CN106059646A - Improved MIMO signal detection method - Google Patents

Abstract

The invention discloses an improved MIMO signal detection method. A ZF-OSIC (Zero Forcing Ordered Successive Interference Cancellation) algorithm is first employed to detect the receiving signal of a V-BLAST system to obtain a detection result of the ZF-OSIC algorithm. The detection result is low in reliability owing to the defect of the ZF-OSIC algorithm. Based on the detection result, the improved MIMO signal detection method employs an ML (Maximum Likelihood) rule to correct one or more than one element in the detection result. The number of correction elements is selected based on practical situations so as to meet demands for complexity and error rate from practical applications. The improved MIMO signal detection method keeps lower complexity, meanwhile obtains a lower error rate, and possesses greater adaptability.

Description

A kind of MIMO signal detection method of improvement

Technical field

The invention belongs to wireless communication technology field, the MIMO signal detection method of a kind of improvement.

Background technology

Multi-input multi-output system is called for short mimo system, and it utilizes space diversity to improve the availability of frequency spectrum, is forth generation The technology of great development prospect in mobile communication.MIMO technology improves the rate of information throughput and data reliability, and has relatively High energy efficiency.Owing to sending and receiving end antenna amount is more, no matter for tradition MIMO or extensive mimo system, receive letter Number detection is all good problem to study.

The reception signal detection of mimo system can be used Maximum Likelihood Detection (Maximum Likelihood, ML), compel Zero detection (Zero Forcing, ZF), counteracting serial interference (Successive Interference Cancellation, SIC), the method such as sphere decoding (sphere decoding, SD).Wherein, ML algorithm is obtained in that receiving diversity completely increases Benefit.ML algorithm is substantially a kind of enumeration, concentrates all possible signaling point by exhaustive modulation constellation, chooses and receive Signal is immediate as detection output, has higher complexity.If with the bit error rate for passing judgment on criterion, then ML is optimum calculation Method.ML algorithm errors is, along with antenna number and modulation constellation exponent number increase, its algorithm complex exponentially rises.ZF calculates Method is a kind of linear algorithm, obtains Signal estimation value with filtering matrix with receiving signal multiplication, and its complexity is relatively low, it is easy to real Existing.But owing to linear null matrix of compeling is non-orthogonal, in the case of low signal-to-noise ratio, it will amplifying noise, therefore this algorithm exists The bit error rate under low signal-to-noise ratio is higher.In order to improve the reliability of ZF algorithm, SIC method is suggested.SIC algorithm based on ZF or MMSE, be divided into ZF counteracting serial interference (Zero Forcing Successive Interference Cancellation, And least mean-square error (Minimum Mean Square Error Successive Interference ZF-SIC) Cancellation, MMSE-SIC).Utilize ZF or MMSE criterion to detect a certain transmission signal, examine from receiving signal is removed The signal measured, then carry out the detection of next signal.In view of the signal detection order impact on signal detecting result, it is proposed that Serial interference elimination (Ordered Successive Interference Cancellation, OSIC) based on sequence is calculated Method.Owing to ZF-OSIC algorithm is wanted first to detect a certain transmission signal by ZF algorithm, also with ZF filtering matrix, so ZF algorithm low reliability originally is not improved, and the bit error rate of ZF-OSIC algorithm still needs to reduce further.

From the foregoing, it will be observed that ML detection algorithm complexity is high, actual application is difficult to.Though ZF-OSIC complexity is relatively low, But degree of accuracy is the lowest, its range of this drawbacks limit.

Summary of the invention

Technical problem solved by the invention is to provide the MIMO signal detection method of a kind of improvement.

The technical solution realizing the object of the invention is: the MIMO signal detection method of a kind of improvement, including following step Rapid:

Step 1, the reception antenna of V-BLAST system receive signal, and transmit a signal to signal processing module；

Step 2, signal processing module use ZF-OSIC algorithm to detect the signal received, and obtain ZF-OSIC algorithm Testing result；

Step 3, ZF-OSIC algorithm testing result is modified, obtains revised detection output, complete pair signals Detection.

