TWI640182B - Receiving device and log-likelihood generating method - Google Patents

Abstract

A receiving device includes an equalizer for generating an equalized signal and a first signal, the equalized signal being located in a decision area, and a sequence estimator for generating an estimate based on the first signal a signal; a decoder; and a pair-to-number ratio ratio calculator for generating a plurality of logarithmic probability ratios based on the equalized signal and the estimated signal, wherein when the estimated signal is not located in the decision region, at least The one-to-one ratio is 0.

Description

Receiving device and logarithmic probability ratio generating method

The present invention relates to a receiving apparatus and a logarithmic probability ratio generating method, and more particularly to a receiving apparatus and a logarithmic probability ratio generating method for avoiding erroneous deferral.

Decision feedback equalizer (DFE) has been widely used in the receiving end. However, the decision feedback equalizer has the disadvantage of Error Propagation. That is to say, when the decision is wrong, the decision feedback equalizer will equalize the received signal according to the wrong decision, thereby reducing the performance of the equalizer and generating more wrong decisions.

On the other hand, the receiving end of the current communication system can decode the signal by using a low density parity check code (LDPC) decoder, and the LDPC decoder needs the correct logarithmic probability ratio to correctly decode, and the decision feedback equalizer False deferral will output more erroneous logarithmic odds ratios to the LDPC decoder, resulting in an increase in the error rate of the LDPC decoder.

Therefore, how to generate a logarithmic ratio that avoids erroneous deferral has become one of the goals of the industry.

Accordingly, it is a primary object of the present invention to provide a receiving apparatus and a logarithmic ratio ratio generating method that avoid erroneous deferral to improve the disadvantages of the prior art.

The embodiment of the invention discloses a receiving device, comprising an equalizer for receiving a received signal and generating an equalized signal and a first signal, wherein the equalized signal corresponds to a plurality of bits, and the equalized signal is located a decision area in a plurality of decision areas, each decision area of the plurality of decision areas corresponding to a set of bit values of a plurality of bits; a sequence estimator coupled to the equalizer for a first signal, generating an estimated signal; a pair of odds ratio calculator coupled to the equalizer and the sequence estimator for generating a corresponding number based on the equalized signal and the estimated signal a plurality of logarithmic probability ratios of the plurality of bits, wherein when the estimated signal is not located in the decision region, the ratio of the at least one pair of the logarithmic odds ratio is 0; and a decoder coupled to the A log-probability ratio calculator for decoding based on the complex log-probability ratio.

The embodiment of the present invention further discloses a log-probability ratio generating method, which is applied to a pair-amount ratio ratio calculator of a receiving device, wherein the log-probability ratio calculator is coupled to the equalizer and a sequence of the receiving device. The estimator, the log-probability ratio generating method includes receiving an equalization signal and an estimation signal from the equalizer and the sequence estimator, wherein the equalizer generates the equalization signal according to a received signal, The equalization signal corresponds to a plurality of bits, and the equalization signal is located in a decision region of the plurality of decision regions, each decision region of the plurality of decision regions corresponding to a set of bit values of the plurality of bits; and according to the plurality of bits And the estimated signal, generating a plurality of log-probability ratios corresponding to the plurality of bits, wherein at least one pair of values in the plurality of log-probability ratios when the estimated signal is not located in the decision region The ratio is 0; wherein a decoder of the receiving device decodes according to the plurality of log-probability ratio numbers.

Please refer to FIG. 1. FIG. 1 is a block diagram of a receiving apparatus 10 according to an embodiment of the present invention. The receiving device 10 is used in a communication system comprising an equalizer 12, a sequence estimator 14, a pairwise probability ratio calculator 16, and a decoder 18. The equalizer 12 can be a decision feedback equalizer (DFE) for receiving a received signal y _{n+ Δ} and generating an equalized signal z _n and a signal r _n according to the received signal y _{n+ Δ} . The sequence estimator 14 is coupled to the equalizer 12 for generating an estimated signal x _t,n according to the signal r _{n ,} and the sequence estimator 14 can be a Maximum Likeli sequence estimator (Maximum Likelihood Sequence Estimator, MLSE). Wherein, the equalization signal z _n and the received signal y _n+Δ respectively represent the _equalized signal received at time _n and the received signal received at time (n+Δ), and the equalized signal z _n can be demodulated ( Demodulated) into a plurality of bits b ₀ ~ b _L-1, i.e. equalized signal z _n may correspond to a plurality of bits b ₀ ~ b _L-1. The log-probability ratio calculator 16 is coupled to the equalizer 12 and the sequence estimator 14 for comparing the equalized signal z _n and the estimated signal x _t,n and corresponding to the plurality of bits according to the comparison result. The complex logarithm of the elements b ₀ to b _L-1 is greater than LLR ₀ to LLR _L-1 . In the subsequent description, the equalization signal z _{n is} compared with the estimated signal x _t,n to compare the distribution of the equalization signal z _n and the estimated signal x _t,n to a constellation plane, which will be described in detail in Rear. The decoder 18 can be a Low Density Parity Check (LDPC) decoder coupled to the Logarithmic Probability Ratio Calculator 16 for decoding according to the logarithmic probability ratio LLR ₀ to LLR _L-1 . .

