CN101980467B - Retransmitting linear precoding processing method for multi-input multi-output system - Google Patents

Abstract

The invention discloses a retransmitting linear precoding processing method for a multi-input multi-output system. The method is characterized by comprising the following steps of: distributing power to sub-channels of which the signal to interference and noise ratio is smaller than a decoding threshold against each sub-channel pairing matrix in the multi-codon supported MIMO system retransmission precoding processing process; performing power distribution on residual power according to a target of minimizing the packet error rate to calculate the minimum packet error rate relative to the sub-channel pairing matrix; and traversing all sub-channel pairing matrixes, selecting the sub-channel pairing matrix corresponding to the minimum value in the minimum packet error rate as an optimal sub-channel pairing matrix, and distributing power to each sub-channel according to the optimal sub-channel pairing matrix. By adopting the method, the packet error rate lower than that of the traditional method can be achieved by reasonably pairing sub-channels and distributing power according to the adaptive modulation coding level selected by each codon.

Description

A kind of multi-input multi-output system retransmits the linear predictive coding processing method

Technical field

The invention belongs to multiple-input and multiple-output (MIMO) mobile communication technology field, particularly technology of the automatic request retransmission (ARQ) in the MIMO communication system and linear predictive coding technology.

Background technology

Multiple-input and multiple-output (MIMO) technology can produce space diversity and spatial multiplexing gain, is the important technology that adopts in the 4th third-generation mobile communication system.The linear predictive coding technology is to realize the important means of MIMO technical advantage, and the design of traditional linear predictive coding mostly is based on the design under the packet single transmission situation.And in automatic request retransmission (ARQ) mechanism that generally adopts in the Modern Communication System error data packets is retransmitted repeatedly.So the design of the linear predictive coding when packet retransmits is a major issue.

" acoustics, voice and signal processing international conference " (IEEE International Conference on Acoustics; Speech; And Signal Processing Proceedings.Volume:3; Pp:iii/425-iii/428,2005) disclosed a kind of MIMO ARQ under desirable channel information situation retransmits the design that the linear predictive coding method for designing is based on maximize capacity; Re-transmission precoding method for designing under a kind of MIMO ARQ system non-ideal communication channel estimation condition that " communication journal " (28th volume the 9th phase 7-14 page or leaf) mentioned is based on the suboptimum design of maximize capacity lower bound.More than two kinds of methods all can give the good relatively subchannel of this channel conditions of subchannel pairing of previous comprehensive channel situation relative mistake and distribute more power.For the mimo system of supporting many codeword transmission; Because each code word can independently be carried out Adaptive Modulation and Coding; The protective capability that each code word requires is relevant with selected Adaptive Modulation and Coding grade, so two kinds recited above retransmit recoding processing methods and just can't reach optimum in the mimo system that is applied to the many codeword transmission of support the time.

Summary of the invention

The present invention proposes a kind of mimo system of supporting many codeword transmission and retransmits recoding processing method, can reasonably match subchannel and distribute power according to the Adaptive Modulation and Coding grade that each code word is selected, and reaches lower Packet Error Ratio performance.

Multi-input multi-output system of the present invention retransmits the linear predictive coding processing method; Comprise: adopt a plurality of code words to transmit at transmitting terminal; When a packet transmitted for the first time, each code word was independently added CRC, and independently carried out Adaptive Modulation and Coding; When packet retransmitted, the first information transmitted of information and date bag that transmitting terminal retransmits was consistent; The transmitting terminal elder generation premultiplication linear predictive coding matrix data of redispatching, the method that the linear predictive coding matrix during the packet initial transmission is poured water through channel singular value decomposition and power obtains; At receiving terminal, elder generation is the associate matrix of the left unitary matrix of premultiplication channel singular value decomposition to received signal, then each code word is carried out the demodulating and decoding operation of receiving terminal: if the transmission first time of a packet is then directly decoded to each code word; If retransmit, then earlier the data of each code word are carried out high specific and merge, then decoding; Have only when all code words all separate to the time, receiving terminal is just successfully indicated to the decoding that transmitting terminal feeds back a bit, otherwise the decoding failure indication of feeding back a bit, transmitting terminal carries out retransmission operation when receiving decoding failure indication;

