JP2022076113A - Log-likelihood ratio calculation circuit and wireless receiver - Google Patents

Abstract

To improve the calculation accuracy of a log-likelihood ratio and improve the error correction capability of low-density parity check.SOLUTION: A log-likelihood ratio LLRN(bit_num) based on thermal noise and a log-likelihood ratio LLRφ(bit_num) based on phase noise are multiplied by weighting coefficients WN and Wφ, respectively, and then added together to calculate a log-likelihood ratio LLRbit_num.SELECTED DRAWING: Figure 1

Description

The present invention relates to a log-likelihood ratio calculation circuit and a wireless receiving device, and more particularly to a circuit for calculating a log-likelihood ratio used in error correction in low-density parity check decoding and a wireless receiving device including the circuit.

Conventionally, in low density parity check (LDPC: short for Low Density Parity Check) decoding, an error is made using the log-likelihood ratio (LLR: short for Log-Likelihood Ratio) calculated based on the distribution of thermal noise of the received signal. A method for making corrections is known (see, for example, Patent Document 1).

Japanese Unexamined Patent Publication No. 2013-201582

By the way, in the high multi-value modulation method, there is a problem that the influence other than the thermal noise such as the phase noise and fading of the received signal is large, and the error correction ability of the low density parity check is reduced. Specifically, the conventional method for calculating the log-likelihood ratio is based on the premise that the amplitude of the received signal fluctuates based on the normal distribution / Gaussian distribution around the ideal symbol point due to the thermal noise of the received signal. However, in this method, it is not possible to consider the fluctuation when the phase of the symbol is rotated due to the phase noise of the received signal, fading, or the like. For this reason, the accuracy of calculating the log-likelihood ratio deteriorates, especially in an environment where phase noise is present, and the error correction capability of the low-density parity check is reduced.

Therefore, the present invention includes a log-likelihood ratio calculation circuit and a radio reception including the log-likelihood ratio calculation circuit, which can improve the calculation accuracy of the log-likelihood ratio and improve the error correction capability of low-density parity check. The purpose is to provide the device.

In order to solve the above problems, the invention according to claim 1 has a log-likelihood ratio based on thermal noise and a log-likelihood ratio based on phase noise, each of which is multiplied by a weighting coefficient and then added to the log-likelihood. It is a log-likelihood ratio calculation circuit characterized by calculating the degree ratio.

According to the second aspect of the present invention, in the logarithmic likelihood ratio calculation circuit according to the first aspect, the logarithmic likelihood ratio based on the phase noise is modulated by a orthogonal phase amplitude modulation method having a multivalued number of 16 on the transmitting side. Of the 4 bits of the received signal, which is distinguished whether it becomes 0 or 1 according to the phase angle of the ideal symbol point, the phase angle of the received signal is used according to the phase angle. It is calculated based on the probability density distribution that becomes 0 and the probability density distribution that becomes 1, and the ideal symbol point that distinguishes whether it becomes 0 or 1 according to the amplitude (called "angle distinction symbol point"). With respect to the bit divided into the ideal symbol point (referred to as "phase distinction symbol point") in which 0 or 1 is distinguished according to the phase angle, the amplitude distinction symbol point and the phase distinction symbol After determining which of the points is based on the amplitude of the received signal, the probability density distribution of the amplitude distinction symbol point becomes 0 according to the amplitude using the amplitude of the received signal. Calculated based on the probability density distribution that becomes 1, and for the phase distinction symbol point, the phase angle of the received signal is used to obtain a probability density distribution that becomes 0 and a probability density distribution that becomes 1 according to the phase angle. It is characterized by being calculated based on.

The invention according to claim 3 is characterized in that, in the log-likelihood ratio calculation circuit according to claim 2, the log-likelihood ratio based on the phase noise of each bit is calculated according to a predetermined mathematical formula.

The invention according to claim 4 is a wireless receiving device comprising the log-likelihood ratio calculation circuit according to any one of claims 1 to 3.

According to the inventions of claims 1 to 3, the degree of mutual priority between the log-likelihood ratio based on thermal noise and the log-likelihood ratio based on phase noise is taken into consideration (specifically, according to the invention. By adding these (after multiplying each log-likelihood ratio by a weighting coefficient) and calculating the log-likelihood ratio of each bit, the log-likelihood ratio is calculated by phase noise and fading in addition to the influence of thermal noise. Since the element of the distribution fluctuation due to is taken into consideration, it is possible to realize a low-density parity inspection that is robust against disturbances including phase noise.

According to the inventions of claims 1 to 3, the probability distribution when the phase of the symbol rotates / fluctuates based on the normal distribution / Gaussian distribution due to the phase noise or fading of the received signal is also taken into consideration. Since the log-likelihood ratio is calculated, it is possible to improve the calculation accuracy of the log-likelihood ratio even in an environment where the phase noise of the received signal is dominant. It is possible to improve the error correction ability.

According to the invention of claim 4, it is possible to exert the above-mentioned effects in a wireless receiving device that performs error correction in low density parity check decoding using a log-likelihood ratio.