In general, if the equalization signal z _n is M-PSK (Phase Shift Keying), M-PAM (Pulse Amplitude Modulation) or M-order The modulated signal is modulated by quadrature amplitude modulation (M-QAM), and the equalized signal z _n can be demodulated into L bits b ₀ to b _L-1 , where M=2 ^L . In addition, for demodulation, a constellation plane (or a complex plane) can be divided into a plurality of decision regions (Decision Region) R ₍₀₎ to R _(M-1) , and decision regions R ₍₀₎ to R _(M) Each of the decision regions in _-1) corresponds to a set of bit values of the bits b ₀ to b _L-1 . For example, please refer to FIG. 3 together. FIG. 3 is a schematic diagram of distribution of four constellation points of QPSK in a constellation plane according to an embodiment of the present invention. Any point s on the constellation plane can demodulate a set of bit values b _{1 .} b ₀ , wherein 4 QPSK constellation points can be Gray Coded, that is, the bit value groups of adjacent QPSK constellation points are different only in a single bit, 4 QPSKs and the bit value groups they represent are drawn. Shown in the upper left of Figure 3. In order to demodulate the bit value group b ₁ b ₀ , the constellation plane can be divided into decision regions R ₍₀₎ to R ₍₃₎ (shown in the lower right diagram of FIG. 3), located in the decision region R ₍₀₎ The constellation points of R ₍₁₎ , R ₍₂₎ , and R ₍₃₎ may respectively represent (demodulated to) b ₁ b ₀ =00, b ₁ b ₀ =01, b ₁ b ₀ =10 or b ₁ b ₀ =11.

In order to avoid equalizer 12 (which may be a DFE) Deferred error (Error Propagation) may be generated, the receiving apparatus 10 using a sequence estimator 14 to estimate sequence to produce estimated signal x _{t, n,} The log-probability ratio calculator 16 can generate a log-probability ratio LLR ₀ -LLR _L-1 based on the comparison of the equalization signal z _n with the estimated signal x _t,n . Specifically, when the equalization signal z _n and the estimated signal x _t,n are respectively located in different decision regions in the decision regions R ₍₀₎ to R _(M-1) , the logarithmic probability ratio LLR ₀ to LLR _L- At least one pair of odds ratios in ₁ is 0 in LLR _k to avoid an erroneous logarithmic probability ratio being passed to decoder 18. That is, the equalization signal z _n is located in a decision region R _(m1) in the decision regions R ₍₀₎ - R _{(M-1 )} and the estimated signal x _t,n is located in the decision region R ₍₀₎ - R _{( M-1) In the case} of the first decision region R _(m2) (R _(m2) ≠ R _(m1) ), the logarithmic generality ratio LLR ₀ to LLR _L-1 is at least one pair of the odds ratio LLR _k is zero.

In detail, for any digital modulation, when the bit b _k represented by the signal s is to be demodulated, the entire constellation plane can be divided into a binary region R _k,0 and a binary region R _{k , 1} , and the binary region R _k,0 and the binary region R _k,1 divide the entire constellation plane (representing the binary region R _k,0 and the binary region R _{k,1 are} mutually exclusive and the binary region R _k,0 The union with the binary region R _k,1 is the entire constellation plane). The binary region R _k,0 represents the region where b _k =0 is demodulated, and the binary region R _k,1 represents the region where b _k =1 is demodulated, in other words, if the signal s is located in the binary region R _{k , 0} , the bit b _k of the demodulated signal s is 0; if the signal s is located in the binary region R _k,1 , the bit b _k of the demodulated signal s is 1 (where the binary region R _{k, 0} and the binary region R _k,1 can be considered to correspond to the bit b _k ). When the equalization signal z _n is located in the binary region R _k,1 and the estimated signal x _t,n is located in the binary region R _k,0 (or the equalization signal z _n is located in the binary region R _k,0 and the estimated signal x _{When t, n} is located in the binary region R _k,1 ), the log-probability ratio calculator 16 outputs a log-probability ratio LLR _k of zero.