Compared with prior art, its remarkable advantage is the present invention: 1) present invention ties with the detection of ML criterion modification ZF-OSIC Really, compare with ZF-OSIC detection, substantially increase the reliability of testing result；2) present invention is with ZF-OSIC testing result as base Plinth is optimized, and compares with ML detection, greatly reduces algorithm complex；3) strong adaptability of the present invention, the modifying factor of the present invention Sub-K can regard the weighting factor between complexity and reliability as, can flexible value in [1, M] as the case may be.Not Lot of complexity can be born but allow the application scenarios of the bigger bit error rate, optional less K value；To bit error rate requirement high but The application scenario of lot of complexity can be born, optional bigger K value.

Below in conjunction with the accompanying drawings the present invention is described in further detail.

Accompanying drawing explanation

Fig. 1 is the system composition diagram of the embodiment of the present invention.

Fig. 2 is the MIMO signal detection method flow chart improved.

Fig. 3 is that the bit error rate of the embodiment of the present invention is with signal to noise ratio change curve.

Detailed description of the invention

In conjunction with accompanying drawing, the MIMO signal detection method of a modification of the present invention, comprise the following steps:

Step 1, the reception antenna of V-BLAST system receive signal, and transmit a signal to signal processing module；

Step 2, signal processing module use ZF-OSIC algorithm to detect the signal received, and obtain ZF-OSIC algorithm Testing result；Use ZF-OSIC algorithm to receive signal detect particularly as follows:

The row l of row Norm minimum in step 2-1, searching V-BLAST system channel group inverse matrices matrix₁, public affairs used Formula is:

$l_i=\underset{j}{\arg \min }\left|\right|{\left(G_i\right)}_j|{|}^2$

In formula, G_iFor the generalized inverse matrix of the channel matrix after processing, G₁=H⁺, H is the channel matrix of V-BLAST system, J is line number, and i is iterations, and initial value is 1；

Step 2-2, determine l_iThe secondary Signal estimation value detectedFormula used is:

$r_{l_i}=w_{l_i}{y}_i$

In formula, y_iSignal to be detected before detecting for i & lt, as i=1, y₁Reception signal for V-BLAST system；

Step 2-3, i & lt detect before signal y to be detected_iIn, delete the signal that i & lt detects, concrete formula For:

$y_{i+1}=y_i-{\hat{a}}_{l_i}{\left(H\right)}_{l_i}$

In formula,L for the channel matrix of V-BLAST system_iRow,The Signal estimation value detected for i & lt Carry out the result after quantifying judgement,Q (*) is quantization function；

Step 2-4, by l in channel matrix_iRow are deleted, and to the channel matrix after processing, ask its generalized inverse, return afterwards Step 2-1 carries out next iteration, and till i=M, M is the transmission antenna number of V-BLAST system.

Step 3, ZF-OSIC algorithm testing result is modified, obtains revised detection output, complete pair signals Detection.It is modified concretely comprising the following steps to ZF-OSIC algorithm testing result:

Step 3-1, determining that modifying factor K, K represent positive element number to be repaired, its span is [1, M]；

Step 3-2, determining the choosing method kind number T choosing K element from M element, formula used is:

$T=C_M^K$

Step 3-3, with modulation constellation concentrate s constellation point replace K the element chosen respectively in exhaustive mode, obtain s^KIndividual new vector, and calculate its distance matrix respectively, formula used is:

$D_q^p=\left|\right|y-H{\tilde{x}}_{ZF-OSIC}^p|{|}^2$

In formula, y is the reception signal of V-BLAST system, and H is the channel matrix of V-BLAST system, and q is from M element The q kind choosing K element is followed the example of, and q is iterations simultaneously, and initial value is 1；For choosing K element from M element Q kind follows the example of the distance matrix of the new vector of middle pth,For s constellation point concentrating with modulation constellation in exhaustive mode Replace the new vector of pth of gained after K the element chosen, p ∈ [1, s respectively^K]；

Step 3-4, return step 3-3 carry out next iteration, till q=T；

Step 3-5, above-mentioned iteration result is judged, choose minimumCorresponding vector is as finally detecting knot Really.

Below in conjunction with embodiment the present invention done further detailed description:

Embodiment

Whole system forms such as Fig. 1.System uses 4 the 4 V-BLAST systems received.Data bit flow a length of 8 × 10⁶, Modulation system is 4QAM.Assuming that transmission channel is flat Rayleigh fading channel, channel condition information is sending with receiving terminal the most Know, and receiving terminal has realized timing and sign synchronization.

