CN101247204B - Robustness pattern detection implementing method of global digital broadcasting system - Google Patents

Abstract

The robustness mode detection method in the global digital broadcasting system belongs to the robustness mode detection technical field, and it is characterized in that, at first four kinds of modes are traversed with the method of linear weighted correlation modulus, and the respective corresponding correlation value sequences are obtained, Then multiply the correlation values obtained by various modes by constants of different sizes so that the global maximum value appears in the correlation value sequence of the matching mode, and finally determine the current robustness according to the global maximum value in the correlation value sequence model. The scheme is simple and intuitive, can reduce the computational complexity, and even if the signal-to-noise ratio is greatly deteriorated, only four constants need to be adjusted, the judgment accuracy rate is still very high, and the judgment basis of the pattern detection is also very simple.

Description

Robust Pattern Detection Implementation Method in Global Digital Broadcasting System

technical field

The present invention relates to a specific implementation scheme of robust mode detection in a global digital broadcasting (DRM) system. By adopting the method, it not only has the characteristics of small amount of calculation, low computational complexity, and small hardware overhead, but also has simple basis for mode judgment. It has the advantages of intuitiveness and high detection accuracy. In the case of extremely deteriorated signal-to-noise ratio (SNR), this method can still be used to perform robust mode detection correctly. It has high robustness and flexibility and belongs to wireless communication receiving technical field.

Background technique

DRM (Digital Radio Mondiale) is a global digital broadcasting standard, which applies to digital AM broadcasting with frequencies below 30Mhz, including shortwave, medium wave and long wave. In April 2001, the system proposal proposed by the DRM Alliance was adopted as a formal recommendation by the International Telecommunication Union (ITU); it was standardized by the European Telecommunications Standardization Organization (ETSI) in October 2001; and it was approved by the International Electrotechnical Association in March 2002. (IEC) passed, and the DRM system specification came into effect, paving the way for the digitization of AM broadcasting. Some transmitter stations of many broadcasting organizations in the world have been officially put into commercial broadcast operation in the form of DRM since June 2003.

The DRM system adopts the Orthogonal Frequency Division Multiplexing (OFDM) modulation method. After the data to be transmitted is modulated by the Quadrature Amplitude Modulation (QAM), it is mapped to different subcarriers together with the pilot information, and then uses the inverse discrete Fourier transform ( IDFT) completes OFDM modulation and converts the frequency domain signal to the time domain. In order to prevent symbol crosstalk and ensure the orthogonality of subcarriers when the symbol windowing position is inaccurate, a guard interval, cyclic prefix CP, is introduced in the time domain. The cyclic prefix CP is obtained by copying the N-point long useful data part after L points, and adding it to the beginning position of the useful data part. In this way, the CP of point L and the useful data part of point N together form a complete OFDM time-domain symbol, and the overall length of the time-domain symbol is N+L. Finally, the obtained time-domain baseband signal is transmitted through radio frequency carrier modulation.

In order to achieve maximum data transmission efficiency and transmission accuracy and reliability under different channel conditions, the DRM system defines four robustness modes A, B, C, and D. In the DRM standard, the recommended robustness modes under different channel conditions are given, as shown in Table 1:

Table 1 Typical conditions for the application of each robustness mode

Robust Mode Typical channel transmission conditions A Gaussian channel with minor fading B Time and frequency selective channel with long delay transmission C Same as B, but with higher Doppler spread D Same as B, but with severe time delay and Doppler spread

Each robustness mode defines its own related parameters, such as the number of symbols contained in each transmission frame, the duration of CP, the duration of useful data part, subcarrier spacing, and so on. The choice of the robust mode is to better resist the multipath delay spread and Doppler frequency shift introduced by the wireless transmission channel. If the baseband processing processes such as signal synchronization, equalization, and demodulation are to be correctly implemented at the receiver, the parameters related to the DRM signal must be known, so the robustness mode must be judged first. It can be seen that the robust pattern detection in the DRM system plays an important role, which is directly related to the extraction of the DRM signal parameters and is the prerequisite for the receiver to work correctly.

The DRM system also defines 6 frequency band occupancy modes, which are 4.5Khz, 5Khz, 9Khz, 10Khz, 18Khz, and 20Khz. When sampling with a certain clock, the useful data part and the duration of the CP can be converted into the numbers N and L of sampling points, respectively. The following subscripts are used to distinguish the N and L values in each robustness mode, then the lengths of the useful data parts corresponding to the four modes A, B, C, and D are Na, Nb, Nc, and Nd respectively, and the corresponding CP lengths are respectively La, Lb, Lc, Ld.