In one embodiment, the log likelihood of the equalized signal according to the estimated signal z _n x _{t, n,} than the result (i.e., judgment signal z _n and the estimated signal x _{t, n} ratio calculator 16 is located The same decision area) produces a complex signal (erase signal) and a logarithmic probability ratio based on the erase signal. Please refer to FIG. 2, which is a block diagram of a pair-to-digits ratio calculator 26 according to an embodiment of the present invention. Log likelihood ratio calculator 26 may be used to achieve the log likelihood ratio calculator 16, which may be used to signal to other z _n (on behalf of the QPSK modulation signal z _n, etc. can be demodulated is demodulated into bits to be b ₀ , b ₁ ). The log-probability ratio calculator 26 includes an erasing unit 260 and a pair-to-number ratio ratio calculating unit 262. The erasing unit 260 is configured to generate an erasing signal q _n according to the equalized signal z _n and the estimated signal x _{t, n} . The erase signal q _n can be expressed as q _n =q _{n, I} +jq _n,Q ,q _n,I represents the real part of the erase signal q _n (hereinafter referred to as the erased real part), q _{n, Q} represents the wipe In addition to the imaginary part of the signal q _n (hereinafter referred to as the erased imaginary part); the log-probability ratio calculation unit 262 is used to generate a log-probability ratio LLR ₀ , LLR ₁ from the erase signal q _n . In an embodiment, the log-probability ratio calculation unit 262 can calculate the log-probability ratios LLR ₀ and LLR ₁ according to the characteristics of the QPSK in the constellation plane, respectively, LLR ₀ = c × q _{n, I} (formula 1.1) and LLR. ₁ = c × q _{n, Q} (Equation 1.2), where c is a constant, and the constant c can be proportional to a Signal-to-Noise Ratio (SNR).

Taking the QPSK signal as an example, as shown in the upper right diagram of FIG. 3, the binary regions R _0,1 , R _0,0 are the right and left half planes of the constellation plane (ie, R _0,1 can be expressed as R _0,1 ={s | Re( s) ≥0}, R _0,0 can be expressed as R _0,0 ={s | Re(s)<0}), as shown in the lower left diagram of Figure 3, the binary region R _1,1 , R _1,0 are the upper and lower half planes of the constellation plane (ie, R _1,1 can be expressed as R _1,1 ={s | Im( s) ≥0}, R _1,0 can be expressed as R _1,0 ={s | Im (s)<0}), where Re(·) and Im(·) represent the real part operation and the imaginary part operation, respectively. In other words, for the QPSK signal, the equalization signal z _n and the estimated signal x _t,n can be determined by judging the real part size or the imaginary part size of the equalization signal z _n and the estimated signal x _{t,n .} Located in the area where you are located. In this case, when the equal real part Re(z _n ) of the equalization signal z _n and the estimated real part Re(x _t,n ) of the estimated signal x _t,n are mutually different (ie, sign( Re (z _n)) not equal _{sign (Re (x t, n} )), where sign (·) for the collection of sign operation, sign (·) can be represented as formula 2), of log likelihood output calculator 26 ratio for The probability ratio LLR ₀ is 0; when the equalized imaginary part Im(z _n ) of the equalization signal z _n and the estimated imaginary part Im(x _t,n ) of the estimated signal x _t,n are mutually different ( That is, sign(Im(z _n )) is not equal to sign(Im (x _t,n ))), and the logarithmic probability ratio calculator 26 outputs a logarithmic probability ratio LLR ₁ of 0. From another point of view, when the equalized real part Re(z _n ) and the estimated real part Re(x _t,n ) are mutually different, the erasing unit 260 can generate the erased real part q _{n, I} is 0, the log-probability ratio calculation unit 262 can generate a log-probability ratio LLR ₀ of 0 according to the formula 1.1. When the equalized imaginary part Im(z _n ) and the estimated imaginary part Im(x _t,n ) are mutually different, the erasing unit 260 can generate the erased imaginary part q _{n , Q} is 0, and the logarithmic probability ratio The calculation unit 262 can generate a log-probability ratio LLR ₁ of 0 according to Equation 1.2. On the other hand, when the equal real part Re(z _n ) and the estimated real part Re(x _t,n ) are identical to each other, the erasing unit 260 can generate the erased real part q _{n, and the I} is equalized. Part Re(z _n ); when the equalized imaginary part Im(z _n ) and the estimated imaginary part Im(x _t,n ) are identical to each other, the erasing unit 260 may generate the erasing imaginary part q _{n, Q} is Equalize the imaginary part Im(z _n ). (Formula 2)