First the signal received system reception antenna carries out ZF-OSIC detection, concretely comprises the following steps:

The row l of row Norm minimum in step 1, searching V-BLAST system channel group inverse matrices matrix₁, formula used For:

$l_i=\underset{j}{\arg \min }\left|\right|{\left(G_i\right)}_j|{|}^2$

In formula, G_iFor the generalized inverse matrix of the channel matrix after processing, G₁=H⁺, H is the channel matrix of V-BLAST system, J is line number, and i is iterations, and initial value is 1；

Step 2, determine l_iThe secondary Signal estimation value detectedFormula used is:

$r_{l_i}=w_{l_i}{y}_i$

In formula, y_iSignal to be detected before detecting for i & lt, as i=1, y₁Reception signal for V-BLAST system；

Step 3, i & lt detect before signal y to be detected_iIn, delete the signal that i & lt detects, concrete formula is:

$y_{i+1}=y_i-{\hat{a}}_{l_i}{\left(H\right)}_{l_i}$

In formula,L for the channel matrix of V-BLAST system_iRow,The Signal estimation value detected for i & lt Carry out the result after quantifying judgement,

Step 4, by l in channel matrix_iRow are deleted, and to the channel matrix after processing, ask its generalized inverse, return step afterwards Rapid 1 carries out next iteration, till i=4.

But ZF filtering matrix G is non-orthogonal, also it is exaggerated noise with receiving while signal y is multiplied.Therefore, originally Invention is after ZF-OSIC, and testing result is modified by application ML criterion.Because transmission antenna number M is 4, so modifying factor Sub-K value is [1,4].The present invention is as a example by modifying factor K is 1 and 2, if m_bRepresentThe b element, then [m can be expressed as₁,m₂,m₃,m₄]^T.To testing resultThat revises concretely comprises the following steps:

(1) during K=1

Step 1, selection K are 1.

Step 2, determine the choosing method kind number T choosing 1 element from 4 elements.Formula used is:

$T=C_4^1=4$

Step 3,4 constellation point (-1-i) concentrated with 4-QAM modulation constellation, (-1+i), (1-i), (1+i) is with exhaustive side Formula replaces 1 element chosen respectively, obtains s^KIndividual new vector, and calculate its distance matrix respectively.Formula used is:

$D_q^p=\left|\right|y-H{\tilde{x}}_{ZF-OSIC}^p|{|}^2$

In formula, y is the reception signal of V-BLAST system, and H is the channel matrix of V-BLAST system.Q is from 4 elements The q kind choosing 1 element is followed the example of.Q is iterations simultaneously, and initial value is 1.For choosing K element from M element Q kind follow the example of the distance matrix of the new vector of middle pth.For 4 constellation point concentrating with modulation constellation with exhaustive side Formula replaces the new vector of pth of gained after 1 element chosen, p ∈ [Isosorbide-5-Nitrae] respectively.

Assume q=1 to represent from 4 elements to choose first element to be modified.Then Then

Step 4, return step 3 carry out next iteration, till q=4.

Step 5, choose minimumCorresponding vector is as final detection result.

(2) during K=2

Step 1, determine that K is 2.

Step 2, determine the choosing method kind number T choosing 2 elements from 4 elements.Formula used is:

$T=C_4^2=6$

Step 3,4 constellation point (-1-i) concentrated with 4-QAM modulation constellation, (-1+i), (1-i), (1+i) is with exhaustive side Formula replaces 2 elements chosen respectively, obtains 4²=16 new vectors, and calculate its distance matrix respectively.Formula used is:

$D_q^p=\left|\right|y-H{\tilde{x}}_{ZF-OSIC}^p|{|}^2$

In formula, y is the reception signal of V-BLAST system, and H is the channel matrix of V-BLAST system.Q is from 4 elements The q kind choosing 1 element is followed the example of.Q is iterations simultaneously, and initial value is 1.For choosing 2 elements from 4 elements Q kind follow the example of the distance matrix of the new vector of middle pth.For 4 constellation point concentrating with modulation constellation with exhaustive side Formula replaces the new vector of pth of gained after 2 elements chosen, p ∈ [1,16] respectively.

Assume q=1 to represent from 4 elements to choose first and second elements to be modified.Then Then

Step 4, return step 3 carry out next iteration, till q=6.

Step 5, choose minimumCorresponding vector final detection result the most.