In the currently used DRM system robust mode detection method, the basic idea is to traverse the N and L parameters of each robust mode, calculate the corresponding correlation value sequence, and then judge the mode by detecting the periodicity of the sequence. Among them, one of the core algorithms is the calculation of the correlation value of two data segments, which respectively refer to the L-long data starting from a certain point and the L-long data starting from an N point in between. In the algorithm for calculating correlation values, there are currently four typical methods: one is the maximum likelihood function method. The disadvantage of this method is that the calculation complexity is high. Calculation of energy; the second is the method of taking the real part of the correlation. The disadvantage of this method is that it cannot eliminate the influence of the fractional frequency offset on the correlation calculation, and it can only be used in the case of a small fractional frequency offset, which has certain limitations. ; The third is to take the sign bit of the data point to calculate the correlation, that is, quantify it to 1+j, -1+j, -1-j, 1-j according to the quadrant where the data point is located, and then do the correlation modulo. Although the duplication and repetition relationship between the CP and the last L points of the useful data part is retained, and the calculation is simple and convenient, if the amplitude of the signal itself is too large, only taking the sign bit will lose the amplitude information of the signal, which will affect the accuracy of the correlation calculation. ; The fourth is the exponential weighted correlation method that takes into account the intersymbol interference. Although this method has a good modeling and approximation characteristic for intersymbol interference, that is, the later points in the CP are less affected by intersymbol interference. The weighting coefficient tends to 1, but because the weighting coefficient is generated by means of an exponential function, the calculation complexity is high, and the obtained weighting coefficient requires high precision. Under the hardware condition of limited word length, the generation and storage of the weighting coefficient will be difficult. loss of precision.

One of the methods currently used for DRM system robustness mode detection is: traverse the N and L parameters of the four robustness modes, and use the maximum likelihood function method to calculate the correlation of each data point in the digital baseband signal of a certain length. Values, four sets of correlation value sequences are obtained, and the case where the correlation peaks are periodically distributed is selected as the result of pattern detection. This method has the following obvious shortcomings: First, the computational complexity is far from the maximum likelihood function, which not only needs to find the correlation, but also needs to solve the energy; second, because the CP of the robust mode A is too short, its peak value is not obvious, and the peak value There is no way to talk about the periodicity between them, so the detection of A mode often fails; the third is that the detection process of the peak periodicity is cumbersome, and the possible false peaks will affect the detection of the period.

The second method of DRM system robustness mode detection currently used is: still similar to method one, the matching robustness mode is determined by detecting the periodicity of the correlation peak obtained by the maximum likelihood method, the difference lies in method two Only traverse B, C, and D modes, and solve the corresponding three sets of correlation value sequences. If there is no periodicity of correlation peaks, the current working mode is considered to be A. Although the accuracy rate of this method is greatly improved, it still has the disadvantages of high computational complexity and complex cycle detection, and because the indirect method of elimination is used to judge the robustness mode A, it is not direct enough and the algorithm efficiency is not high , and if there is no signal at the current frequency point, because it cannot be judged as B, C, or D mode, it will be wrongly judged as the current A mode, so that there are certain limitations in robust mode detection.

Contents of the invention

In order to solve the defect that the existing DRM system robust mode detection method cannot accurately identify the A mode, and overcome the high complexity of related calculations, the difficulty in realizing the exponential weighting coefficient, and the cumbersomeness of mode judgment based on periodic detection, the present invention proposes a A new implementation method of robust pattern detection in the DRM system, traversing the N and L parameters of four robust patterns, adopting the calculation form of linear weighted correlation modulo to solve the respective correlation value sequences, and then the four groups of correlation The value sequence is multiplied by different constants so that the global maximum appears in the correlation value sequence of the matching pattern, and finally the global maximum peak in the four sets of correlation value sequences is searched to determine the current robustness mode. The implementation method of robust mode detection proposed by the invention not only has simple correlation calculation, low calculation complexity and small hardware overhead, but also uses the global maximum value as the basis for judging the mode is very simple and intuitive. Using this implementation method not only reduces the complexity of correlation calculations in the currently used methods, but also greatly simplifies the decision basis for robust pattern detection. The accuracy of robust pattern detection is very high, and the signal-to-noise ratio deteriorates significantly. In the case of , if the constant to be multiplied with the correlation value sequence is correctly selected, the current robustness mode can still be accurately judged. Therefore, the implementation method of robust pattern detection in the DRM system proposed by the present invention has the obvious advantages of simplicity, accuracy, flexibility and strong ability to resist the decline of signal-to-noise ratio.