It is a functional block diagram which shows the schematic structure of the radio reception device in embodiment which includes the LLR calculation part as the log-likelihood ratio calculation circuit which concerns on embodiment of this invention. It is a functional block diagram which shows the schematic structure of the LLR calculation part as a log-likelihood ratio calculation circuit which concerns on embodiment. It is a figure which shows the arrangement of the ideal symbol point of a 16QAM system, and the bit string assigned to each ideal symbol point. It is a figure explaining the setting of the variable related to the calculation of the log-likelihood ratio. It is a figure explaining the concept of the calculation formula of the log-likelihood ratio of the 1st bit, and in particular, it is a figure explaining the region where the 1st bit is 0 and the region where the 1st bit is 1. It is a figure explaining the concept of the calculation formula of the log-likelihood ratio of the 1st bit. It is a figure explaining the concept of the calculation formula of the log-likelihood ratio of the 2nd bit, and in particular, it is a figure explaining a symbol which puts bit information on an amplitude, and a symbol which puts bit information on a phase. It is a figure explaining the concept of the calculation formula of the log-likelihood ratio of the 2nd bit. It is a figure explaining the concept of the weighting coefficient of a log-likelihood ratio. (A) is a figure explaining the case where the log-likelihood ratio based on thermal noise is prioritized. (B) is a figure explaining the case where the log-likelihood ratio based on the phase noise is prioritized. It is a figure explaining the problem of the conventional method of calculating the log-likelihood ratio, and the measures by the log-likelihood ratio calculation circuit which concerns on this invention. It is a functional block diagram which shows the schematic structure of the evaluation system used in the verification example of the effectiveness of the log-likelihood ratio calculation circuit which concerns on this invention. It is a figure which shows the error rate of the code as the evaluation result in the verification example of the effectiveness of the log-likelihood ratio calculation circuit which concerns on this invention.

Hereinafter, the present invention will be described based on the illustrated embodiment. In the following, the characteristic configuration of the present invention will be described, and the description of the conventional mechanism for performing wireless communication will be omitted. Further, in each figure, the signal line and the imaginary part (Q signal; in other words, the orthogonal component and the Q signal component) that transmit the real part (I signal; in other words, the in-phase component and the I signal component) constituting the complex signal are transmitted. ) Is displayed as one signal line together with the signal line to be transmitted. Further, "greater than or equal to" and "greater than or equal to" in the following description may be replaced with each other, and further, "less than or equal to" and "less than or equal to" may be replaced with each other.

FIG. 1 is a functional block diagram showing a schematic configuration of a wireless receiver 1 according to an embodiment, including an LLR calculation unit 20 as a log-likelihood ratio calculation circuit according to an embodiment of the present invention.

The antenna 10 receives a signal wave transmitted from a radio transmission device (not shown) and transfers the signal wave as a reception signal to the channel filter 11.

Here, the wireless transmission device applies low-density parity inspection coding to the wireless frame to be transmitted and uses a quadrature amplitude modulation method (that is, 16QAM method; QAM is an abbreviation for Quadrature Amplitude Modulation) having a multi-value number of 16. It is modulated and then transmitted from the antenna. That is, the wireless receiving device 1 receives a multi-value modulation signal that has been subjected to low-density parity check coding and is modulated by a quadrature-amplitude modulation method (16QAM method) having a multi-value number of 16.

The channel filter 11 is a filter for limiting the reception band, and may be specifically configured by, for example, a bandpass filter. The channel filter 11 receives the input of the received signal transferred from the antenna 10, performs band limiting processing on the received signal, extracts the received signal in a desired frequency band from the received signal, and outputs the received signal.

The mixer 13 receives the input of the received signal output from the channel filter 11 and receives the input of the local signal output from the local oscillator 12, and multiplies the received signal by the local signal to obtain the frequency of the received signal. Convert (specifically, down-convert) and output.

The automatic gain control unit 14 receives the input of the received signal output from the mixer 13 and performs a gain adjusting process on the received signal so that the level of the received signal becomes constant at a predetermined level to adjust the gain. The received signal (which is an analog signal) after processing (in other words, after level correction) is output.

The A / D converter 15 (Analog-to-Digital converter) receives the input of the received signal output from the automatic gain control unit 14, performs analog-digital conversion processing on the received signal, and converts the received signal from the analog signal. Convert to a digital signal. That is, the A / D converter 15 outputs a digital received signal.

The digital orthogonal detection unit 16 receives an input of a digital reception signal output from the A / D converter 15 and converts the received signal into a baseband signal by orthogonal detection. Specifically, the digital orthogonal detection unit 16 has a multiplier in which a cos wave is input from a numerically controlled oscillator and synchronously detects in-phase components of the received signal, and a sine wave is input from the numerically controlled oscillator to detect the received signal. It is equipped with a multiplier that synchronously detects orthogonal components. The digital orthogonal detection unit 16 performs orthogonal detection of multi-value modulation by digital signal processing, and is expressed by IQ orthogonal coordinates of the real number component (I signal) constituting the real part and the imaginary number component (Q signal) constituting the imaginary part. Complex signals (in other words, detection signals of in-phase components and orthogonal components) are generated and output.

The numerically controlled oscillator of the digital orthogonal detection unit 16 is an oscillator that oscillates at a frequency corresponding to a set value, and generates cos waves and sine waves whose phases differ from each other by 90 ° to each of the two multipliers. Supply.

