Soft information demodulation method for 8PSK and 16APSK of satellite communication

Abstract

The invention relates to the field of satellite communication and satellite broadcast high-order modulation mode demodulation, in particular to a 8PSK and 16APSK soft information demodulation method for satellite communication. The method mainly aims at reducing the number of times of solving the square of the Euclidean distance in a max-log scheme and reducing the number of used comparators. In 8PSK, the square of 8 Euclidean distances required to be calculated in a max-log scheme is reduced to the square of only 4 Euclidean distances required to be calculated, and the number of comparators is reduced from 14 to 1. At 16 APSK, the square of 16 Euclidean distances required to be calculated in the max-log scheme is reduced to the square of 8 Euclidean distances required to be calculated, and the number of comparators is reduced from 32 to 8. And the operation speed is improved.

Description

Technical Field

[0001] The invention relates to the field of satellite communication and satellite broadcasting high-order modulation mode demodulation, and in particular to a soft information demodulation method for 8PSK and 16APSK of satellite communication. Background Art

[0002] With the development and progress of satellite communications and satellite broadcasting, the available spectrum resources are becoming increasingly scarce. One solution is to improve spectrum utilization. It is difficult to improve spectrum utilization with traditional low-order modulation methods, which requires the selection of high-order modulation methods for satellite communications. Since satellite communications are over long distances and the signal-to-noise ratio is often low, most of them choose 8PSK and 16APSK modulation methods. In the satellite broadcasting (DVB) protocol, the encoding and decoding method is a combination of BCH and LDPC. The decoder requires the demodulator to output soft information for each bit of each symbol, namely the log-likelihood ratio (LLR).

[0003] The calculation formula of LLR is:

[0004]

[0005] Where r is the received signal, s i For all constellation points, σ 2 is the noise power.

[0006] As can be seen from the formula, in the process of calculating LLR, it is necessary to calculate the Euclidean distance between the received signal point and all constellation points, find the exponent, sum, and then take the logarithm. The amount of calculation is very large and the complexity is very high. It is difficult to implement on the hardware platform of satellite communication, so it is more practical to study its simplified solution. In order to simplify the calculation, many studies have provided various methods to reduce the complexity of LLR calculation, among which the most commonly used is the Max-log approximation scheme.

[0007] The Max-log approximation is:

[0008]

[0009] Where r is the received signal, s i For all constellation points, σ 2 is the noise power.

[0010] The formula is interpreted as: When the bit b is required i When LLR is calculated, r to b i = 1, and the minimum distance min1 is obtained by the comparator, and the distance from r to b is calculated. i = 0, the minimum value min0 is obtained by the comparator, and then the bit b is obtained i of The following is the Max-log approximation scheme, two modulation modes of 8PSK and 16APSK, the calculation amount of Euclidean distance and the number of comparators.

[0011] 8PSK standard constellation diagram Figure 1 As shown, the relationship between the data bit and the output carrier phase of the 8PSK modulator is shown in Table 1.

[0012] Table 1 Relationship between data bit and 8PSK modulator output carrier phase

[0013] serial number b2b1b0 Carrier Phase 1 000 +45° 2 100 +90° 3 110 +135° 4 010 +180° 5 011 +225° 6 111 +270° 7 101 +315° 8 001 +0°

[0014] Q i is the square of the Euclidean distance between the received signal r and the constellation diagram numbered i.

[0015] 1. Log-likelihood ratio of b0:

[0016]

[0017] 2. Log-likelihood ratio of b1:

[0018]

[0019] 3. Log-likelihood ratio of b2:

[0020]

[0021] The Max-log method needs to calculate 8 square distances for each received signal, and a total of 18 comparators are used to find the minimum value. When the comparators are reused, the number of comparators can be optimized and reduced to 14.

[0022] 16PSK standard constellation diagram Figure 2 As shown in FIG. 1 , the 4+12 constellation diagram of 16APSK consists of an inner circle and an outer circle, the inner circle radius is R1, and the outer circle radius is R2. The relationship between the data bit and the output carrier amplitude and phase of the 16APSK modulator is shown in Table 2.