By that analogy, as K=3 and 4, detection process is similar to the above.

Fig. 3 is that the bit error rate of several signal detecting method is with the change curve of average signal-to-noise ratio at reception antenna.In figure, ZF-OSIC 1c and ZF-OSIC 2c represents the situation that modifying factor K is 1 and 2 respectively.It can be seen that ML detection be reliability High, the present invention improves about 1dB when K=1 than ZF-OSIC performance, improves about 3dB than ZF-OSIC during K=2.It is visible, The present invention achieves the preferable performance less than the ZF-OSIC algorithm bit error rate, and the bit error rate of K=2 is than bit error rate during K=1 also Decrease.Modifying factor K is the biggest, and the figure place of correction is the most, and testing result will be the most accurate, and performance curve will be calculated closer to ML Method.

From complexity and Performance Simulation Results it can be seen that the present invention increases to cost with the least complexity, achieve Less than the bit error rate of ZF-OSIC detection, performance is improved, and is a kind of well MIMO signal detection method.

Claims (3)

1. the MIMO signal detection method improved, it is characterised in that comprise the following steps:

Step 1, the reception antenna of V-BLAST system receive signal, and transmit a signal to signal processing module；

Step 2, signal processing module use ZF-OSIC algorithm to detect the signal received, and obtain the detection of ZF-OSIC algorithm Result；

Step 3, ZF-OSIC algorithm testing result is modified, obtains revised detection output, the inspection of complete pair signals Survey.

The MIMO signal detection method of improvement the most according to claim 1, it is characterised in that use ZF-OSIC in step 2 Algorithm to receive signal detect particularly as follows:

The row l of row Norm minimum in step 2-1, searching V-BLAST system channel group inverse matrices matrix₁, formula used is:

$l_i=\underset{j}{\mathrm{argmin}}\left|\right|{\left(G_i\right)}_j|{|}^2$

In formula, G_iFor the generalized inverse matrix of the channel matrix after processing, G₁=H⁺, H is the channel matrix of V-BLAST system, and j is Line number, i is iterations, and initial value is 1；

Step 2-2, determine l_iThe secondary Signal estimation value detectedFormula used is:

$r_{l_i}=w_{l_i}{y}_i$

In formula, y_iSignal to be detected before detecting for i & lt, as i=1, y₁Reception signal for V-BLAST system；

Step 2-3, i & lt detect before signal y to be detected_iIn, delete the signal that i & lt detects, concrete formula is:

$y_{i+1}=y_i-{\hat{a}}_{l_i}{\left(H\right)}_{l_i}$

In formula,L for the channel matrix of V-BLAST system_iRow,The Signal estimation value detected for i & ltCarry out Quantify the result after judgement,Q (*) is quantization function；

Step 2-4, by l in channel matrix_iRow are deleted, and to the channel matrix after processing, ask its generalized inverse, return step afterwards 2-1 carries out next iteration, and till i=M, M is the transmission antenna number of V-BLAST system.

The MIMO signal detection method of improvement the most according to claim 1, it is characterised in that in step 3, ZF-OSIC is calculated Method testing result is modified concretely comprising the following steps:

Step 3-1, determining that modifying factor K, K represent positive element number to be repaired, its span is [1, M]；

Step 3-2, determining the choosing method kind number T choosing K element from M element, formula used is:

$T=C_M^K$

Step 3-3, with modulation constellation concentrate s constellation point replace K the element chosen respectively in exhaustive mode, obtain s^KIndividual New vector, and calculate its distance matrix respectively, formula used is:

$D_q^p=\left|\right|y-H{\tilde{x}}_{ZF-OSIC}^p|{|}^2$

In formula, y is the reception signal of V-BLAST system, and H is the channel matrix of V-BLAST system, and q is for choose from M element The q kind of K element is followed the example of, and q is iterations simultaneously, and initial value is 1；For choosing the q of K element from M element Plant the distance matrix following the example of the new vector of middle pth,S the constellation point for concentrating with modulation constellation is distinguished in exhaustive mode The new vector of the pth of gained, p ∈ [1, s after K the element that replacement is chosen^K]；

Step 3-4, return step 3-3 carry out next iteration, till q=T；

Step 3-5, above-mentioned iteration result is judged, choose minimumCorresponding vector is as final detection result.