively correct each training symbol block or data symbol block cross decomposition;

5) After the orthogonal decomposition, the training symbol block is transformed into the frequency domain for channel estimation, and the data symbol block is equalized in the frequency domain according to the channel estimation result, and then transformed back to the time domain after removing the influence of channel interference;

6) After equalization, the data symbol block converted back to the time domain is sent to the Viterbi decoder for demodulation to obtain the source sequence.

The steps of obtaining the binary sequence from the data source, mapping the binary sequence into a modulation symbol sequence, forming the modulation symbols or training symbols into symbol blocks, and sending them to the continuous phase modulator block by block according to the length of the symbol block are:

1) Map the binary sequence a _n obtained from the data source to the data symbol sequence I _n , specifically map 1 to +1, and map 0 to -1;

2) every 252 data symbols I _n and 4 return-to-zero symbols form a data symbol block, and the length of the data symbol block is 256, as shown in Figure 2, as can be seen, in a symbol block, the number of data symbols 252, divided into two sections, 229 (201) in the first section, 23 (203) in the second section, and a group of zeroing symbols (202, 204) in the middle and at the end respectively;

3) Insert a training character block every 20 data symbol blocks, and form a frame structure with the training character block as the starting point, and the training character block is still composed of a training character symbol with a length of 252 and 4 return-to-zero symbols;

$q\left(t\right)={\&Integral;}_{-\&infin;}^{t}g\left(\mathrm{\&tau;}\right)\mathrm{d\&tau;}$ 为成形函数，I_n为数据符号块或训练序列符号块，h＝0.5。4) Each symbol block enters the continuous phase modulator in turn, and the phase at the starting point of each symbol block is 0, and the phase at the 231st (256-25) symbol of each symbol block is also 0, and the continuous phase modulation signal waveform The expression is in

$q\left(t\right)={\&Integral;}_{-\&infin;}^{t}g\left(\mathrm{\&tau;}\right)\mathrm{d\&tau;}$ is a shaping function, _In is a data symbol block or a training sequence symbol block, h=0.5.

Fig. 3 illustrates the way of adding the cyclic prefix in the embodiment of the present invention, since the CPM signal is modulated in phase, it is necessary to add the cyclic prefix after the CPM modulation, specifically copying the modules 304 and 305 in Fig. 3 to form 301 get the cyclic prefix. In order to ensure the continuity of the phase of the CPM signal, a return-to-zero symbol needs to be added to return the phase state to zero, see 303 and 305 in FIG. 3 .

After each symbol block completes the continuous phase modulation, the steps of inserting the cyclic prefix waveform to obtain the continuous phase modulation signal are:

1) The size of the cyclic prefix length should be greater than the maximum dispersion length of the channel, and the maximum dispersion length of the frequency selective fading channel is set to 25 symbols, so the length of the cyclic prefix is set to 25 symbols;

2) Return the phase to zero at the 229th (256-25-2) symbol, which will use up 2 phase-zero symbols;

3) At the end of each symbol block, another 2 return-to-zero symbols need to be added to return the phase to 0 from any state;

4) For each symbol block, copy the continuous phase modulation signal waveform of the last 25 symbols to the front of the symbol block to form a cyclic prefix, see 301 in FIG. 3 .

The steps of sending the continuous phase modulation signal on the communication channel, and the continuous phase modulation signal undergoing frequency selective fading are as follows:

1) Continuous phase modulation signal After experiencing the frequency selective fading signal, it is expressed as: $r\left(t\right)=\underset{l=0}{\overset{m_{\mathrm{D.}}-1}{\mathrm{\&Sigma;}}}{h}_l\mathrm{the\; s}(t-\mathrm{lT})+z\left(t\right),$ Where z(t) is zero-mean complex Gaussian noise with variance N ₀ ; M _D is the maximum dispersion length of the channel, and h _l is the gain of the L-th path of the channel.