[0023] Table 2 Relationship between data bits and 16APSK modulator output carrier amplitude and phase

[0024]

[0025]

[0026] Q i is the square of the Euclidean distance between the received signal r and the constellation diagram numbered i.

[0027] 1. Log-likelihood ratio of b0:

[0028]

[0029] 2. Log-likelihood ratio of b1:

[0030]

[0031] 3. Log-likelihood ratio of b2:

[0032]

[0033] 4. Log-likelihood ratio of b3:

[0034]

[0035] The above Max-log method requires 16 square distances for each received signal, and a total of 56 comparators are used to find the minimum value. When the comparators are multiplexed, the number of comparators can be optimized and reduced to 32.

[0036] The disadvantage of the Max-log approximate solution for the log-likelihood ratio LLR is that the number of times the square of the Euclidean distance is solved is too large, and more comparators are used when selecting the minimum value, resulting in higher complexity. Summary of the invention

[0037] The invention aims to reduce the number of times of solving the square of Euclidean distance and the number of comparators to simplify the calculation of LLR and improve the operation speed. A soft information demodulation method of 8PSK and 16APSK for satellite communication is proposed.

[0038] The technical solution adopted by the present invention is:

[0039] A soft information demodulation method for 8PSK and 16APSK of satellite communication includes the following processes:

[0040] The 8PSK processing process is as follows:

[0041] (101) Rotate the 8PSK standard constellation diagram clockwise by π / 8;

[0042] (102) Rotate the received signal r clockwise by π / 8 to obtain r1, r1 = r*e (-1j*π / 8) ;

[0043] (103) Take the absolute value of the real and imaginary parts of the rotated point r1 and map it to the first quadrant to obtain r2:

[0044] r2=abs(real(r1))+1j*abs(imag(r1));

[0045] (104) Select four points on the constellation diagram whose bits are 001, 000, 100 and 110 after clockwise rotation and number them 1, 2, 3, 4, and calculate the square of the Euclidean distance between r2 and the selected four points;

[0046] (105)8PSK three bits are represented as b2b1b0;

[0047] Calculate the log-likelihood ratio of bit position b0:

[0048]

[0049] Calculate the log-likelihood ratio of bit position b1:

[0050]

[0051] Calculate the log-likelihood ratio of bit position b2:

[0052]

[0053] Among them, Q i is the square of the Euclidean distance between r2 and the constellation point numbered i among the four selected points;

[0054] (106) Determine the sign of the real and imaginary parts of point r1. If the real part of r1 is negative, then LLR1 = -LLR1; if the imaginary part of r1 is negative, then LLR0 = -LLR0;

[0055] The 16APSK processing process is:

[0056] (201) Take the absolute value of the real and imaginary parts of the received signal r and map it to the first quadrant to obtain r1:

[0057] r1=abs(real(r))+1j*abs(imag(r));

[0058] (202) Select eight points with bits 0101, 0100, 0000, 1000, 1010, 1101, 1100 and 1110 on the 16APSK standard constellation diagram and number them from 1 to 8 in sequence, and calculate the square of the Euclidean distance between r1 and the selected 8 points;

[0059] (203) 16APSK four bits are represented as b3b2b1b0

[0060] Calculate the log-likelihood ratio of bit position b0:

[0061]

[0062] Calculate the log-likelihood ratio of bit position b1:

[0063]

[0064] Calculate the log-likelihood ratio of bit position b2:

[0065]

[0066] Calculate the log-likelihood ratio for bit position b3:

[0067]

[0068] Among them, Q i is the square of the Euclidean distance between r1 and the constellation point numbered i among the 8 selected points;

[0069] (204) Determine the sign of the real and imaginary parts of the input signal r. If the real part of r is negative, then LLR1 = -LLR1; if the imaginary part of r is negative, then LLR0 = -LLR0.

[0070] The advantages of the present invention compared with the prior art are:

[0071] The present invention is mainly devoted to reducing the number of times the max-log scheme calculates the square of the Euclidean distance and reducing the number of comparators used. In 8PSK, the max-log scheme needs to calculate 8 squares of the Euclidean distances to only need to calculate 4 squares of the Euclidean distances, and the number of comparators is reduced from 14 to 1. In 16APSK, the max-log scheme needs to calculate 16 squares of the Euclidean distances to only need to calculate 8 squares of the Euclidean distances, and the number of comparators is reduced from 32 to 8.