If mimo system transmitting terminal antenna number is M _T, receiving terminal antenna number M _RMore than or equal to the transmitting terminal antenna number, the information symbol behind the Adaptive Modulation and Coding is designated as S wherein _i(i=1 ..., M _T) i code word of expression, the ratio of signal power and noise power is γ;

Error code word rate performance curve with each Adaptive Modulation and Coding grade in error code word rate model formation (1) the match system:

$f_i\left(\mathrm{SINR}\right)=\left\{\begin{array}{ccc}1& \mathrm{if}& 0<\mathrm{SINR}<{\mathrm{\&gamma;}}_{\mathrm{th}}^i\\ a_i\exp (-g_i*\mathrm{SINR})& \mathrm{if}& \mathrm{SINR}>{\mathrm{\&gamma;}}_{\mathrm{th}}^i\end{array}\right.---\left(1\right)$

Wherein SINR is for receiving the Signal Interference and Noise Ratio of code word, a _iAnd g _iBe respectively multiplier constant and exponential constant that the corresponding error code word rate match of Adaptive Modulation and Coding grade that i code word select obtains,

It is the corresponding decoding threshold of Adaptive Modulation and Coding grade that i code word selected;

The channel that system estimation was come out when if packet the m time retransmitted for

then channel estimation errors modeling formula do

$E_m=H_m-{\hat{H}}_m---\left(2\right)$

H wherein _mBe the mimo channel of reality, m equals the initial transmission of 0 expression packet, and m is greater than the re-transmission of 0 expression packet; Error matrix E _mEach element obey that average is 0, variance does

Multiple Gaussian distribution, if desirable channel information, then variance

Be 0;

The channel

that system estimation went out when packet was retransmitted for the m time carries out singular value decomposition and gets

${\hat{H}}_m=U_m{\mathrm{\&Lambda;}}_m^{1/2}{V}_m^H---\left(3\right)$

Equal 0 for m, that is, packet transmits for the first time, and the linear predictive coding matrix does

$F_0=V_0{\tilde{F}}_0^H---\left(4\right)$

Wherein the power division matrix does

0 _{I, i}The element value of the capable i row of the i of the matrix in the expression bracket, () ₊That bigger value in value and 0 two numerical value in the bracket, the algorithm horizontal plane parameter c of pouring water are got in expression ₀Selection need satisfy the normalized power restrictive condition

It is characterized in that: greater than 0, when promptly retransmitting for the m time, the linear predictive coding expression matrix does for m

$F_m=V_m{P}^H{\tilde{F}}_m^H---\left(5\right)$

Wherein permutation matrix P is a subchannel pairing matrix, and power division matrix

satisfies normalized Power Limitation condition

The reception Signal Interference and Noise Ratio of i code word was designated as when the j time was retransmitted

The total reception Signal to Interference plus Noise Ratio that then retransmits i code word after merging for the m time does

The previous reception error probability that defines i code word does $f_i^{\mathrm{Previous}}=f_i\left({\mathrm{\&gamma;}}_{G,i}^{m-1}\right);$

Support the mimo system re-transmission linear predictive coding matrix of many code words to obtain in order to following method:

Initialization: the optimum permutation matrix P of initialization _GBe unit matrix, that is, and P _G=I, overall minimum Packet Error Ratio

The set of all permutation matrixes of mark is S;

Step 1:

If the permutation matrix set then jumps to step 2 for empty set; Otherwise, from the permutation matrix S set, select a permutation matrix to give permutation matrix P its assignment arbitrarily; Diagonal matrix after order is replaced through permutation matrix P does

${\widetilde{\mathrm{\&Lambda;}}}^m={\mathrm{P\&Lambda;}}_m{P}^H/(1+{\mathrm{\&sigma;}}_E^2\mathrm{\&gamma;})---\left(6\right)$