The roll-off filter 17 receives the input of the baseband signal (in other words, the detection signal of the in-phase component and the orthogonal component) output from the digital orthogonal detection unit 16, and the in-phase component and the orthogonal component of the baseband signal are respectively. The baseband signal is subjected to a process for removing the inter-code interference, and the spectrum waveform of the baseband signal is shaped and output so as to have a desired roll-off.

The equalizer 18 receives the input of the baseband signal output from the roll-off filter 17, and removes intersymbol interference due to frequency selective fading for each of the in-phase component and the orthogonal component of the baseband signal. The baseband signal (received signal) after the adaptive equalization process is output.

The decoding unit 19 receives the input of the log-likelihood ratio (LLR: an abbreviation of Log-Likelihood Ratio) output from the LLR calculation unit 20, and uses the log-likelihood ratio to reduce the density according to, for example, the sum-product decoding method. Perform parity check / decryption processing.

The LLR calculation unit 20 is a circuit for calculating the log-likelihood ratio used in error correction in low-density parity check decoding, and is incorporated in the wireless receiver 1.

FIG. 2 is a functional block diagram showing a schematic configuration of the LLR calculation unit 20 as the log-likelihood ratio calculation circuit according to the embodiment.

The LLR calculation unit 20 according to the embodiment multiplies the log-likelihood ratio LLR _{N (bit_num)} based on thermal noise and the log-likelihood ratio LLR φ _{(bit_num)} based on phase noise by the weighting coefficients W _N and W φ, respectively. The log-likelihood ratio LLR _{bit_num} is calculated by adding the above.

Further, the LLR calculation unit 20 according to the embodiment has a 4-bit received signal in which the logarithmic likelihood ratio LLRφ _{(bit_num)} based on phase noise is modulated by a orthogonal phase amplitude modulation method having a multivalued number of 16 on the transmitting side. For the bit that is distinguished whether it becomes 0 or 1 according to the phase angle φ _i of the ideal symbol point, the probability density distribution that becomes 0 according to the phase angle using the phase angle xφ of the received signal. It is calculated based on the probability density distribution that becomes 1 and the ideal symbol point (called "amplitude distinction symbol point") that distinguishes whether it becomes 0 or 1 according to the amplitude, and according to the phase angle. For a bit that is divided into an ideal symbol point (called a "phase distinction symbol point") in which 0 or 1 is distinguished, it is either an amplitude distinction symbol point or a phase distinction symbol point. After determining whether or not the phase is based on the amplitude x _R of the received signal, the probability density distribution that becomes 0 and the probability density distribution that becomes 1 according to the amplitude are used for the amplitude distinction symbol points using the amplitude x _R of the received signal. The phase distinction symbol point is calculated based on the probability density distribution that becomes 0 and the probability density distribution that becomes 1 according to the phase angle, using the phase angle xφ of the received signal. There is.

First, FIG. 3 shows the arrangement of ideal symbol points in the 16QAM system and the bit sequence assigned to each ideal symbol point (in other words, the signal space diagram of 16QAM). In FIG. 3, A _i is a homeomorphic component of the ideal symbol point arrangement, and B _i is an orthogonal component of the ideal symbol point arrangement (however, i = 1, 2, 3, 4).

The LLR calculation unit 20 calculates the log-likelihood ratio LLR _{bit_num} of the (bit_num) bit according to the following formula 1.
(Number 1) LLR _{bit_num} = W _N x LLR _{N (bit_num)} + Wφ x LLRφ _{(bit_num)}
Here, bit_num: bit number (however, bit_num = 1, 2, 3, 4)
LLR _{N (bit_num)} : Log-likelihood ratio based on thermal noise
LLRφ _{(bit_num)} : Log-likelihood ratio based on phase noise
W _N : Weighting coefficient of log-likelihood ratio based on thermal noise
Wφ: Weighting coefficient of log-likelihood ratio based on phase noise

The log-likelihood ratio LLR _{N (bit_num)} based on the thermal noise of the (bit_num) bit is specifically calculated according to the following formulas 2A to 2D, for example.

The meanings of the symbols and variables in the above formulas 2A to 2D are as follows.
LLR _N1 : Log-like likelihood ratio based on 1-bit thermal noise LLR _N2 : Log-like likelihood ratio based on 2-bit thermal noise LLR _N3 : Log-like likelihood ratio based on 3-bit thermal noise LLR _N4 : 4-bit thermal noise Logarithmic likelihood ratio based on x _i : In-phase component of received signal x _q : Orthogonal component of received signal A _i : In-phase component of ideal symbol point arrangement (however, i = 1, 2, 3, 4)
B _i : Orthogonal component of ideal symbol point arrangement (however, i = 1, 2, 3, 4)
σ ² : Noise dispersion

The log-likelihood ratio LLRφ _{(bit_num)} based on the phase noise of the (bit_num) bit is specifically calculated according to, for example, the following formulas 3 to 6. Here, FIG. 4 shows the setting of the variables used in the following formulas 3 to 6, which are defined for the arrangement of the ideal symbol points shown in FIG.

