CN112636803A - Modified vector disturbance soft demodulation method - Google Patents

Abstract

The invention belongs to the technical field of wireless communication, and particularly relates to a design of a soft demodulation scheme of VP. In the method provided by the invention, the original LLR function is corrected, and the modulus part is considered in the function, so that the corrected function can meet the probability distribution requirement of data after the modulus is taken by a VP receiving end, and the output soft information is corrected. Compared with the method for directly carrying out hard decision and original soft decision on the VP, the scheme provided by the invention can provide good performance gain.

Description

Modified vector disturbance soft demodulation method

Technical Field

The invention belongs to the technical field of wireless communication, and particularly relates to a modified vector disturbance soft demodulation method.

Background

Recently, a new non-linear precoding technique, Vector Perturbation (VP) precoding, has been proposed, and the key idea is to select an additive Perturbation Vector to shape the transmitted symbol, and obtain the optimal Perturbation Vector through a search algorithm such as LRA. By the vector perturbation technology, the coding technology can achieve the performance close to the total capacity. On the other hand, soft information such as Log-Likelihood Ratio (LLR) is used in the soft decision technology, and soft information is output in combination with the related channel coding technology, so that the overall error code performance of the system is greatly improved compared with that of hard decision and soft decision.

However, the VP technique uses a modulo operation so that the data at the receiving end does not follow a gaussian distribution.

Disclosure of Invention

Aiming at the problems, the invention provides a new LLR function for soft decision of the VP to improve the overall error code performance of the VP system by restoring the original LLR.

For ease of understanding, the modified soft decision scheme employed by the present invention is illustrated as follows:

the VP technique uses a perturbation vector to perturb the original data, i.e., v ═ s + τ l, where v denotes the perturbed data, s denotes the sign vector, l denotes the perturbation vector, whose dimension is the same as s,

τ is a positive real number, usually chosen as τ 2| c-_maxAnd + delta, wherein the module value of the constellation point symbol with the maximum amplitude in the QAM constellation set represents the minimum Euclidean distance between the constellation point symbols. Assuming ZF precoding is used, i.e. the precoding matrix is

The transmitted power is normalized to 1, then the transmitted signal is

Wherein the content of the first and second substances,

then the received data is

The received data is multiplied by a power normalization factor, and then the disturbance vector is removed through the modular transportation to obtain

The modulo operation is defined as

This makes it possible to find that the distribution to which the noise is subjected at this time is a gaussian modulo distribution, and LLRs cannot be used directly as soft information.

The technical scheme of the invention is as follows:

a modified vector perturbation soft demodulation method, comprising the steps of:

s1, defining the vector of the sending symbols as S, one symbol corresponding to n bits b_n...b₂，b₁，b₀}；

S2, for the ith element S in the vector S_i，s_iCorresponding to the ith bit b_iLet b_iCorresponding value probability and s_iIs related to the in-phase component of, i.e.

m is b_iSymbol s corresponding to 0_iThe value of the in-phase component of (a);

s3, calculation b_iProbability of 0:

wherein M is c_iAll values of (a), (b), (c) and (d)_iIs composed of

X is

The real part of the ith element of (1), τ is a positive real number,

is a Gaussian variable with a mean value of 0 and a variance of 2 sigma, n_iThe ith element of n is expressed, and the modulus operation is defined as

For ln (P { x | b)_i0) weighted:

wherein l is the disturbance amount;

adding a limit to the value of M to make M + -tau E [ -tau, tau]Similarly, calculate b_iProbability of 1, get b_iSoft information of

The final log-likelihood ratio (LLR) is obtained as:

wherein M is₀，M₁Are respectively b_i0 and b_iC when 1 corresponds to_iAll values of (a);

s4, respectively obtaining b by adopting the method of the step S3_kAnd k is not equal to the soft information of i, and after the soft information of the whole code word is obtained, a decoder is used for decoding to obtain the estimation of the code word.

The method has the advantages that the original LLR function is corrected, the modulus part is considered in the function, so that the corrected function can meet the probability distribution requirement of data after the modulus of a VP receiving end is acquired, and the output soft information is corrected. Compared with the method for directly carrying out hard decision and original soft decision on VP, the scheme provided by the invention can provide good performance gain

Drawings

FIG. 1 is a block diagram of a VP soft decision system;

fig. 2 is a schematic diagram showing the bit error rate performance comparison of the hard decision, the original soft decision and the modified soft decision scheme when the number of the transmitting antennas is 4 and the number of users is 4, when BPSK modulation is adopted and 1/3-rate LDPC coding is used;

fig. 3 is a schematic diagram showing the error rate performance comparison of the hard decision, the original soft decision and the modified soft decision scheme when the number of the transmitting antennas is 4 and the number of users is 4, QPSK modulation is adopted, and 1/3-rate LDPC coding is used;