Give each code word preassignment power

$P_i^m=\max (({\mathrm{\&gamma;}}_{\mathrm{th}}^i-{\mathrm{\&gamma;}}_{G,i}^{m-1})/\left(\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i}\right),0)---\left(7\right)$

If

then from the permutation matrix S set, get rid of permutation matrix P jumps to step 1;

If the remaining power According to Lagrange power allocation formula (8)

${\tilde{F}}_i^m={(-\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i}}{c}_m+\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i}}\ln f_i^{\mathrm{previous}}{g}_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i})}_{+}---\left(8\right)$

The Power Limitation condition is satisfied in searching Parameter c _m, according to minimum Packet Error Ratio formula (9)

${\mathrm{per}}_{\min }^P=\underset{i=1}{\overset{M_T}{\mathrm{\&Sigma;}}}\min (\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i}}{e}^{c_m},f_i^{\mathrm{previous}})---\left(9\right)$

Obtain the minimum Packet Error Ratio

of corresponding permutation matrix P

If the minimum Packet Error Ratio of corresponding permutation matrix P is less than the minimum Packet Error Ratio of the overall situation, that is,

Then upgrading overall minimum Packet Error Ratio is the corresponding minimum Packet Error Ratio of permutation matrix P, that is, Upgrade optimum permutation matrix P _GBe permutation matrix P, that is, and P _G=P gets rid of permutation matrix P from the permutation matrix S set, jump to step 1 then;

Step 2:

Permutation matrix P according to optimum _G, carry out power division;

Order is through optimum permutation matrix P _GDiagonal matrix after the displacement does

Give each code word (7) preassignment power by formula

$P_i^m=\max (({\mathrm{\&gamma;}}_{\mathrm{th}}^i-{\mathrm{\&gamma;}}_{G,i}^{m-1})/\left(\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^m\right)}_{i,i}\right),0)$

If

is the i-th subchannel allocated power

jump to step three;

If

Dump power then

The dump power of each code assignment is shown in the formula (8), that is,

Horizontal plane parameter c wherein _mSelection must satisfy the Power Limitation condition

Upgrade power division matrix element value

Step 3:

Obtaining the power division matrix does

Retransmitting the linear predictive coding matrix does $F_m=V_m{P}_G^H{\tilde{F}}_m^H;$

When retransmitting for the m time, transmitting terminal carries out the precoding processing operation to original code word information s: take advantage of pre-coding matrix on the original code word information left side, use F _mS sends signal.

Conventional retransmission linear predictive coding method for designing can be given the good relatively subchannel of subchannel this channel conditions of pairing of previous comprehensive channel situation relative mistake; Form the reverse pair relationhip of previous comprehensive channel situation and current retransfer channel situation, and can pour water the algorithm more power to the good subchannel of this channel conditions by distribution ratio independent design power.The target of conventional method is a maximize capacity; But for the mimo system of supporting many code words; The Adaptive Modulation rank difference that many code words of particularly supporting are selected is apart from bigger situation; It is very high that traditional method may support that the protective capability of the subchannel of low Adaptive Modulation and Coding grade is carried to certain, and cause the ability protection of the subchannel of supporting higher Adaptive Modulation and Coding grade inadequately, so just can't reach lower re-transmission Packet Error Ratio.Compare with traditional method; The present invention retransmits in the precoding processing process at the mimo system of supporting many code words; Traveled through all subchannel pairing matrixes, and to each subchannel pairing matrix, the present invention distributes earlier power not reach the subchannel of decoding threshold to Signal to Interference plus Noise Ratio; The reason of doing like this is not reach decoding threshold if there is subchannel to receive later merging Signal to Interference plus Noise Ratio, then the packet of this reception necessarily can't separate right; The subchannel that the present invention does not reach decoding threshold to signal to noise ratio carries out again remaining power being carried out power division according to the target that minimizes Packet Error Ratio after the preassignment power, tries to achieve the minimum Packet Error Ratio of this subchannel pairing matrix relatively then; The present invention can travel through all subchannel pairing matrixes, selects the corresponding subchannel of minimum value in the minimum Packet Error Ratio to match matrix as optimum subchannel pairing matrix, distributes power for each subchannel according to the subchannel pairing matrix of optimum then.The Adaptive Modulation and Coding grade that the present invention selects according to each code word is reasonably matched subchannel and optimum distribution power; Adaptive Modulation and Coding rank difference distance regardless of a plurality of code words selections; Adopt method of the present invention can match optimum channel; Distribute optimum power to give each channel, reach the Packet Error Ratio performance lower than conventional method.