The meanings of the symbols / variables in the following formulas 3 to 7 are as follows. The "φ" of the phase angle of the ideal symbol point among the following is displayed as "Φ" in the drawing.
xφ: Phase angle of received signal x _R : Amplitude of received signal φ _i : Phase angle of ideal symbol point (however, the subscript i of φ i = 0, 1, 2, ..., 11)
R _i : Amplitude of ideal symbol point (however, subscript i of R = 0, 1, 2)
σ ² : Noise variance R _thre1 : First amplitude threshold R _thre2 : Second amplitude threshold

The log-likelihood ratio LLRφ ₁ based on the phase noise of the first bit is calculated according to the following mathematical formula 3.

The log likelihood ratio LLR φ ₂ based on the second bit phase noise is such that the amplitude x _R of the received signal is equal to or greater than the first amplitude threshold R _thre 1 or equal to or less than the second amplitude threshold R _thre 2 (that is, x _R ≧ R _{thre 1} or x _R ). When _{≤R thre2} ), it is calculated according to the following formula 4A, and the amplitude x _R of the received signal is less than the first amplitude threshold R _thre1 and larger than the second amplitude threshold R _thre2 (that is, R _thre2 <x _R <R). In the case of _thre1 ), it is calculated according to the following formula 4B.

The log-likelihood ratio LLRφ ₃ based on the phase noise of the third bit is calculated according to the following mathematical formula 5.

The log likelihood ratio LLR φ4 based on the phase noise of the _4th bit is such that the amplitude x _R of the received signal is equal to or greater than the first amplitude threshold R _thre1 or equal to or less than the second amplitude threshold R _thre2 (that is, x _R ≧ R _thre1 or x _R ). _{≤R thre2} ), calculated according to the following formula 6A, the amplitude x _R of the received signal is less than the first amplitude threshold R _thre1 and larger than the second amplitude threshold R _thre2 (that is, R _thre2 <x _R <R). In the case of _thre1 ), it is calculated according to the following formula 6B.

The above formulas 3 to 6 are based on the following ideas (see FIGS. 5 to 8).

In the first bit, the phase angle is from −90 ° to + 90 ° (in other words, from 270 ° to 360 ° / 0 ° to 90 °; that is, in FIG. 5, the phase angles Φ ₉ to Φ ₁₁ and Φ ₀ of the ideal symbol point are set. It becomes ₀ in the 4th and 1st quadrants of the IQ orthogonal coordinate system including up to Φ2, and the phase angle is from + 90 ° to -90 ° (in other words, from 90 ° to 180 ° to 270 °; that is. , In FIG. 5, it becomes 1 in the second and third quadrants of the IQ orthogonal coordinate system including the phase angles Φ ₃ to Φ ₅ and Φ ₆ of the ideal symbol point to Φ ₈ . Therefore, assuming that the phase of the received signal fluctuates based on the normal distribution / Gaussian distribution, the probability density that the 1st bit becomes 0 is shown as shown in FIG. 6A, and the probability density that the 1st bit becomes 1 is shown in FIG. 6B. Is shown. Based on this, by considering these probability density distributions as each term, the above equation 3 for calculating the log-likelihood ratio LLRφ ₁ based on the phase noise of the first bit is derived (see FIG. 6C).

The 3rd bit is based on the same idea as the 1st bit, that is, the 3rd bit has a phase angle of 0 ° to + 90 ° and then + 180 ° (that is, the phase angle of the ideal symbol point in FIG. 5 is Φ ₀ to Φ ₂ , It becomes 0 in the first and second quadrants of the IQ orthogonal coordinate system including Φ ₃ and up to Φ ₅ , and the phase angle is 0 ° from + 180 ° to + 270 ° (that is, the ideal symbol point in FIG. 5). It becomes ₁ when the phase angle of Φ ₆ to Φ ₈ and Φ ₉ is included in the IQ orthogonal coordinate system (third quadrant and fourth quadrant). Based on this, the log-likelihood ratio LLRφ ₃ based on the phase noise of the 3rd bit is calculated by considering the probability density distribution in which the 3rd bit becomes 0 and the probability density distribution in which the 3rd bit becomes 1 as each term. The above equation 5 is derived.

Further, since the same idea can be applied to all the quadrants of the IQ orthogonal coordinate system for the second bit, the first quadrant of the IQ orthogonal coordinate system will be described as an example. As shown in FIG. 7, the second bit is an ideal symbol point in which whether it becomes 0 or 1 is distinguished by the amplitude (“symbol with bit information on the amplitude” in the figure; the amplitude distinction. An ideal symbol point (a symbol point) and whether it becomes 0 or 1 is distinguished by the phase (“symbol with bit information on the phase” in the figure; the phase distinction symbol point. ) And. Therefore, by comparing the amplitude x _R of the received signal with the first amplitude threshold R _thre1 and the second amplitude threshold R _thre2 , the received signal / symbol puts the bit information on either the amplitude or the phase. The method of calculating the logarithmic likelihood ratio is changed for each of them. The probability density distribution in which the second bit becomes 0 and the probability density distribution in which the second bit becomes 1 are shown as shown in FIG. 8A for the symbol on which the bit information is placed on the amplitude. The probability density distribution in which the second bit becomes 0 and the probability density distribution in which the second bit becomes 1 are shown as shown in FIG. 8B, respectively. Based on this, by considering these probability density distributions as each term, the above equation 4 for calculating the log-likelihood ratio LLRφ ₂ based on the phase noise of the second bit is derived (see FIG. 8C).