Fig. 4 shows the CPM signal orthogonal decomposition block diagram in the embodiment of the present invention, can adopt Gram-Schmidt orthogonal decomposition method to obtain the orthogonal basis, the number of the orthogonal basis is 2 in our embodiment, therefore, adopt two Orthogonal decomposition is carried out in the way of road correlation, and two road vectors are obtained. After the received CPM signal is decomposed, the correlation of the CPM signal itself is eliminated, so that channel equalization can be performed better.

The continuous phase modulation signal after receiving the fading, removes the cyclic prefix, and obtains the training symbol block and the data symbol block that have experienced fading, and uses the orthogonal base stored locally to do each training symbol block or data symbol block The steps of orthogonal decomposition are:

1) Receive the continuous phase modulation signal r(t) of the frequency selective fading channel, and remove the cyclic prefix for r(t);

2) plus tilt phase Make the state of the phase grid time-invariant, and get r'(t);

3) The orthogonal basis stored locally is f ₀ (t)=1, $f_1\left(t\right)=\frac{1}{\sqrt{1-\frac{4}{{\mathrm{\&pi;}}^2}}}(e^{\frac{\mathrm{j\&pi;t}}{T}}-\frac{j2}{\mathrm{\&pi;}}),$ 0≤t<T, where T is the symbol period; correlate the received signal r′(t) with the tilted phase with two orthogonal bases within the symbol period, and obtain two sets of vectors r ₀ (n) and r ₁ (n).

Fig. 5 shows the process of frequency domain equalization of the CPM signal. After the CPM signal is orthogonally decomposed, two vectors r ₀ (n) and r ₁ (n) are obtained, and these two vectors are transformed into the frequency domain (501, 504). According to the result of channel estimation and the MMSE criterion, in the frequency domain Upper equalization eliminates the influence of the channel (502, 505), transforms the equalized frequency domain data into the time domain for demodulation (503, 506), and completes frequency domain equalization.

After the orthogonal decomposition, transform the training symbol block into the frequency domain for channel estimation, and perform equalization in the frequency domain on the data symbol block according to the channel estimation result, and then transform back to the time domain after removing the influence of channel interference for:

1) Orthogonal decomposition decomposes the training character block and the data symbol block into two vectors at the same time; extract the vector segments r ₀ ^T (n) and r ₁ ^T (n) of the training character block from the two vectors respectively;

2) Transform r ₀ ^T (n), r ₁ ^T (n) and the locally stored training symbol block vectors s ₀ ^T (n), s ₁ ^T (n) into the frequency domain to obtain R ₀ ^T (k ), R ₁ ^T (k) and S ₀ ^T (k), S ₁ ^T (k);

3) According to the minimum mean square error criterion, the estimated value of the channel frequency domain after orthogonal decomposition is:

${\tilde{h}}_0=\frac{h_0^{*}}{{\left|h_0\right|}^2+N_0},$ ${\tilde{h}}_1=\frac{{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">h</figure-callout>}}_1^{*}}{{\left|{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">h</figure-callout>}}_1\right|}^2+N_0}$ ( $h_0=R_0^T/S_0^T,$ ${\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">h</figure-callout>}}_1=R_1^T/S_1^T$ );

$\widetilde{S_0}=\widetilde{H_0}*R_0,$ $\widetilde{S_1}=\widetilde{H_1}*R_1;$ 4) Transform the vectors r ₀ (n) and r ₁ (n) obtained after the orthogonal decomposition of the data symbol block into the frequency domain to obtain R ₀ , R ₁ , and remove the influence of the channel in the frequency domain according to the channel estimation results to obtain Two balanced outputs

$\widetilde{S_0}=\widetilde{h_0}*R_0,$ $\stackrel{~}{{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">S</figure-callout>}}_1}=\stackrel{~}{{\mathrm{<figure-callout\; id="1"\; label="h"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">h</figure-callout>}}_1}*{\mathrm{<figure-callout\; id="1"\; label="R"\; filenames="H2009101545692E00014.png"\; state="\{\{state\}\}">R</figure-callout>}}_1;$

变换到时域，送进解调器解调。5) Will

Transform to the time domain and send it to the demodulator for demodulation.