The erase unit 260 is not limited to being implemented with a specific circuit structure. For example, please refer to FIG. 4 , which is a block diagram of an erasing unit 460 according to an embodiment of the present invention. The erasing unit 460 can be used to implement the erasing unit 260 . The erasing unit 460 includes a real unit R1, R2, an imaginary unit I1, I2, a sign unit S1 to S4, a judging unit D1, D2, a multiplexer M1, M2, and a combining unit CB, and the real unit R1, R2 respectively The equalized real part Re(z _n ) is generated, and the real part Re(x _t,n ) is estimated, and the imaginary part units I1 and I2 respectively generate the equalized imaginary part Im(z _n ) and the estimated imaginary part Im(x _{t, n} ), the sign units S1, S2, S3, S4 will equalize the real part Re(z _n ), estimate the real part Re(x _t,n ), equalize the imaginary part Im(z _n ), estimate the virtual The part Im(x _{t, n} ) is substituted into the formula 2 to obtain positive and negative values s1, s2, s3, and s4. The determining unit D1 is configured to determine whether the positive/negative value s1 is equal to the positive or negative value s2. If yes, the determining unit D1 controls the multiplexer M1 to output the erased real part q _{n, where I} is equalized real part Re(z _n ); if not, The judging unit D1 controls the multiplexer M1 to output the erased real part q _{n, and I} is 0. Similarly, the determining unit D2 is used to determine whether the positive/negative value s3 is equal to the positive or negative value s4, and if so, the determining unit D2 controls the multiplexer M2 to output the erased imaginary part q _{n , Q} is the equalized imaginary part Im(z _n ); If not, the judging unit D2 controls the multiplexer M2 to output the erasing imaginary part q _{n , which} is 0. The combining unit CB outputs the erase signal q _n as q _n =q _n,I +jq _n,Q .

From the above, when the other portions of the real Re (z _n) with the estimated real part Re (x _{t, n)} mutually different numbers, and other representatives of b bits for demodulating the real part Re ₀ to (z _n It may be wrong, so the erase unit 260/460 erases the equalized real part Re(z _n ), and instead, the erase unit 260/460 outputs the erased real part q _{n , I} is 0 and according to the formula 1.1 A logarithmic probability ratio LLR ₀ can be generated to be 0. Similarly, when the equalized imaginary part Im(z _n ) and the estimated imaginary part Im(x _t,n ) are mutually different, the equalized imaginary part Im(z _n ) for demodulating the bit b ₁ is represented. It may be wrong, so the erasing unit 260/460 erases the equalized imaginary part Im(z _n ), and the erase unit 260/460 outputs the erased imaginary part q _{n , Q} is 0 and can be according to the formula 1.2 Generate a logarithmic probability ratio LLR ₁ to zero. In short, the erasing unit 260/460 first generates the erase signal q _n and then generates a logarithmic probability ratio LLR ₀ , LLR ₁ according to the erase signal q _n .

In another embodiment, a log likelihood ratio calculator 16 can calculate the first equalized signal z _n in accordance with a pre-estimate of the degree of re-alignment of the equalized signal estimate signal z _n ratio _{LLR 0 '~ LLR L-1} ', embodiment x _t,n is the position on the constellation plane, and it is decided whether to erase the pre-logarithm probability ratio LLR ₀ '~LLR _L-1 ', where the pre-logarithm probability ratio LLR ₀ '~LLR _{L- 1} 'Second-first logarithmic probability is erased than LLR _k ' to represent the output log-probability ratio LLR _k is 0 (k=0,...,L-1). Please refer to FIG. 5. FIG. 5 is a block diagram of a one-to-seven-degree ratio calculator 56 according to an embodiment of the present invention. The log-probability ratio calculator 56 includes a pair-to-number ratio ratio calculating unit 562 and an erasing unit 560. Of log likelihood ratio calculation unit 562 may first calculate corresponds to the equalized signal z _n equalized signal z _n represents bit b ₀ ~ b advance log likelihood ratio _{LLR 0 '~ LLR L-1} ' of the _L-1, In detail, the log-probability ratio calculation unit 562 can apply the conventional formula for calculating the log-probability ratio, and calculate the pre-number likelihood ratios LLR ₀ ' to LLR _L-1 ' based on the equalization signal z _n . In general, the pre-number likelihood LLR _k ' can be expressed as Equation 3, and for different modulation methods, the log-probability ratio calculation unit 562 can apply a formula (this formula is related to the illuminant value group and its corresponding constellation) Calculate the pre-number likelihood LLR ₀ '~LLR _L-1 ' by focusing on the constellation plane arrangement). For example, for M-PSK, the pre-number likelihood LLR _k ' can be expressed as Equation 4, and for 16-QAM, in an embodiment, the pre-number likelihood LLR ₀ ' can be expressed as formula 5, wherein, α, β of the transmission _{signal, α I, β I (α} Q, β Q) to transmit a signal [alpha], the real part beta] (the imaginary ^{_{part), S k (0) /}} S k (1) corresponding to The set of constellation points with b _k =0/1, SNR is the signal to noise ratio.