The present invention is characterized in that the method comprises the following steps in sequence:

Step (1) carries out the traversal of four modes to the digital baseband signal obtained through radio frequency front-end processing and de-IQ modulation, and the steps during each traversal are as follows:

Step (1.1) selects the corresponding useful data part length N and CP length L in the traversal mode;

Step (1.2) obtains the digital baseband signal data obtained after the IQ modulation, so that the time domain position serial number corresponding to the data point is θ, and obtains its corresponding correlation value according to the following formula:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Among them, r(k) is the kth data of the received DRM baseband signal, r ^* (k) represents the conjugate operation, weight (mode, k-θ) is a weight function, which is a function of the working mode and k, and its expression The formula is:

$\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; 2</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 6</span>}}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; 2</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; <</span>\frac{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; Z</span>}\\ \mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>& \frac{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; Z</span>}\end{array}\right.$

L _mode is La, Lb, Lc, and Ld, which represent the CP lengths in A, B, C, and D modes, respectively.

The size of the weighting coefficient is related to the distribution of intersymbol interference (ISI) interference energy, which reflects the strength of the correlation between the data points in the CP and the corresponding copied data points. Since the earlier point in the CP is more affected by the ISI brought by the previous symbol, that is, the greater the ISI interference energy, the weaker the correlation with the copied data point, and the smaller the corresponding weighting coefficient. The weighting function adopted in the present invention simplifies the exponential weighting function and performs linearization processing on it. In each robustness mode, the first half of CP corresponds to a linear increasing function distributed in the range of 0 to 1, and the slope is to the power of 2 to facilitate calculation and reduce hardware overhead; the weighting coefficient corresponding to the second half of CP is 1, this is not only because the point behind the CP is less affected by ISI, but also because the ISI interference energy is mainly concentrated in the first half of the CP, while the energy in the second half of the CP is so small that it can be ignored within consideration.

Step (2) Multiply the four sets of correlation value sequences obtained when traversing each robust mode in step (1) by four constants respectively. If the constant corresponding to mode D is 1, then the constant corresponding to mode A can take a value in the range of 1.6 to 2.6, the constant corresponding to mode B can take a value in the range of 0.9 to 1.7, and the constant corresponding to mode C can be in the range of 0.9 to 1.7 value within the range. The reason for multiplying by 4 constants is to make the global maximum appear in the matching pattern, so that step (3) can judge the pattern by searching the global maximum. If the current signal is in mode A, since the CP of the mode A signal is very short, the correlation peak calculated by using the N and L parameters of mode A is not obvious, and the difference from the maximum value calculated in other modes is very small, and the global maximum value does not necessarily appear In the correlation value sequence of mode A, it is necessary to expand the correlation value of mode A by a certain factor to make it appear an obvious global maximum; and if the current signal is B, C or D mode, use the matching mode N, L The parameters can get a very obvious peak value, and there is a certain redundancy in the magnitude of the global maximum value, even if it is attenuated by a certain multiple, it will not be affected, but the attenuated size cannot be lower than the expanded size of the related value of A mode. So there are certain constraints among the four constants.

Step (3) Search for the maximum value in the four sets of correlation value sequences processed by step (2), and select the global maximum value. The mode in which the global maximum value is located is the judgment result. So far, the robust mode detection is completed.

The method for implementing robust pattern detection in the DRM system proposed by the present invention has the advantages that the introduction of linear weighting coefficients can offset the influence of part of the intersymbol crosstalk, which not only suppresses false peaks, but also has computational complexity compared to exponential weighting. The form has been greatly reduced, the hardware overhead is small, and it is beneficial to the realization of the hardware. In the case of limited hardware word length, the weighting coefficient can be accurately stored; The size of the sequence is used to search for the global maximum, and the simple and intuitive physical quantity of the global maximum is used to perform robust pattern detection. constant, it can still correctly search out that the global maximum value appears in the matching pattern, so as to achieve the purpose of correctly detecting the current pattern. Therefore, the method proposed in the present invention not only has high accuracy, but also has a strong ability to resist the decline of the signal-to-noise ratio. Robust pattern detection with high flexibility.