The 4th bit is also based on the same idea as the 2nd bit. For the symbol which puts the bit information on the amplitude, it is the same as the 2nd bit, and for the symbol which puts the bit information on the phase, the 4th bit becomes 0. The phase angles of the ideal symbol points are Φ ₂ , Φ ₃ , Φ ₈ , and Φ ₉ , and the phase angles of the ideal symbol points are Φ ₀ , Φ ₅ , Φ ₆ , and Φ ₁₁ . Based on the fact that there is, the probability density distribution where the 4th bit becomes 0 for the symbol which puts the bit information on the amplitude, the probability density distribution where the 4th bit becomes 1, and the 4th bit is 0 for the symbol which puts the bit information on the phase. By considering the probability density distribution in which the 4th bit becomes 1 and the probability density distribution in which the 4th bit becomes 1 as each term, the above equation 6 for calculating the logarithmic likelihood ratio LLRφ ₄ based on the phase noise of the 4th bit is derived.

The first amplitude threshold R _thre1 and the second amplitude threshold R _thre2 are not limited to specific values. For example, the first amplitude threshold R _thre1 is between the amplitudes _R1 and _R2 of the ideal symbol point. The second amplitude threshold R _thre2 is set between the amplitudes R ₀ and R ₁ of the ideal symbol point, and then the bit information is put on the amplitude and the bit information is put on the phase. It is appropriately set to an appropriate value after considering that it can be satisfactorily distinguished from the symbol. Specifically, for example, it is conceivable that the first amplitude threshold value R _thre1 is set as in the following formula 7A, and the second amplitude threshold value R _thre2 is set as in the following formula 7B.
(Number 7A) R _thre1 = R ₁ + (R ₂ -R ₁ ) / 2
(Number 7B) R _thre2 = _Ro + (R ₁ -R ₀ ) / 2

Then, by performing the low density parity check decoding process using the log-likelihood ratio calculated according to the above equation 1, the log-likelihood ratio LLR _N based on the thermal noise distribution and the logarithm based on the phase noise distribution are obtained. Since the likelihood ratio LLRφ is added, the robustness to phase noise is improved. Further, since the magnitude of the influence of the phase noise differs depending on the magnitude of the thermal noise and the position of the symbol, the log-likelihood based on either the thermal noise or the phase noise is determined by the weighting coefficients WN and _Wφ of each log-likelihood ratio. Whether to prioritize the degree ratio is controlled.

The weighting coefficients W _N and W φ of each log-likelihood ratio are not limited to specific values, respectively, and the phase noise is affected by the magnitude of the thermal noise and the position of the symbol. It is appropriately set to an appropriate value after considering how much priority is given to the log-likelihood ratio based on which of noise and phase noise.

The weighting coefficients WN and _Wφ of each logarithmic likelihood ratio may be calculated by predetermined formulas, and specifically, for example, the weighting coefficient of the logarithmic likelihood ratio based on thermal noise according to the following equation 8A. W _N may be calculated and the weighting coefficient W _N of the logarithmic likelihood ratio based on the phase noise may be calculated according to the following equation 8B.

The meanings of the symbols and variables in the above formulas 8A and 8B are as follows.
N: Dispersion of thermal noise (that is, √N corresponds to the average amplitude of thermal noise)
S _d : QAM method inter-symbol spacing α: Priority adjustment coefficient β: Equivalent adjustment coefficient

The priority adjustment coefficient α determines when the average value (that is, √N) of the amplitude fluctuation due to thermal noise is less than or equal to what percentage of the symbol spacing Sd of the QAM method, the logarithmic likelihood ratio _LLRφ based on the phase noise is prioritized. It is a coefficient for.

The priority adjustment coefficient α is not limited to a specific value, and is appropriately set to an appropriate value after considering the degree of mutual priority between thermal noise and phase noise. Specifically, for example, the priority adjustment coefficient α may be set to any value in the range of about 0.1 to 0.5, to give just one example.

The equivalence adjustment coefficient β is a coefficient for equalizing the magnitude of each log-likelihood ratio when the priority of the log-likelihood ratio due to thermal noise and the log-likelihood ratio due to phase noise are the same.

The equivalence adjustment coefficient β is not limited to a specific value, and the priority of the log-likelihood ratio due to thermal noise and the log-likelihood ratio due to phase noise are the same on the premise of the value of the priority adjustment coefficient α. Sometimes it is set to a value that makes the magnitude of each log-likelihood ratio equal.

The priority adjustment coefficient α and the equivalent adjustment coefficient β are set in advance for the weighting coefficient control unit 24 of the LLR calculation unit 20.