Description of drawings

Performance curve when accompanying drawing 1 all uses the Adaptive Modulation and Coding class 5 for following two code words of ideal communication channel information state.

Performance curve when accompanying drawing 2 uses Adaptive Modulation and Coding class 5 and grade 1 respectively for following two code words of ideal communication channel information state.

Performance curve when accompanying drawing 3 uses Adaptive Modulation and Coding class 5 and MCS grade 1 respectively for following two code words of non-ideal communication channel information state.

Embodiment

Embodiment below in conjunction with description of drawings this method.

Embodiment 1:

Use the Turbo encoding and decoding structure in Long Term Evolution (LTE) technical standard the 8th edition of third generation communication technology cooperative programme (3GPP) in the present embodiment, five kinds of Adaptive Modulation and Coding (MCS) grade of introducing is shown in first three columns in the table 1.The later data length of these five kinds of MCS grade rate-matched is 4008 bits all, the error code word rate performance curve of the various MCS grades of emulation under the Gaussian channel scene at first in the present embodiment, and according to error code word rate model formation (1), promptly

$f_i\left(\mathrm{SINR}\right)=\left\{\begin{array}{ccc}1& \mathrm{if}& 0<\mathrm{SINR}<{\mathrm{\&gamma;}}_{\mathrm{th}}^i\\ a_i\exp (-g_i*\mathrm{SINR})& \mathrm{if}& \mathrm{SINR}>{\mathrm{\&gamma;}}_{\mathrm{th}}^i\end{array}\right.$

The performance curve of the various MCS grades of match, the parameter of match is shown in three row of back in the table 1.

The various Adaptive Modulation and Coding grade of table 1 fitting parameter table

The MCS grade	Modulation system	Code check	The multiplier constant	Exponential constant	Decoding threshold (dB)
						MCS1	QPSK	1/2	9.9771*10 26	43.009	1.60
MCS2	QPSK	3/4	1.8254*10 9	8.1076	4.20
						MCS3	16QAM	9/16	6.3022*10 16	6.7222	7.60
MCS4	16QAM	3/4	5.7952*10 11	2.4702	10.40
						MCS5	64QAM	3/4	4.1937*10 11	0.7371	15.60

The multi-input multi-output system antenna configurations is transmitting terminal antenna number M in the present embodiment _T=2, receiving terminal antenna number M _R=2, system supports the transmission of two code words, that is,

The Adaptive Modulation and Coding grade of two code word uses all is the MCS5 in the table 1; Each time of packet channel transmitted is that average is 0, and variance is 1 independent same distribution Rayleigh channel, and channel information is desirable, that is, and and the error matrix element variance Actual mimo channel is identical with the mimo channel of system estimation

The tentation data bag retransmits once.

Multi-input multi-output system of the present invention retransmits linear predictive coding processing method and traditional " acoustics, voice and signal processing international conference " (IEEE International Conference on Acoustics when the packet initial transmission; Speech; AndSignal Processing Proceedings.Volume:3; Pp:iii/425-iii/428,2005) the neutral line recoding processing method is identical, all adopts singular value decomposition and the power algorithm of pouring water to obtain.Channel singular value decomposition process is formula (3); That is,

power process of pouring water does

${\tilde{F}}_i^0={(c_0-\frac{1}{\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_0\right)}_{i,i}})}_{+},i=\mathrm{1,2}$

Wherein 0 _{I, i}The element value of the capable i row of the i of the matrix in the expression bracket, c ₀Selection need satisfy the Power Limitation condition

Initial transmission power allocation matrix then

The linear predictive coding of initial transmission is shown in the formula (4), that is,

When a packet transmitted for the first time, transmitting terminal carried out the precoding processing operation to sending signal, sends signal F ₀S.Receiving terminal at first multiply by

on the left side that receives signal and respectively each code word is decoded then during initial transmission; Having only two code words all to separate decode and successfully indicates just feeding back a bit to transmitting terminal; Otherwise feed back bit decoding failure indication, transmitting terminal receives decoding when successfully indicating, and continues to send new data; When transmitting terminal receives decoding failure indication, data retransmission s.