(Formula 3)

(Formula 4)

(Equation 5)

In addition, the erasing unit 560 is coupled to the equalizer 12 and the sequence estimator 14 for comparing the equalized signal z _n with the estimated signal x _t,n (the equalizing signal z _n and the estimated signal x Whether _t,n is located in the same decision area), and based on the comparison result of the equalization signal z _n and the estimated signal x _t,n , the output logarithmic probability ratio LLR _k is a pre-logarithm probability ratio LLR _k 'or 0 . Specifically, when the erasing unit 560 determines that the equalization signal z _n and the estimated signal x _t,n one of the signals is located in the binary region R _k,0 corresponding to the bit b _k and the other signal is located corresponding to When the binary region R _{k,1 of the} bit b _k , the erasing unit 560 outputs a logarithmic probability ratio LLR _k is 0; and when the erasing unit 560 determines that the equalized signal z _n and the estimated signal x _{t,n are} simultaneously located When the same binary region (R _{k, 0} or R _{k, 1} ) is used, the erasing unit 560 outputs a log-probability ratio LLR _{k which} is a pre-logarithm probability ratio LLR _k '.

For example, please refer to Figure 3 (taking QPSK as an example), assuming that the equalization signal z _n is located in the decision region R ₍₃₎ , if the estimated signal x _t,n is located in the decision region R ₍₁₎ , the erase unit 560 output logarithmic probability ratio LLR ₀ is the pre-logarithm probabilistic ratio LLR ₀ '(LLR ₀ = LLR ₀ ') and the output log-probability ratio LLR ₁ is 0 (LLR ₁ =0); if the estimated signal x _t,n Located in the decision area R ₍₂₎ , the erasing unit 560 outputs a logarithmic probability ratio LLR ₀ of 0 (LLR ₀ =0) and outputs a logarithmic probability ratio LLR ₁ to a pre-logarithm probability ratio LLR ₁ '(LLR ₁ =LLR ₁ '); if the estimated signal x _t,n is located in the decision region R ₍₀₎ , the erasing unit 560 outputs a log-probability ratio LLR ₀ of 0 (LLR ₀ =0) and outputs a log-probability ratio LLR ₁ of 0 (LLR) ₁ =0).

Please refer to FIG. 6. FIG. 6 is a schematic diagram showing the distribution of four constellation points of the 4-PAM in the constellation plane according to the embodiment of the present invention. The four 4-PAMs and the bit value groups they represent are shown in FIG. In the upper left figure, the decision areas R ₍₀₎ to R ₍₃₎ corresponding to each set of bit values are shown in the lower right diagram of FIG. Similarly, the binary region R _0,0 corresponding to the bit b ₀ and the binary region R _{0,1 are} shown in the upper right diagram of FIG. 6 , and the binary region R _1,0 corresponding to the bit b ₁ and The binary region R _{1,1 is} shown in the lower left diagram of Fig. 6. Assuming that the equalization signal z _n is located in the decision region R ₍₃₎ , if the estimated signal x _t,n is located in the decision region R ₍₁₎ , the erasing unit 560 outputs a logarithmic probability ratio LLR ₀ to LLR ₀ 'and outputs a logarithmic probability ratio LLR ₁ is 0; if the estimated signal x _{t, n} positioned decision region R _(2), erasing unit 560 outputs a log likelihood ratio of the LLR ₀ is 0 and outputs a log likelihood ratio LLR ₁ as LLR ₁ '; if the estimate The measurement signal x _t,n is located in the decision region R ₍₀₎ , and the erasing unit 560 outputs a log-probability ratio LLR ₀ 0 and outputs a log-probability ratio LLR ₁ of zero.

Please refer to FIG. 7. FIG. 7 is a schematic diagram showing the distribution of eight constellation points of the 8-PSK in the constellation plane according to the embodiment of the present invention. The eight constellation points of the 8-PSK and the bit value group and the decision region R _{(represented) 0)} ～R _{(7) is} shown in the upper left diagram of FIG. 7, and the binary region R _0,0 corresponding to the bit b ₀ and the binary region R _{0,1 are} shown in the upper right diagram of FIG. 7 . The binary region R _1,0 and the binary region R _1,1 corresponding to the element b ₁ are shown in the lower left diagram of FIG. 7 , and the binary region R _2,0 corresponding to the bit b ₂ and the binary region R _{2, 1 is} shown in the lower right diagram of Figure 7. Assuming that the equalization signal z _n is located in the decision region R ₍₇₎ , and according to the foregoing principle, the estimation signal x _t,n is located in the decision regions R ₍₀₎ to R ₍₇₎ , the erasing unit 560 outputs a logarithmic probability ratio LLR ₀ to LLR _{2 are} shown in Table 1. Table I <TABLE border="1"borderColor="#000000"width="85%"><TBODY><tr><td></td><td>R₍₀₎</td><td>R₍₁₎</td><td>R₍₂₎</td><td>R₍₃₎</td><td>R₍₄₎</td><td>R₍₅₎</td><td>R₍₆₎</td><td>R₍₇₎</td></tr><tr><td>LLR₀</Td><td> 0 </td><td>LLR₀'</td><td> 0 </td><td>LLR₀'</Td><td> 0 </td><td>LLR₀'</td><td> 0 </td><td>LLR₀'</Td></tr><tr><td>LLR₁</td><td> 0 </td><td> 0 </td><td>LLR₁'</td><td>LLR₁'</td><td> 0 </td><td> 0 </td><td>LLR₁'</td><td>LLR₁'</td></tr><tr><td>LLR₂</td><td> 0 </td><td> 0 </td><td> 0 </td><td> 0 </td><td>LLR₂'</td><td>LLR₂'</td><td>LLR₂'</td><td>LLR₂'</td></Tr></TBODY></TABLE>