By making the weighting coefficients WN and _Wφ of each log-likelihood ratio calculated according to the above equations 8A and 8B, as shown in FIG. 9A, the average amplitude √N of the thermal noise is between the symbols of the QAM method. When the interval S _d is larger than α × 100 [%], the weighting coefficient W _N for the log-likelihood ratio based on thermal noise becomes large (specifically, it becomes larger than 1), and conversely, it becomes phase noise. The weighting coefficient Wφ with respect to the log-likelihood ratio based on is smaller (specifically, smaller than 1), that is, the log-likelihood ratio based on thermal noise is prioritized and the final log-likelihood ratio LLR _{bit_num} is calculated. It will be. In FIG. 9, α = 0.5 is set as an example only.

On the other hand, as shown in FIG. 9B, when the average amplitude _√N of the thermal noise is α × 100 [%] or less of the inter-symbol spacing Sd of the QAM method, the weighting coefficient Wφ with respect to the logarithmic likelihood ratio based on the phase noise is It becomes larger (specifically, it becomes 1 or more), and conversely, the weighting coefficient W _N for the logarithmic likelihood ratio based on thermal noise becomes smaller (specifically, it becomes 1 or less), that is, it is based on phase noise. The logarithmic likelihood ratio is prioritized and the final logarithmic likelihood ratio LLR _{bit_num} is calculated.

In Equation 8, the weighting coefficient WN for the logarithmic likelihood ratio based on thermal noise and the weighting coefficient _Wφ for the logarithmic likelihood ratio based on phase noise are based on the magnitude of thermal noise (in other words, with thermal noise as a variable). It is designed to be calculated, but based on the magnitude of the phase noise (in other words, with the phase noise as a variable), the weighting factor W _N for the log-possibility ratio based on the thermal noise and the log-probability ratio based on the phase noise. The weighting coefficient Wφ may be calculated.

The thermal noise LLR calculation unit 21 of the LLR calculation unit 20 receives the input of the baseband signal (received signal) output from the equalizer 18, and the in-phase component x _i of the received signal from the baseband signal and the orthogonality of the received signal. The component x _q is acquired, and the logarithmic likelihood ratio LLR _{N (bit_num)} based on the thermal noise of the (bit_num) bit th is calculated and output according to the above equations 2A to 2D.

The phase noise LLR calculation unit 22 receives the input of the baseband signal (received signal) output from the equalizer 18, acquires the phase angle _xφ of the received signal and the amplitude xR of the received signal from the baseband signal. Considering the magnitude relationship between the amplitude x _R of the received signal and the first amplitude threshold R _thre1 and the second amplitude threshold R _thre2 , the phase noise of the (bit_num) bit according to the above equations 3 to 6 is set. Calculates and outputs the based logarithmic likelihood ratio LLRφ _{(bit_num)} .

The thermal noise power estimation unit 23 receives the input of the baseband signal (received signal) output from the equalizer 18, estimates the thermal noise power in the wireless receiver 1 based on the baseband signal, and estimates the thermal noise power. Outputs the estimated amount of.

The weighting coefficient control unit 24 receives the input of the baseband signal (received signal) output from the equalizer 18, and also receives the input of the estimated amount of the thermal noise power output from the thermal noise power estimation unit 23. Using the baseband signal and the estimated thermal noise power (specifically, the dispersion N of the thermal noise is calculated), the weighting coefficient W _N of the logarithmic likelihood ratio based on the thermal noise is calculated according to the above equation 8A. At the same time, the weighting coefficient Wφ of the logarithmic likelihood ratio based on the phase noise is calculated and output according to the above equation 8B.

The first multiplier 25 receives an input of a logarithmic likelihood ratio LLR _{N (bit_num)} based on the thermal noise output from the thermal noise LLR calculation unit 21, and is based on the thermal noise output from the weight coefficient control unit 24. Upon receiving the input of the weighting coefficient W _N of the logarithmic likelihood ratio, the logarithmic likelihood ratio LLR _{N (bit_num)} based on the thermal noise was multiplied by the weighting coefficient WN of the logarithmic likelihood ratio based on the thermal noise and multiplied. The result (that is, W _N × LLR _{N (bit_num)} ) is output.

The second multiplier 26 receives an input of a logarithm likelihood ratio LLRφ _{(bit_num)} output from the phase noise LLR calculation unit 22 and a logarithm based on the phase noise output from the weighting coefficient control unit 24. Upon receiving the input of the weighting coefficient Wφ of the likelihood ratio, the result of multiplying and multiplying the logarithmic likelihood ratio LLRφ _{(bit_num)} based on the phase noise and the weighting coefficient Wφ of the logarithmic likelihood ratio based on the phase noise (that is,). Wφ × LLRφ _{(bit_num)} ) is output.

The adder 27 receives the input of the multiplication result (that is, W _N × LLR _{N (bit_num)} ) output from the first multiplier 25, and the multiplication result output from the second multiplier 26 (that is, that is, the adder 27). Upon receiving the input of Wφ × LLRφ _{(bit_num)} ), these multiplication results are added according to the above formula 1, and the result of the addition is the logarithmic likelihood ratio LLR _{bit_num} (= W _N × LLR _{N (} bit_num)) of the (bit_num) bit. ₎ + Wφ × LLRφ _{(bit_num)} ) is output.