For the linear predictive coding design of the retransmission processes first time, the linear predictive coding enclosed that conventional method obtains is expressed as

$F_1^{\mathrm{Trad}}=V_1{P}_1^H\mathrm{diag}(\sqrt{{\mathrm{\&lambda;}}_1},\sqrt{{\mathrm{\&lambda;}}_2})P_0$

${\mathrm{\&lambda;}}_i={(\mathrm{\&mu;}-\frac{1+\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_0^{\mathrm{Trad}}\right)}_{i,i}}{\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_1^{\mathrm{Trad}}\right)}_{i,i}})}_{+},i=\mathrm{1,2}---\left(10\right)$

Wherein conventional method retransmits the sending direction matrix V for the first time ₁The unitary matrix that obtains for the singular value decomposition of channel

Permutation matrix P ₀Make the matrix after the initial transmission singular value is replaced Diagonal element by from big to small sequence arrangement, permutation matrix P ₁Make the matrix that retransmits for the first time after singular value is replaced

Diagonal element by from small to large sequence arrangement, the value of conventional method horizontal plane parameter μ need satisfy Power Limitation condition λ ₁+ λ ₂=1.When conventional method retransmits; Receiving terminal at first receiving the data that

multiply by on the left side before to have been received each code word high specific merging respectively then of signal, is decoded then.

For the design of linear predictive coding of the retransmission processes first time, the present invention is according to following method design linear predictive coding:

To channel by formula (3) carry out singular value decomposition, that is,

The Signal to Interference plus Noise Ratio of i code word of initial transmission does

The corresponding fitting parameter of each code word is a ₁=a ₂=4.1937*10 ¹¹, g ₁=g ₂=0.7371, Previous error probability does

Obtain pre-coding matrix according to following method:

Initialization: the optimum permutation matrix of initialization is a unit matrix, that is, and and P _G=I, overall minimum Packet Error Ratio

The set of all permutation matrixes of mark does

Step 1:

If the permutation matrix set is an empty set, then jump to step 2, otherwise, from the permutation matrix S set, select a permutation matrix to give permutation matrix P its assignment arbitrarily, the diagonal matrix after replacing through permutation matrix P by formula (6) obtains, that is, ${\widetilde{\mathrm{\&Lambda;}}}^1={\mathrm{P\&Lambda;}}_1{P}^H;$

Give each code word (7) preassignment power by formula; That is,

If

then from permutation matrix set s, get rid of permutation matrix P jumps to step 1;

If

the surplus power is

according to Lagrange power allocation formula (8)

${\tilde{F}}_i^1={(-\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{c}_1+\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}\ln f_i^{\mathrm{previous}}{g}_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i})}_{+}$

The Power Limitation condition is satisfied in searching

Parameter c ₁, according to minimum Packet Error Ratio formula (9)

${\mathrm{per}}_{\min }^P=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\min (\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{e}^{c_1},f_i^{\mathrm{previous}})$

Obtain the minimum Packet Error Ratio

of corresponding permutation matrix P

If the minimum Packet Error Ratio of corresponding permutation matrix P is less than the minimum Packet Error Ratio of the overall situation, that is,

Then upgrading overall minimum Packet Error Ratio is the corresponding minimum Packet Error Ratio of permutation matrix P, that is,

Upgrade optimum permutation matrix P _GBe permutation matrix P, that is, and P _G=P gets rid of permutation matrix P from permutation matrix set s, jump to step 1 then;

Step 2:

Permutation matrix P according to optimum _G, carry out power division; Order is through optimum permutation matrix P _GDiagonal matrix after the displacement does