[0072] 8PSK:

[0073] Find the degree of squared distance Number of comparators Max-log 8 14 Improved Max-log 4 1

[0074] 16APSK:

[0075] Find the degree of squared distance Number of comparators Max-log 16 32 Improved Max-log 8 8 BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 This is the standard constellation diagram for 8PSK.

[0077] Figure 2 This is the 16APSK standard constellation diagram.

[0078] Figure 3 Detailed flowchart for calculating soft information for the improved max-log scheme for 8PSK.

[0079] Figure 4 The constellation diagram of 8PSK standard constellation diagram is rotated by π / 8.

[0080] Figure 5 Schematic diagram of selecting reference points for 8PSK rotation constellation diagram.

[0081] Figure 6 Detailed flowchart for calculating soft information for the improved max-log scheme for 16APSK.

[0082] Figure 7 Schematic diagram of selecting reference points for the 16APSK standard constellation diagram. DETAILED DESCRIPTION

[0083] The present invention will be further explained below in conjunction with the accompanying drawings.

[0084] A soft information demodulation method for 8PSK and 16APSK of satellite communication includes the following processes:

[0085] like Figure 3 As shown, the 8PSK processing process is:

[0086] (101) By observing the standard constellation diagram of 8PSK, it can be seen that after rotating the standard constellation diagram clockwise by π / 8, Figure 4 As shown, the three bits of information corresponding to the constellation point have axisymmetric or central symmetric characteristics. The three bits are b2b1b0, b0 is symmetric about the x-axis, b1 is symmetric about the y-axis, and b2 is symmetric about the origin;

[0087] (102) Rotate the received signal r clockwise by π / 8 to obtain r1, r1 = r*e (-1j*π / 8) ;

[0088] (103) Take the absolute value of the real and imaginary parts of the rotated point r1 and map it to the first quadrant to obtain r2:

[0089] r2=abs(real(r1))+1j*abs(imag(r1));

[0090] (104) Select the four points where the bits on the constellation diagram are 001, 000, 100 and 110 after clockwise rotation, such as Figure 5 Select the four points in the yellow area and calculate the square of the Euclidean distance between r2 and the selected four points. The numbering of each point is shown in the figure; Table 3 shows the correspondence between the serial number of the 8PSK reference point and the data bit and carrier phase.

[0091] Table 38 Correspondence between the sequence number of PSK reference point and data bit and carrier phase

[0092]

[0093]

[0094] (105)8PSK three bits are represented as b2b1b0;

[0095] Calculate the log-likelihood ratio of bit position b0:

[0096]

[0097] Calculate the log-likelihood ratio of bit position b1:

[0098]

[0099] Calculate the log-likelihood ratio of bit position b2:

[0100]

[0101] Among them, Q i is the square of the Euclidean distance between r2 and the constellation point numbered i among the four selected points;

[0102] (106) Determine the sign of the real and imaginary parts of point r1. If the real part of r1 is negative, then LLR1 = -LLR1; if the imaginary part of r1 is negative, then LLR0 = -LLR0;

[0103] This improved max-log scheme needs to calculate the square of four Euclidean distances, and only needs one comparator, that is, only the minimum square distance between the second and third points needs to be compared.

[0104] In principle, the floating-point performance of this scheme should be exactly the same as that of the max-log scheme, and simulations have shown that the performance is exactly the same.

[0105] like Figure 6 As shown in the figure, by observing the standard constellation diagram of 16APSK, it can be seen that the 4 bits of information corresponding to the constellation points in the figure have axial symmetry or central symmetry characteristics. The 4 bits are b3b2b1b0, b0 is symmetrical about the x-axis, b1 is symmetrical about the y-axis, and b2 and b3 are symmetrical about the origin. The 16APSK processing process is:

[0106] (201) Take the absolute value of the real and imaginary parts of the received signal r and map it to the first quadrant to obtain r1:

[0107] r1=abs(real(r))+1j*abs(imag(r));

[0108] (202) Select eight points on the 16APSK standard constellation diagram, where the bits are 0101, 0100, 0000, 1000, 1010, 1101, 1100 and 1110, as follows: Figure 7 The 8 points in the yellow area are marked, and the square of the Euclidean distance between r1 and the selected 8 points is calculated. The numbering of each point is shown in the figure; the corresponding relationship between the serial number of the 16APSK reference point and the data bit and carrier phase is shown in Table 4.