In addition, the operation of the logarithmic probability ratio calculator 16 can be summarized as a one-to-severity ratio generation method 80, please refer to FIG. The log-probability ratio generation method 80 can be performed by the log-probability ratio calculator 16. For details of the logarithmic probability ratio generation method 80, please refer to the aforementioned related paragraphs, and details are not described herein again.

In addition, regarding the structure of the equalizer 12, please refer to FIG. 9, which shows a block diagram of the equalizer 12 of the embodiment of the present invention. The equalizer 12 is a decision feedback equalizer, and the equalizer 12 includes a feed forward equalizer (FFE) 120, a decision unit 122, a feedback equalizer (FBE) 124, and an addition. The 126, 128 and the subtractor 129. The feedforward equalizer 120 is configured to generate a feedforward output signal a _n according to the received signal y _{n+ Δ} , and the feedforward output signal a _n can be expressed as a _n = f _i y _{n+ Δ - i} , where f _i represents the ith filter coefficient of the feedforward equalizer 120 and Nf represents the filter length of the feedforward equalizer 120. The decision unit 122 for equalized signal z _n, generates a decision signal x _{h, n.} The feedback equalizer 124 is configured to generate a feedback output signal u _n and a partial (Sial) signal v _n according to the decision signal x _h,n , and the feedback output signal u _n can be expressed as u _n = b _i x _h,ni , where b _i represents the ith filter coefficient of the feedback equalizer 124 and Nb represents the filter length of the feedback equalizer 124. In addition, the partial sum signal v _n is the feedback output signal u _n minus the multiplication result of the mth filter coefficient and the decision signal x _{h, nm} , ie v _n = b _i x _h,ni - b _m x _h,nm . The adder 126 to the output signal of the feedforward output and the feedback signal a _n u _n summed to produce an equalized signal z _n is the addition result a _n feedforward output signal and the feedback of the output signal u _n. The adder 128 to the feedforward output signal a _n v _n partial signals and summed to produce the addition result signal r _n is a _n feedforward output signal and the portion of the signal v _n. The equalizer 12 outputs a signal r _n to the sequence estimator 14, sequence estimator 14 can generate estimated signal according to the signal x _t r _{n, n.} In addition, the subtracter 129 calculates the subtraction result of the equalization signal z _n and the decision signal x _h,n to generate an error signal e, the filter coefficient f _{i of the} feedforward equalizer 120, and the filtering of the feedback equalizer 124. The coefficient b _i can be adjusted according to the error signal e. For the operation of the equalizer 12 and the sequence estimator 14, please refer to the contents disclosed by the applicant in the Republic of China Patent Application No. 104128970 and the Republic of China Patent Application No. 105102644, and details are not described herein.

In summary, in order to avoid the error deferral caused by the decision feedback equalizer, the present invention uses the log-probability ratio calculator to determine whether the equalization signal and the estimated signal are located in the same decision area, thereby generating a logarithmic probability. For example, when the equalization signal and the estimated signal are located in different decision regions, at least one pair of odds ratios is zero. The above are only the preferred embodiments of the present invention, and all changes and modifications made to the scope of the present invention should be within the scope of the present invention.