Give each code word (7) preassignment power by formula, that is,

If

is the i-th subchannel allocated power

jump to step three;

If

then each code word of dump power

by formula (8) are distributed dump power; Promptly

${\tilde{F}}_i^1={(-\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{c}_1+\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}\ln f_i^{\mathrm{previous}}{g}_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i})}_{+}$

Horizontal plane parameter c wherein ₁Selection must satisfy the Power Limitation condition

Upgrade power division matrix element value

Step 3:

Get the power allocation matrix is

retransmission linear pre-coding matrix is

Transmitting terminal uses F when retransmitting for the first time ₁The s re-transmitted signal.Receiving terminal at first receiving the data that

multiply by on the left side before to have been received each code word high specific merging respectively then of signal, is decoded then.

Fig. 1 is that two code words are all used the Adaptive Modulation and Coding class 5, under desirable channel information situation, retransmits the Packet Error Ratio performance comparison that once later multi-input multi-output system of the present invention retransmits linear predictive coding processing method and conventional retransmission linear predictive coding processing method.Wherein solid line A is the Packet Error Ratio performance curve of conventional retransmission linear predictive coding processing method, and the dotted line B of solid line A below retransmits the Packet Error Ratio performance curve of linear predictive coding processing method for multi-input multi-output system of the present invention.Under the as can be seen from Figure 1 identical state of signal-to-noise, multi-input multi-output system of the present invention retransmits the linear predictive coding processing method can reach littler Packet Error Ratio performance.Conventional retransmission linear predictive coding processing method is always oppositely matched channel when retransmitting; Promptly; Channel conditions subchannel preferably when the pairing of the relatively poor subchannel of initial transmission channel conditions retransmits, the relatively poor subchannel of channel conditions when the subchannel pairing preferably of initial transmission channel conditions retransmits; And subchannel more power preferably when conventional method can be distributed to re-transmission according to subchannel displacement and power division formula (10).The adaptive coding and modulating grade of two code word employings is identical in the present embodiment, and just two code words need same protective capability ability correct decoding.And conventional retransmission linear predictive coding processing method can't make the protective capability of two code words reach identical.Multi-input multi-output system of the present invention retransmits the linear predictive coding processing method and can match and power distribution strategies according to the require adaptive adjusting subchannel of each code word to protective capability, adopts the inventive method can reach littler Packet Error Ratio performance.Be 0.01 o'clock at Packet Error Ratio as can be seen from Figure 1, multi-input multi-output system of the present invention retransmits the linear predictive coding processing method has the gain about 1dB than conventional method.

Embodiment 2:

Identical among the encoding and decoding structure of using in the present embodiment and the embodiment 1, identical among the parameter of employed five kinds of Adaptive Modulation and Coding grades and match and the embodiment 1, can be referring to shown in the table 1.Different with embodiment 1 is: the Adaptive Modulation and Coding grade that two code words supporting in the present embodiment are used is respectively MCS5 and MCS1 in the table 1, and the parameter of first code word match is a ₁=4.1937*10 ¹¹, g ₁=0.7371,

The parameter of second code word match is a ₂=9.9771*10 ²⁶, g ₂=43.009,

Other process all is identical with process among the embodiment 1.

Fig. 2 is that two code words are all used Adaptive Modulation and Coding grade MCS5 and MCS1 respectively; Under the ideal communication channel information state, retransmit the Packet Error Ratio performance comparison that once later multi-input multi-output system of the present invention retransmits linear predictive coding processing method and conventional retransmission linear predictive coding processing method.Wherein solid line C is the Packet Error Ratio performance curve of conventional retransmission linear predictive coding processing method, and the dotted line D of solid line C below retransmits the Packet Error Ratio performance curve of linear predictive coding processing method for multi-input multi-output system of the present invention.