Description of drawings

Fig. 1 shows the circuit principle modules involved in the implementation method of robust pattern detection in the DRM system.

Fig. 2 is a flowchart of a method for implementing robust pattern detection in a DRM system.

Detailed ways

The implementation method of robust pattern detection in the DRM system proposed by the present invention, its corresponding circuit principle module is composed of four parts: a memory, a correlator, a multiplier, and a comparator:

1) The memory is used to store digital baseband signals, and the subsequent robust mode detection processing is performed through these data;

2) The correlator calculates the correlation value corresponding to each point when traversing the four robustness modes, so that the correlation value sequence corresponding to each robustness mode can be obtained. The correlator uses the following formula to calculate the correlation value of each point:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Among them, N and L are the useful data part length and CP length corresponding to the robustness mode currently traversed respectively, weight(mode, k-θ) is the weighting function corresponding to the robustness mode traversed currently, and θ is the The time domain position number of the data point, r(k) is the kth data of the received DRM baseband signal, and r ^* (k) represents the conjugate operation. The two parts of data for correlation operation in the correlator are L data starting from the data point and L data starting from N data points after the data point. Linear weighting is used to process each product item in the correlation operation, the purpose is to offset part of the intersymbol interference, suppress the generation of false peaks, and simplify the computational complexity of the exponential weighting form; the modulus of the linear weighted correlation results can be removed. The influence of fractional frequency offset;

3) Use a multiplier to multiply the four sets of correlation value sequences obtained through traversal by four definite constants to adjust their amplitudes, so that not only can the B, C, and D modes with obvious global maximums be correctly detected, At the same time, it can correctly detect the A mode whose peak is not obvious because the CP length is too short;

4) The comparator is used to search for the global maximum in the adjusted four sets of correlation value sequences, and the pattern corresponding to the global maximum is the result of robust pattern detection.

In the process of traversing correlations above, the method of determining the linear weighting coefficient is:

$\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; 2</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 6</span>}}\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; 2</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>& \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; <</span>\frac{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; Z</span>}\\ \mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>& \frac{{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; mode</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; Z</span>}\end{array}\right.$

Wherein, L _mode is La, Lb, Lc, and Ld, which represent the CP lengths of A, B, C, and D modes, respectively.

In each robustness mode, the weighting coefficient is a linear increasing function distributed in the range of 0 to 1 in the first half of CP, and the slope is to the power of 2 to facilitate calculation and reduce hardware overhead; in the second half of CP is 1, this is not only because the point behind the CP is less affected by ISI, but also because the ISI interference energy is mainly concentrated in the first half of the CP, while the energy in the second half of the CP is so small that it can be ignored within consideration.

In the above process of adjusting the correlation value sequence, the constants corresponding to each robustness mode are determined as follows:

The reason for the failure of the currently used robust mode detection method is that the CP length of the A mode is too short, so that the peak value and its periodicity are not obvious, thus causing a misjudgment of the A mode. Therefore, the robust mode detection implementation method proposed in the present invention aims at this problem, changes the mode decision basis, and detects the mode by searching for the global maximum value. The four sets of correlation value sequences obtained by ergodic correlation are multiplied by four constants, so that the global maximum must appear in the matching pattern, which is convenient for pattern detection. If the current signal is in mode A, since the CP of the mode A signal is very short, the correlation peak calculated by using the N and L parameters of mode A is not obvious, and the difference from the maximum value calculated in other modes is very small, and the global maximum value does not necessarily appear In the correlation value sequence of mode A, it is necessary to expand the correlation value of mode A by a certain factor to make it appear an obvious global maximum; and if the current signal is mode B, C or D, use the N and L parameters of the matching mode Both can get very obvious peaks, and there is a certain redundancy in the magnitude of the global maximum value, even if it is attenuated by a certain multiple, it will not be affected, but the attenuated size cannot be lower than the expanded size of the A-mode correlation value, so There are certain constraints among the four constants.