10‧‧‧ Receiving device

12‧‧‧ Equalizer

14‧‧‧Sequence estimator

16, 26, 56‧‧‧ logarithmic ratio ratio calculator

18‧‧‧Decoder

120‧‧‧Feed-feed equalizer

122‧‧‧Decision unit

124‧‧‧Feedback equalizer

126, 128‧‧ ‧ adder

129‧‧‧Subtractor

260, 460, 560‧‧‧ erasing unit

262, 562‧‧‧ logarithmic odds ratio calculation unit

b ₀ ～b ₂ ‧‧‧ bits

CB‧‧‧ combination unit

D1, D2‧‧‧ judgment unit

I1, I2‧‧‧ imaginary unit

LLR ₀ ~ LLR _L-1 ‧‧‧ logarithmic odds ratio

LLR ₀ '~LLR _L-1 '‧‧‧pre-logarithm ratio

M1, M2‧‧‧ multiplexer

q _n,I ‧‧‧Erasing the real part

q _n,Q ‧‧‧Erase the imaginary part

R1, R2‧‧‧ real unit

R ₍₀₎ ~ R ₍₇₎ ‧ ‧ decision area

R _0,0 to R _2,0 , R _0,1 to R _2,1 ‧‧‧ binary region

a _n , e , q _n , r _n , u _n , v _n , x _h,n , x _t,n , y _n+Δ , z _n ‧‧‧ signals

S1～S4‧‧‧positive unit

FIG. 1 is a block diagram of a receiving apparatus according to an embodiment of the present invention. 2 is a block diagram of a pair-to-digit ratio ratio calculator according to an embodiment of the present invention. FIG. 3 is a schematic diagram showing a plurality of constellation points according to an embodiment of the present invention. FIG. 4 is a block diagram of an erasing unit according to an embodiment of the present invention. Figure 5 is a block diagram of a pairwise probability ratio calculator in accordance with an embodiment of the present invention. FIG. 6 is a schematic diagram showing a plurality of constellation points according to an embodiment of the present invention. FIG. 7 is a schematic diagram showing a plurality of constellation points according to an embodiment of the present invention. FIG. 8 is a flow chart of a method for generating a one-to-one ratio ratio according to an embodiment of the present invention. FIG. 9 is a block diagram of an equalizer according to an embodiment of the present invention.

Claims (17)