As can be seen from Figure 2, under identical state of signal-to-noise, adopt multi-input multi-output system of the present invention to retransmit the linear predictive coding processing method and can reach littler Packet Error Ratio performance.The adaptive coding and modulating rank difference that two code words adopt in the present embodiment is apart from bigger, uses the code word of MCS1 to need the lower protective capability just can correct decoding, uses the code word of MCS5 to need higher protective capability ability correct decoding.Conventional retransmission linear predictive coding processing method is always oppositely matched channel when retransmitting; Promptly; Channel conditions subchannel preferably when the pairing of the relatively poor subchannel of initial transmission channel conditions retransmits, the relatively poor subchannel of channel conditions when the subchannel pairing preferably of initial transmission channel conditions retransmits; And subchannel more power preferably when conventional method can be distributed to re-transmission according to subchannel displacement and power division formula (10); Then conventional retransmission linear predictive coding processing method can the protective capability of the code word of using Adaptive Modulation and Coding grade MCS1 carry higher; Can the waste part energy, and cause using the not enough correct decoding of protective capability of the code word of Adaptive Modulation and Coding grade MCS5.Can match and power distribution strategies according to the require adaptive adjusting subchannel of each code word and adopt multi-input multi-output system of the present invention to retransmit the linear predictive coding processing method, therefore adopt multi-input multi-output system of the present invention to retransmit the linear predictive coding processing method and can reach littler Packet Error Ratio performance protective capability.Be 0.01 o'clock at Packet Error Ratio as can be seen from Figure 2, multi-input multi-output system of the present invention retransmits the linear predictive coding processing method has the performance boost more than the 4dB than conventional method.The performance chart of comparative example 1 and embodiment 2 can find out that two code words use Adaptive Modulation and Coding rank differences apart from big more, adopt the performance boost of the inventive method obvious more.

Embodiment 3:

Identical among the encoding and decoding structure of using in the present embodiment and the embodiment 1, identical among the parameter of employed five kinds of Adaptive Modulation and Coding grades and match and the embodiment 1, shown in the visible table 1.

The multi-input multi-output system antenna configurations is transmitting terminal antenna number M in the present embodiment _T=2, receiving terminal antenna number M _R=2, system supports the transmission of two code words, that is,

The Adaptive Modulation and Coding grade of two code word uses is respectively MCS5 and MCS1 in the table 1; Each time of packet channel transmitted is that average is 0, and variance is 1 independent same distribution Rayleigh channel, and the channel information ideal is imperfect, has channel estimation errors, the error matrix element variance

The tentation data bag retransmits once.

To retransmit the linear predictive coding processing method identical with traditional " communication journal " (the 28th rolls up the 9th phase 7-14 page or leaf) neutral line recoding processing method for multi-input multi-output system of the present invention when the packet initial transmission, all adopts singular value decomposition and the power algorithm of pouring water to obtain.Channel singular value decomposition process is formula (3); That is,

power process of pouring water does

${\tilde{F}}_i^0={(c_0-\frac{1+\mathrm{\&gamma;}{\mathrm{\&sigma;}}_E^2}{\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_0\right)}_{i,i}})}_{+},i=\mathrm{1,2}$

Wherein 0 _{I, i}The element value of the capable i row of the i of the matrix in the expression bracket, the horizontal plane parameter c ₀Selection need satisfy the Power Limitation condition

Initial transmission power allocation matrix then

The linear predictive coding of initial transmission is that formula (4) provides, that is,

Receiving terminal at first multiply by on the left side that receives signal during initial transmission

Respectively each code word is decoded then; Having only two code words all to separate decode and successfully indicates just feeding back a bit to transmitting terminal; Otherwise feed back bit decoding failure indication, transmitting terminal receives decoding when successfully indicating, and continues to send new data; When transmitting terminal receives decoding failure indication, data retransmission s.