The present invention is a system for CPM modulation using frequency domain equalization method against the influence of frequency selective fading channel. The method and circuit described here can be completely traditional for the single component of CPM modulation and demodulation. We require the combination of it and frequency domain channel equalization to be protected as an invention. Frequency domain channel equalization includes CPM signal orthogonal decomposition, frequency domain channel estimation and frequency domain equalization. The above is only a specific implementation method for a specific application, but the true spirit and scope of the present invention are not limited thereto. Any person skilled in the art can modify the specific method of a single component to achieve CPM frequency domain in different applications Balanced modem system. The present invention is limited only by the appended claims and their equivalents, which we claim to be protected as the present invention.

Claims (4)

1. A continuous phase modulation frequency domain equalization modulation and demodulation method is characterized in that comprising the steps: 1) Obtain a binary sequence from the data source, map the binary sequence into a modulation symbol sequence, form a symbol block with the modulation symbol and the training symbol, and send it to the continuous phase modulator block by block according to the length of the symbol block; 2) After each symbol block completes the continuous phase modulation, insert the cyclic prefix waveform to obtain the continuous phase modulation signal; 3) Sending the continuous phase modulation signal on the communication channel, and the continuous phase modulation signal undergoes frequency selective fading; 4) Receive the fading continuous phase modulation signal, remove the cyclic prefix, and obtain the training symbol blocks and data symbol blocks that have experienced fading, and use the orthogonal basis stored locally to positively correct each training symbol block and data symbol block cross decomposition; 5) After the orthogonal decomposition, the training symbol block is transformed into the frequency domain for channel estimation, and the data symbol block is equalized in the frequency domain according to the channel estimation result, and then transformed back to the time domain after removing the influence of channel interference; 6) After equalization, the data symbol block converted back to the time domain is sent to the Viterbi decoder for demodulation to obtain the source sequence. 2. A kind of continuous phase modulation frequency domain equalization modulation and demodulation method according to claim 1, it is characterized in that, described obtain binary sequence from data source, binary sequence is mapped into modulation symbol sequence, modulation symbol and training The character symbols form a symbol block, which is sent to the continuous phase modulator block by block according to the length of the symbol block. The steps are: 1) Map the binary sequence a _n obtained from the data source to the data symbol sequence I _n , specifically map 1 to +1, and map 0 to -1; 2) every 252 data symbols I _n and 4 return-to-zero symbols form a data symbol block, and the length of the data symbol block is 256; 3) Insert a training character block every 20 data symbol blocks, and form a frame structure with the training character block as the starting point, and the training character block is still composed of a training character symbol with a length of 252 and 4 return-to-zero symbols; 4) Each symbol block enters the continuous phase modulator in turn, and the phase at the starting point of each symbol block is 0, and the phase at the 231st symbol of each symbol block is also 0, and the continuous phase modulation signal waveform expression is

in

is the shaping function, _In is the data symbol block and the training sequence symbol block,

h=0.5, T is the symbol period. 3. A kind of continuous phase modulation frequency domain equalization modulation and demodulation method according to claim 1, it is characterized in that, after completing continuous phase modulation, each symbol block is inserted into a cyclic prefix waveform to obtain a continuous phase modulation signal The steps are: 1) The size of the cyclic prefix length should be greater than the maximum dispersion length of the channel, and the maximum dispersion length of the frequency selective fading channel is set to 25 symbols, so the length of the cyclic prefix is set to 25 symbols; 2) Return the phase to zero at the 229th symbol, which will use up 2 phase-zero symbols; 3) At the end of each symbol block, another 2 return-to-zero symbols need to be added to return the phase to 0 from any state; 4) For each symbol block, copy the CPM signal waveform of the last 25 symbols to the front of the symbol block to form a cyclic prefix. 4. A method of continuous phase modulation frequency domain equalization modulation and demodulation according to claim 1, characterized in that, the described continuous phase modulation signal is sent on the communication channel, and the continuous phase modulation signal undergoes a frequency selective fading step For: continuous phase modulation signal

After experiencing the frequency selective fading signal, it is expressed as:

Where z(t) is zero-mean complex Gaussian noise with variance N ₀ ; M _D is the maximum dispersion length of the channel, and h _l is the gain of the L-th path of the channel.

Images