[0109] Table 4 Correspondence between the sequence number of 16APSK reference point and data bit and carrier phase

[0110]

[0111]

[0112] (203)16APSK four bits are represented as b3b2b1b0;

[0113] Calculate the log-likelihood ratio of bit position b0:

[0114]

[0115] Calculate the log-likelihood ratio of bit position b1:

[0116]

[0117] Calculate the log-likelihood ratio of bit position b2:

[0118]

[0119] Calculate the log-likelihood ratio for bit position b3:

[0120]

[0121] Among them, Q i is the square of the Euclidean distance between r1 and the constellation point numbered i among the 8 selected points;

[0122] (204) Determine the sign of the real and imaginary parts of the input signal r. If the real part of r is negative, then LLR1 = -LLR1; if the imaginary part of r is negative, then LLR0 = -LLR0.

[0123] This improved max-log scheme needs to calculate the square of 8 Euclidean distances. When multiplexing comparators, min{Q2,Q3,Q4,Q7} of LLR0 and LLR1 can share a set, and min{Q3,Q4} of LLR2 can use min{Q3,Q4} of min{Q2,Q3,Q4,Q7} of LLR0. Therefore, the number of comparators is 8.

[0124] In principle, the floating-point performance of this scheme should be exactly the same as that of the max-log scheme, and simulations have shown that the performance is exactly the same.

Claims

1. A soft information demodulation method for 8PSK and 16APSK in satellite communication, characterized in that: The process includes: The 8PSK processing process is as follows: (101) Rotate the 8PSK standard constellation diagram clockwise by π / 8; (102) Rotate the received signal r clockwise by π / 8 to obtain r1, r1 = r*e (-1j*π / 8) ; (103) Take the absolute value of the real and imaginary parts of the rotated point r1 and map it to the first quadrant to obtain r2: r2=abs(real(r1))+1j*abs(imag(r1)); (104) Select four points on the constellation diagram whose bits are 001, 000, 100 and 110 after clockwise rotation and number them 1, 2, 3, 4, and calculate the square of the Euclidean distance between r2 and the selected four points; (105)8PSK three bits are represented as b2b1b0; Calculate the log-likelihood ratio of bit position b0: Calculate the log-likelihood ratio of bit position b1: Calculate the log-likelihood ratio of bit position b2: Among them, Q i is the square of the Euclidean distance between r2 and the constellation point numbered i among the four selected points; (106) Determine the sign of the real and imaginary parts of point r1. If the real part of r1 is negative, then LLR1 = -LLR1; if the imaginary part of r1 is negative, then LLR0 = -LLR0; The 16APSK processing process is: (201) Take the absolute value of the real and imaginary parts of the received signal r and map it to the first quadrant to obtain r1: r1=abs(real(r))+1j*abs(imag(r)); (202) Select eight points with bits 0101, 0100, 0000, 1000, 1010, 1101, 1100 and 1110 on the 16APSK standard constellation diagram and number them from 1 to 8 in sequence, and calculate the square of the Euclidean distance between r1 and the selected 8 points; (203)16APSK four bits are represented as b3b2b1b0; Calculate the log-likelihood ratio of bit position b0: Calculate the log-likelihood ratio of bit position b1: Calculate the log-likelihood ratio of bit position b2: Calculate the log-likelihood ratio for bit position b3: Among them, Q i is the square of the Euclidean distance between r1 and the constellation point numbered i among the 8 selected points; (204) Determine the sign of the real and imaginary parts of the input signal r. If the real part of r is negative, then LLR1 = -LLR1; if the imaginary part of r is negative, then LLR0 = -LLR0.

Images