Then, the decoding unit 19 receives the input of the log-likelihood ratio LLR _{bit_num} output from the LLR calculation unit 20, and uses the log-likelihood ratio LLR _{bit_num} to perform low-density parity check decoding processing according to, for example, the sum-project decoding method. I do.

According to the LLR calculation unit 20 as the log-likelihood ratio calculation circuit according to the embodiment, the mutual of the log-likelihood ratio LLR _{N (bit_num)} based on thermal noise and the log-likelihood ratio LLR φ _{(bit_num)} based on phase noise. (Specifically, multiply the log-likelihood ratio LLR _{N (bit_num) by} the weighting coefficient WN and multiply the log-likelihood ratio LLRφ _{(bit_num)} by the weighting coefficient Wφ). By adding these and calculating the log-likelihood ratio LLR _{bit_num} of each bit, the factors of distribution fluctuation due to phase noise and fading are taken into consideration in addition to the influence of thermal noise in the calculation of the log-likelihood ratio. Therefore, it is possible to realize a low-density parity inspection that is robust against disturbances including phase noise.

According to the LLR calculation unit 20 as the log-likelihood ratio calculation circuit according to the embodiment, when the phase of the symbol rotates / fluctuates based on the normal distribution / Gaussian distribution due to phase noise or fading of the received signal. Since the log-likelihood ratio is calculated in consideration of the probability distribution, it is possible to improve the calculation accuracy of the log-likelihood ratio even in an environment where the phase noise of the received signal is dominant. , It is possible to improve the error correction ability of low density parity inspection.

Specifically, as shown in FIG. 10, in the conventional method for calculating the log-likelihood ratio, the amplitude of the received signal fluctuates based on the normal distribution / Gaussian distribution around the ideal symbol point due to the thermal noise of the received signal. (Refer to the broken line circle Cf in the figure). Therefore, if there is a fluctuation when the phase of the symbol is rotated due to the phase noise of the received signal (see the arc arrow Fp in the figure), the signal Sf judged to have a high reception probability is the actual transmission signal Ss. It will be different. On the other hand, in the LLR calculation unit 20 as the log-likelihood ratio calculation circuit according to the embodiment, the distribution of fluctuations due to the phase noise of the received signal is taken into consideration, that is, due to the phase noise of the received signal. Since the log-likelihood ratio is calculated in consideration of the situation where the symbol fluctuates on the envelope E, the fluctuation when the phase of the symbol is rotated by the phase noise of the received signal (arc arrow Fp in the figure). Even if there is (see), it is possible to improve the error correction ability of the low density parity inspection by matching the signal determined to have a high reception probability with the actual transmission signal Ss.

An example of verifying the effectiveness of the log-likelihood ratio calculation circuit according to the present invention will be described below.

In this verification example, the calculation accuracy of the log-likelihood ratio between the conventional method for calculating the log-likelihood ratio and the log-likelihood ratio calculation circuit according to the present invention is compared in an environment where thermal noise and phase noise exist. The evaluation system shown in FIG. 11 was used for the purpose of. In the evaluation system of this verification example, a predetermined sample signal is modulated by the 16QAM method (reference numeral 31 in the figure), phase noise and thermal noise are added (reference numerals 32 and 33), and then the thermal noise of the received signal is measured. A conventional method (reference numeral 34) for calculating a logarithmic likelihood ratio based on a distribution, and a logarithmic likelihood ratio calculation circuit (reference numeral 20) according to the present invention for calculating a logarithmic likelihood ratio in consideration of phase noise of a received signal. It was constructed as a system to compare the calculation accuracy of each logarithmic likelihood ratio of.

The evaluation conditions for this verification example were set as follows.
Modulation method: 16QAM method Carrier-to-noise ratio (CNR: Abbreviation for Carrier-Noise Ratio): 15 to 25 dB
Phase noise level: -60 dBc (value at 1 kHz offset)
QAM method inter-symbol spacing S _d : 2
Priority adjustment coefficient α: 0.125
Equivalent adjustment coefficient β: 100

Here, the log-likelihood ratio takes a positive value when the probability that the transmission bit is 0 is high, and takes a negative value when the probability that the transmission bit is 1 is high. Taking advantage of this, the actual transmission bit is compared with the code of the logarithmic likelihood ratio calculated by the conventional method (reference numeral 34 in FIG. 9) to count the number of times the code is incorrect (reference numeral 35). The actual transmission bit was compared with the code of the logarithmic likelihood ratio calculated by the logarithmic likelihood ratio calculation circuit (reference numeral 20) according to the present invention, and the number of times the code was incorrect was counted (reference numeral 36).

As a result of the evaluation system shown in FIG. 11, as shown in FIG. 12, it was confirmed that the log-likelihood ratio calculation circuit according to the present invention has a smaller code error rate in each carrier-to-noise ratio. It was also confirmed that the higher the carrier-to-noise ratio and the more dominant the phase noise, the greater the improvement rate by the log-likelihood ratio calculation circuit according to the present invention. From this result, it was confirmed that the log-likelihood ratio calculation circuit according to the present invention reduces the code error rate as compared with the conventional log-likelihood ratio calculation method.

Although the embodiment of the present invention has been described above, the specific configuration is not limited to the above-described embodiment, and even if there is a design change or the like within a range that does not deviate from the gist of the present invention. Included in the invention.