FIG. 4 is a schematic diagram showing the bit error rate performance comparison of the hard decision, the original soft decision and the modified soft decision schemes when the number of transmitting antennas is 4 and the number of users is 4, and when the 1/3 code rate LDPC coding is used and 16-QAM modulation is adopted;

Detailed Description

The technical scheme of the invention is described in detail in the following with reference to the accompanying drawings and embodiments:

the technical scheme of the invention mainly provides a new LLR function, and the specific derivation process is as follows:

for a transmitted symbol vector s, only one symbol s of s is considered_iRepresents the ith element in the vector s, while taking into account its corresponding ith bit b_i。b_iCorresponding value probability and s_iIs related to the in-phase or quadrature component of (1), provided that it is related to the in-phase component

m is b_iSymbol s corresponding to 0_iOf the in-phase component of (b), more specifically b_iThe probability of a value of 0 being equal to s_iThe probability that the corresponding component is m.

Thus, it is possible to provide

Representing a set of integers, wherein x is

The real part of the ith element of (1), c_iIs b is_iWhen equal to 0

The value of (a).

Will s_iIs substituted into the whole formula and the formula is divided into two parts, namely a part with k equal to 0 and the rest. Namely the above type

Wherein M is c_iAll values of (a).

Order to

By

Taking logarithm of the original formula to obtain the formula

The reduction of the molecule of the second term in the second additional term ln of the above formula is achieved by several methods:

a) taking the largest of all the terms of the sum, i.e. numerator equal to

k is an integer, k is not equal to 0, and the formula can be simplified to account for monotonicity of the exponential operation

b) Taking the sum of the maximum two of all the terms of the sum operation, the monotonicity numerator taking account of the exponential operation being equal to

c) And taking the sum of the maximum finite term in all the terms of the summation operation.

Scheme b) is used in the invention to obtain the formula

While taking into account

Wherein l is the disturbance amount; therefore, it can be used for ln (P { x | b)_i0) to get a more accurate LLR:

in order to reduce complexity, a limit needs to be added to the value of M so that M + -tau epsilon [ -tau, tau]Similarly, calculate b_iProbability of 1, get b_iSoft information of

The final log-likelihood ratio (LLR) is obtained as:

wherein M is₀，M₁Are respectively b _i0 and b_iC when 1 corresponds to_iAll values of (a).

Using the above functions, b is obtained_kAnd k is not equal to the soft information of i, and after the soft information of the whole code word is obtained, a decoder is used for decoding to obtain the estimation of the code word.

Examples

In this example, the number of transmitting antennas is 4, the number of users is 4, each user is configured with a receiving antenna, the modulation mode uses the normalized 16QAM modulation using gray code, one symbol of the modulation mode corresponds to 4 bits { b }₃，b₂，b₁，b₀}，

The channel coding uses convolutional codes and the decoding uses viterbi decoding.

S1, calculating b first₀Probability of 0, b₀Corresponds to s_iIs a same phase component of

That is to say M { -d, -3d }.

S2, calculating

The probability is weighted to obtain

Considering M value limit to obtain final LLR

Calculating b in the same way₀The probability of 1 is given as the probability of,

that is, M ═ d,3d, ln (P { x | b } is obtained₀＝1})。

S3, calculation b₀Soft information of

S4, obtaining b₃，b₂，b₁After the soft information of the whole code word is obtained, a decoder is used for decoding to obtain the estimation of the code word.

Claims (1)

1. A modified vector perturbation soft demodulation method, comprising the steps of:

s1, defining the vector of the sending symbols as S, one symbol corresponding to n bits b_n...b₂，b₁，b₀}；

S2, for the ith element S in the vector S_i，s_iCorresponding to the ith bit b_iLet b_iCorresponding value probability and s_iIs related to the in-phase component of, i.e.

m is b_iSymbol s corresponding to 0_iThe value of the in-phase component of (a);

s3, calculation b_iProbability of 0:

wherein M is c_iAll values of (a), (b), (c) and (d)_iIs b is_iWhen equal to 0

X is

The real part of the ith element of (1), τ is a positive real number,

is a Gaussian variable with a mean value of 0 and a variance of 2 sigma, n_iThe ith element of n is expressed, and the modulus operation is defined as

For ln (P { x | b)_i0) weighted:

wherein l is the disturbance amount;

adding a limit to the value of M to make M + -tau E [ -tau, tau]Similarly, calculate b_iProbability of 1, get b_iSoft information of

The final log-likelihood ratio (LLR) is obtained as:

wherein M is₀，M₁Are respectively b_i0 and b_iC when 1 corresponds to_iAll values of (a);

s4, respectively obtaining b by adopting the method of the step S3_kAnd k is not equal to the soft information of i, and after the soft information of the whole code word is obtained, a decoder is used for decoding to obtain the estimation of the code word.

Images