A receiving device, comprising: an equalizer for receiving a received signal, and generating an equalized signal and a first signal, wherein the equalized signal corresponds to a plurality of bits, and the equalized signal is located in a plurality of decision regions a first decision area, each decision area of the plurality of decision areas corresponds to a set of bit values of a plurality of bits; a sequence estimator coupled to the equalizer for using the first signal, Generating an estimated signal; and a pair of odds ratio calculator coupled to the equalizer and the sequence estimator for generating a plurality of bits corresponding to the equalized signal and the estimated signal a plurality of logarithmic probabilities ratio, wherein when the estimated signal is not located in the decision region, the at least one pair of odds ratios of the plurality of logarithmic probabilities are 0; wherein the logarithmic probability ratio is coupled to the calculator A decoder for decoding based on the plurality of logarithmic odds ratios. The receiving device of claim 1, wherein the logarithm ratio ratio calculator comprises: an erasing unit coupled to the equalizer and the sequence estimator; and a pair-to-number ratio ratio calculating unit coupled to The erase unit. The receiving device of claim 2, wherein the erasing unit generates an erasing signal according to the equalized signal and the estimated signal, and the logarithmic probability ratio calculating unit generates the plurality of logarithmic probabilities according to the erasing signal For example, when the estimated signal is not located in the decision area, an erased imaginary part of the erase signal or a erased real part is zero. The receiving device of claim 3, wherein when the realized real part of the equalized signal and the estimated real part of the estimated signal are mutually different, the erasing unit generates the erased signal Erasing the real part to 0; when the first-order imaginary part of the equalization signal and the estimated imaginary part of the estimated signal are mutually different, the erasing unit generates the erasing imaginary part of the erasing signal as 0. The receiving device of claim 4, wherein when the real part and the estimated real part are identical to each other, the erasing unit generates the erasing real part as the real part; when the equalizing When the imaginary part and the estimated imaginary part are identical to each other, the erasing unit generates the imaginary part as the imaginary part. The receiving device of claim 5, wherein the erasing unit comprises: a first real unit for generating the real part; and a second real unit for generating the estimated real part; a first imaginary unit for generating the imaginary part; a second imaginary unit for generating the estimated imaginary part; a first sign unit for generating a corresponding one of the real part a positive and negative value; a second sign unit for generating a second positive and negative value corresponding to the estimated real part; a third sign unit for generating a third positive and negative value corresponding to the imaginary part; a fourth sign unit for generating a fourth positive and negative value corresponding to the estimated imaginary part; a first multiplexer, when the first positive and negative value is not equal to the second positive and negative value, the first plurality The workpiece outputs the erased real part to 0, and when the first positive and negative value is equal to the second positive and negative value, the first multiplexer outputs the erased real part as the real part; and a second When the third positive and negative value is not equal to the fourth positive and negative value, the second multiplexer outputs the erased imaginary part to 0, and When the first positive and negative value is equal to the second positive and negative value, the second multiplexer outputs the erased imaginary part as the imaginary part. The receiving apparatus according to claim 2, wherein the logarithmic probability ratio calculating unit generates a plurality of pre-logarithm ratio ratios corresponding to the plurality of bits according to the equalization signal, the erasing unit according to the plurality of pre-logarithms The ratio is greater than the ratio of the logarithm of the plurality of bits, wherein the one bit of the plurality of bits corresponds to a first binary region and a second binary region, and the first binary region and the second binary region are divided. a constellation plane, when the equalized signal region is located in the first binary region and the estimated signal is located in the second region, the erasing unit outputs a pair of numbers corresponding to the bit in the plurality of logarithmic probabilities The probability ratio is 0. The receiving device according to claim 7, wherein when the estimated signal is located in the first region, the erasing unit outputs the log-probability ratio corresponding to the bit to correspond to the plurality of pre-logarithm probabilities A pre-logarithm probabilistic ratio of the bit. The receiving device of claim 1, wherein the equalizer is a decision feedback equalizer, comprising: a feedforward equalizer for generating a feedforward output signal according to the received signal; And generating a decision signal according to the equalized signal; a feedback equalizer for generating a feedback output signal and a portion of the sum signal according to the decision signal; and an adder for outputting the feedforward signal And adding the partial sum signal to generate the first signal as an addition result of the feedforward output signal and the partial sum signal. The receiving device of claim 1, wherein the sequence estimator is a maximum likelihood sequence estimator. A logarithmic ratio ratio generating method is applied to a pairwise ratio ratio calculator of a receiving device, wherein the logarithmic probability ratio calculator is coupled to an equalizer of the receiving device and a sequence estimator, the logarithm The method for generating an overview ratio includes: receiving an equalization signal and an estimation signal from the equalizer and the sequence estimator, wherein the equalizer generates the equalization signal according to a received signal, the equalization signal corresponding to the plurality a bit, the equalization signal is located in a decision region of the plurality of decision regions, each decision region of the plurality of decision regions corresponds to a set of bit values of the plurality of bits; and the estimated signal and the estimate Measuring a signal, generating a plurality of logarithmic probabilities ratio corresponding to the plurality of bits, wherein when the estimated signal is not located in the decision region, the at least one pair of odds ratios of the plurality of logarithmic odds ratios is 0; And a decoder of the receiving device decodes according to the plurality of log-probability ratio numbers. The method for generating a logarithmic probability ratio according to claim 11, wherein the step of generating a plurality of logarithmic probability ratios corresponding to the plurality of bits according to the equalized signal and the estimated signal comprises: according to the equalized signal And the estimator signal, wherein an erase signal is generated, wherein when the estimated signal is not located in the decision area, an erase imaginary part or a erased real part of the erase signal is 0; and the complex number is generated according to the erase signal Logarithmic probabilities. The method for generating a logarithmic probability ratio according to claim 12, wherein the step of generating the erase signal according to the equalization signal and the estimated signal comprises: when the equalization real part of the equalization signal and the estimation When the estimated real part of the signal is different from each other, the erased real part of the erased signal is 0; and when the equalized imaginary part of the equalized signal and an estimated imaginary part of the estimated signal are mutually When it is an alien sign, the erased imaginary part of the erase signal is 0. The method according to claim 13, wherein the step of generating the erase signal according to the equalization signal and the estimated signal comprises: when the real part is identical to the estimated real part When the number is generated, the erased real part is the real part; and when the imaginary part is identical to the estimated imaginary part, the erased imaginary part is generated as the imaginary part. The method for generating a logarithmic probability ratio according to claim 11, wherein the step of generating a plurality of logarithmic probability ratios corresponding to the plurality of bits according to the equalized signal and the estimated signal comprises: according to the equalized signal Generating a plurality of pre-logarithm probabilities ratio corresponding to the plurality of bits; and outputting the plurality of logarithmic probabilities ratio according to the plurality of pre-logarithm probabilities. The method for generating a logarithmic probability ratio according to claim 15, wherein one bit of the plurality of bits corresponds to a first binary region and a second binary region, the first binary region and the second second The meta-region is divided into a constellation plane, and according to the plurality of pre-logarithm probabilities, the step of outputting the plurality of log-probability ratios includes: when the equalized signal region is located in the first binary region and the estimated signal is located in the In the second binary region, the ratio of the pair of numbers corresponding to the bit in the complex logarithmic probability ratio is zero. The method for generating a logarithmic probability ratio according to claim 15, wherein the step of outputting the plurality of logarithmic probability ratios according to the plurality of pre-logarithm probabilities ratio comprises: when the estimated signal is located in the first region, corresponding to The log-probability ratio for the bit is a pre-logarithm probabilities ratio corresponding to the bit in the plurality of pre-logarithm probabilities.