Specifically, in the above-described embodiment, the log-likelihood ratio calculation circuit according to the present invention is incorporated as the LLR calculation unit 20 in the wireless receiver 1 whose schematic configuration is shown in FIG. The configuration of the wireless device into which the log-likelihood ratio calculation circuit can be incorporated is not limited to the wireless receiver 1 whose schematic configuration is shown in FIG. 1, and the log-likelihood ratio calculation circuit according to the present invention has another configuration. It may be incorporated into a wireless device.

Further, in the above embodiment, the logarithmic likelihood ratio LLR _{N (bit_num)} based on thermal noise is calculated according to the above equations 2A to 2D, but the logarithmic likelihood ratio LLR _N (bit_num) based on thermal noise is calculated. The calculation formula / calculation method of _{bit_num)} is not limited to the formulas 2A to 2D, and in the above embodiment, the logarithmic likelihood ratio LLRφ _{(bit_num)} based on the phase noise is the above formulas 3 to 6 However, the calculation formula / calculation method of the logarithmic likelihood ratio LLRφ _{(bit_num)} based on the phase noise is not limited to the formulas 3 to 6. That is, the main point of the present invention is to consider the degree of mutual priority between the log-likelihood ratio LLR _{N (bit_num)} based on thermal noise and the log-likelihood ratio LLR φ _{(bit_num)} based on phase noise (specifically). , The log-likelihood ratio LLR _{N (bit_num)} is multiplied by the weighting coefficient W _N and the log-likelihood ratio _{LLRφ (bit_num)} is multiplied by the weighting coefficient Wφ). Is to calculate, and other configurations including the calculation formula / calculation method of the log-likelihood ratio LLR _{N (bit_num)} based on thermal noise and the log-likelihood ratio LLR φ _{(bit_num)} based on phase noise are suitable for a specific configuration. Not limited.

1 Wireless receiver 10 Antenna 11 Channel filter 12 Local oscillator 13 Mixer 14 Automatic gain control unit 15 A / D converter 16 Digital orthogonal detector 17 Roll-off filter 18 Equalizer 19 Decoding unit 20 LLR calculation unit 21 Thermal noise LLR calculation Part 22 Phase noise LLR calculation part 23 Thermal noise power estimation part 24 Weight coefficient control part 25 First multiplier 26 Second multiplier 27 Adder

Claims (4)

The log-likelihood ratio is calculated by multiplying each of the log-likelihood ratio based on thermal noise and the log-likelihood ratio based on phase noise by a weighting coefficient and then adding them.
A log-likelihood ratio calculation circuit characterized by this. The log-likelihood ratio based on the phase noise,
Of the 4-bit received signals modulated by the quadrature amplitude modulation method with a multi-valued number of 16 on the transmitting side
For bits that are distinguished whether they are 0 or 1 depending on the phase angle of the ideal symbol point,
Using the phase angle of the received signal, it is calculated based on the probability density distribution that becomes 0 and the probability density distribution that becomes 1 according to the phase angle.
An ideal symbol point that distinguishes between 0 and 1 depending on the amplitude (called an "amplitude distinction symbol point") and an ideal that distinguishes between 0 and 1 depending on the phase angle. For bits that can be divided into symbol points (called "phase-distinguishing symbol points")
After determining which of the amplitude-distinguishing symbol point and the phase-distinguishing symbol point is based on the amplitude of the received signal,
The amplitude distinction symbol point is calculated using the amplitude of the received signal based on the probability density distribution that becomes 0 and the probability density distribution that becomes 1 according to the amplitude.
The phase distinction symbol point is calculated based on the probability density distribution that becomes 0 and the probability density distribution that becomes 1 according to the phase angle, using the phase angle of the received signal.
The log-likelihood ratio calculation circuit according to claim 1. The log-likelihood ratio based on the phase noise is calculated according to the following mathematical formula 1 for the 1st bit, which is a bit that is distinguished whether it becomes 0 or 1 depending on the phase angle of the ideal symbol point, and the 3rd bit is described below. Calculate the log-likelihood ratio based on the phase noise according to Equation 3 of
The logarithmic likelihood ratio based on the phase noise is calculated according to the following formula 2 for the second bit, which is the bit divided into the amplitude distinction symbol point and the phase distinction symbol point, and the fourth bit is based on the phase noise according to the following formula 4. Calculate the logarithmic likelihood ratio (where equation 2A and equation 4A are used when x _R ≥ R _thre1 or x _R ≤ R _thre2 , and equations 2B and 4B are used when R _thre2 <x _R <R _thre1 . Used),

Here,
xφ: Phase angle of received signal x _R : Amplitude of received signal φ _i : Phase angle of ideal symbol point (however, the subscript i of φ i = 0, 1, 2, ..., 11)
R _i : Amplitude of ideal symbol point (however, subscript i of R = 0, 1, 2)
σ ² : Noise dispersion R _thre1 : First amplitude threshold value R _thre2 : Second amplitude threshold value The log-likelihood ratio calculation circuit according to claim 2. The log-likelihood ratio calculation circuit according to any one of claims 1 to 3 is provided.
A wireless receiver characterized by that.

Images