Below in conjunction with accompanying drawing, describe content of the present invention in detail:

Fig. 1 shows the circuit principle modules involved in the implementation method of robust pattern detection in the DRM system. As shown in Figure 1, the implementation of robust pattern detection follows the RF front-end processing and IQ modulation modules. First store a certain length of digital baseband signals obtained by de-IQ modulation, such as baseband signals with a length of 5 A-mode symbols, and use these data to perform subsequent robust mode detection processing. Then traverse the N and L parameters of various robustness modes. When a robustness mode is selected, select the corresponding N and L values of the mode, and sequentially use the linear weighted correlation adopted by this method for the stored data points. The modulo expression calculates its corresponding correlation value through the correlator, so that four sets of correlation value sequences corresponding to the four robustness modes can be obtained. In order to correctly identify the A mode with a very short CP without affecting the judgment of other modes, four constants are introduced to adjust the four sets of correlation value sequences, and the multipliers are used to multiply the constants to the corresponding correlation value sequences , so that the subsequent comparator can correctly search for the global maximum value. Finally, a comparator is used to search for the global maximum in the correlation value sequence output by the multiplier. No matter in which robust mode, the searched global maximum value must appear in the matching mode, and the detection of the robust mode is completed correctly.

Fig. 2 is a flowchart of a method for implementing robust pattern detection in a DRM system. As shown in Figure 2, the specific process of the robust pattern detection implementation method in the DRM system proposed by the present invention is:

1) Read data of a certain length from the digital baseband signal obtained by the RF front end of the receiver and the IQ modulation process, such as 5 A-mode symbol lengths, that is, 5 (Na+La) data points, and store them in the memory for Lu Used for sticky mode detection;

2) Take the useful data part length Na and CP length La corresponding to the A mode, and for the data points of the front part stored in the memory, such as the first 5 (Na+La)-(Na+La) data points, pass through the correlator in turn To calculate the value of linear weighted correlation modulo, the formula used is:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Among them, N=Na, L=La, θ is the time-domain position sequence number of the data point,

weight factor $\mathrm{weight}(\mathrm{mode},\mathrm{no})=\left\{\begin{array}{cc}{2}^{-6}\mathrm{no}+(1-2^{-5}{L}_{\mathrm{mode}}),& 0\&le;\mathrm{no}<\frac{L_{\mathrm{mode}}}{2},\mathrm{no}\&Element;Z\\ 1,& \frac{L_{\mathrm{mode}}}{2}\&le;\mathrm{no}<L_{\mathrm{mode}},\mathrm{no}\&Element;Z\end{array}\right.$

In this way, a set of correlation value sequences corresponding to the A mode will be obtained;

3) Take the useful data part length Nb and CP length Lb corresponding to the B mode, and store the data points in the front part of the memory, such as the first 5 (Na+La)-(Nb+Lb) data points, through the correlator in turn To calculate the value of linear weighted correlation modulo, the formula used is:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Among them, N=Nb, L=Lb, θ is the time-domain position number of the data point,

weight factor $\mathrm{weight}(\mathrm{mode},\mathrm{no})=\left\{\begin{array}{cc}{2}^{-6}\mathrm{no}+(1-2^{-5}{L}_{\mathrm{mode}}),& 0\&le;\mathrm{no}<\frac{L_{\mathrm{mode}}}{2},\mathrm{no}\&Element;Z\\ 1,& \frac{L_{\mathrm{mode}}}{2}\&le;\mathrm{no}<L_{\mathrm{mode}},\mathrm{no}\&Element;Z\end{array}\right.$

In this way, a set of correlation value sequences corresponding to the B mode will be obtained;

4) Take the useful data part length Nc and CP length Lc corresponding to the C mode, and store the data points in the front part of the memory, such as the first 5 (Na+La)-(Nc+Lc) data points, through the correlator in turn To calculate the value of linear weighted correlation modulo, the formula used is:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Among them, N=Nc, L=Lc, θ is the time-domain position number of the data point,

weight factor $\mathrm{weight}(\mathrm{mode},\mathrm{no})=\left\{\begin{array}{cc}{2}^{-6}\mathrm{no}+(1-2^{-5}{L}_{\mathrm{mode}}),& 0\&le;\mathrm{no}<\frac{L_{\mathrm{mode}}}{2},\mathrm{no}\&Element;Z\\ 1,& \frac{L_{\mathrm{mode}}}{2}\&le;\mathrm{no}<L_{\mathrm{mode}},\mathrm{no}\&Element;Z\end{array}\right.$

In this way, a set of correlation value sequences corresponding to the C mode will be obtained;

5) Take the useful data part length Nd and CP length Ld corresponding to the D mode, and store the data points in the front part of the memory, such as the first 5 (Na+La)-(Nd+Ld) data points, through the correlator in turn To calculate the value of linear weighted correlation modulo, the formula used is:

$\mathrm{<span\; class="notranslate">\; Cor</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; mode</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; r</span>}}^{<span\; class="notranslate">\; *</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span><span\; class="notranslate">\; |</span>$

Wherein, N=Nd, L=La, θ is the time-domain position sequence number of the data point,

weight factor $\mathrm{weight}(\mathrm{mode},\mathrm{no})=\left\{\begin{array}{cc}{2}^{-6}\mathrm{no}+(1-2^{-5}{L}_{\mathrm{mode}}),& 0\&le;\mathrm{no}<\frac{L_{\mathrm{mode}}}{2},\mathrm{no}\&Element;Z\\ 1,& \frac{L_{\mathrm{mode}}}{2}\&le;\mathrm{no}<L_{\mathrm{mode}},\mathrm{no}\&Element;Z\end{array}\right.$

In this way, a set of correlation value sequences corresponding to the D mode will be obtained;

6) Using a multiplier to multiply the correlation value sequence under the A mode by a definite constant;

7) Using a multiplier to multiply the correlation value sequence under the B mode by a definite constant;

8) Using a multiplier to multiply the correlation value sequence under the C mode by a definite constant;

9) using a multiplier to multiply the correlation value sequence in the D mode by a certain constant;

10) search for the global maximum in the four groups of correlation value sequences by a comparator;

11) Judgment mode, the mode corresponding to the global maximum value is the robustness mode detection result.

Claims (1)

1. the robustness pattern detection method in the global digital broadcasting system is characterized in that, described method realizes in the digital integrated circuit of receiving terminal successively according to the following steps

Step (1) accesses from memory through radio-frequency front-end and handles, separates the digital baseband signal that the IQ modulation obtains, and described digital baseband signal is sent to a correlator carries out A, B, C, the D detection of totally four kinds of robustness patterns, and its step is as follows when detecting at every turn:

Step (1.1) is chosen useful data partial-length N and Cyclic Prefix part length L in the digital baseband signal that correspondence is set under this pattern;

Step (1.2) is got the data q that is positioned at the front preseting length in the digital baseband signal that global digital broadcasting system that separating described in the step (1) obtain after the IQ modulation uses, wherein the length of q is useful data partial-length and the Cyclic Prefix part length sum that the integral multiple of A pattern Baud Length deducts the robustness pattern correspondence that is traveling through again, make wherein the time-domain position sequence number of data point correspondence is θ, obtain the value Cor (θ) of the relevant delivery of the pairing linear weighted function of data q described in this step (1.2) as follows

$\mathrm{Cor}\left(\mathrm{\&theta;}\right)=|\underset{k=\mathrm{\&theta;}}{\overset{\mathrm{\&theta;}+L-1}{\mathrm{\&Sigma;}}}\mathrm{weight}(\mathrm{mode},k-\mathrm{\&theta;})\&CenterDot;[r\left(k\right)r^{*}(k+N)\left]\right|,$

Wherein:

Weight (mode, k-θ) is a weighting function, is the function of described robustness pattern and k, represents with following formula:

$\mathrm{weight}(\mathrm{mode},n)=\left\{\begin{array}{cc}{2}^{-6}n+(1-2^{-5}{L}_{\mathrm{mode}})& ,0\&le;n<\frac{L_{\mathrm{mode}}}{2},n\&Element;Z\\ 1& ,\frac{L_{\mathrm{mode}}}{2}\&le;n<L_{\mathrm{mode}},n\&Element;Z\end{array}\right.,$

N is an integer, is limited to 0 under it, and the upper limit depends on the circulating prefix-length L under the current traversal mode _Mode, L _ModeBe La, Lb, Lc, Ld, represent the CP length under A, B, C, the D pattern respectively; R is a received signal; * be the complex conjugate symbol;

Step (2) multiply by the constant of the fixed size of a setting respectively on four groups of sequence of correlation values that step (1.2) traversal obtains with multiplier, wherein the constant of D pattern correspondence is 1, the constant of A pattern correspondence is value in 1.6 to 2.6 scopes, the constant of B pattern correspondence is value in 0.9 to 1.7 scope, the constant of C pattern correspondence is value in 0.9 to 1.7 scope, makes global maximum necessarily appear in the robustness pattern that is complementary;

Step (3) is handled the global maximum in four groups of sequence of correlation values that obtain with a comparator search process step (2), the pairing pattern of this global maximum is the robustness pattern that judgement obtains.

Images