For the linear predictive coding design of the retransmission processes first time, the linear predictive coding enclosed that conventional method obtains is expressed as

$F_1^{\mathrm{Trad}}=V_1{P}_1^H\mathrm{diag}(\sqrt{{\mathrm{\&lambda;}}_1},\sqrt{{\mathrm{\&lambda;}}_2})P_0$

${\mathrm{\&lambda;}}_i={(\mathrm{\&mu;}-\frac{1+\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_0^{\mathrm{Trad}}\right)}_{i,i}/(1+\mathrm{\&gamma;}{\mathrm{\&sigma;}}_E^2)}{\mathrm{\&gamma;}{\left({\mathrm{\&Lambda;}}_1^{\mathrm{Trad}}\right)}_{i,i}/(1+\mathrm{\&gamma;}{\mathrm{\&sigma;}}_E^2)})}_{+},i=\mathrm{1,2}---\left(11\right)$

Wherein conventional method retransmits V for the first time ₁The unitary matrix that obtains for the singular value decomposition of channel

Permutation matrix P ₀Make the matrix after the initial transmission singular value is replaced

Diagonal element by from big to small sequence arrangement, permutation matrix P ₁Make the matrix that retransmits for the first time after singular value is replaced

Diagonal element by from small to large sequence arrangement, the value of conventional method horizontal plane parameter μ need satisfy Power Limitation condition λ ₁+ λ ₂=1.When conventional method retransmits; Receiving terminal at first receiving the data that multiply by on the left side before to have been received each code word high specific merging respectively then of signal, is decoded then.

For the design of linear predictive coding of the retransmission processes first time, the present invention is according to following method design linear predictive coding:

To channel by formula (3) carry out singular value decomposition

The Signal to Interference plus Noise Ratio of i code word of initial transmission does

The parameter of first code word match is a ₁=4.1937*10 ¹¹, g ₁=0.7371,

The parameter of second code word match is a ₂=9.9771*10 ²⁶, g ₂=43.009,

Previous error probability does

Obtain pre-coding matrix according to following algorithm:

Initialization: the optimum permutation matrix of initialization is a unit matrix, that is, and and P _G=I, overall minimum Packet Error Ratio

The set of all permutation matrixes of mark does

Step 1:

If the permutation matrix set is an empty set, then jump to step 2, otherwise, from the permutation matrix S set, select a permutation matrix to give permutation matrix P its assignment arbitrarily, the diagonal matrix after replacing through permutation matrix P by formula (6) obtains, that is, ${\widetilde{\mathrm{\&Lambda;}}}^1={\mathrm{P\&Lambda;}}_1{P}^H/(1+\mathrm{\&gamma;}{\mathrm{\&sigma;}}_E^2);$

Give each code word (7) preassignment power by formula; That is,

If

then from permutation matrix set s, get rid of permutation matrix P jumps to step 1;

If

the surplus power is

according to Lagrange power allocation formula (8)

${\tilde{F}}_i^1={(-\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{c}_1+\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}\ln f_i^{\mathrm{previous}}{g}_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i})}_{+}$

The Power Limitation condition is satisfied in searching

The horizontal plane parameter c ₁According to minimum Packet Error Ratio formula (9)

${\mathrm{per}}_{\min }^P=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\min (\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{e}^{c_1},f_i^{\mathrm{previous}})$

Obtain the minimum Packet Error Ratio of corresponding permutation matrix P

If the minimum Packet Error Ratio of corresponding permutation matrix P is less than the minimum Packet Error Ratio of the overall situation, that is,

Then upgrading overall minimum Packet Error Ratio is the corresponding minimum Packet Error Ratio of permutation matrix P, that is,

Upgrade optimum permutation matrix P _GBe permutation matrix P, that is, and P _G=P gets rid of permutation matrix P from permutation matrix set s, jump to step 1 then;

Step 2:

According to the permutation matrix of optimum, carry out power division; Order is through optimum permutation matrix P _GDiagonal matrix after the displacement does

Give each code word (7) preassignment power by formula, that is,

$P_i^1=\max (({\mathrm{\&gamma;}}_{\mathrm{th}}^i-{\mathrm{\&gamma;}}_{G,i}^0)/\left(\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}\right),0);$

If is the i-th subchannel allocated power

jump to step three;

If then each code word of dump power

by formula (8) are distributed dump power; Promptly

${\tilde{F}}_i^1={(-\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}{c}_1+\frac{1}{g_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i}}\ln f_i^{\mathrm{previous}}{g}_i\mathrm{\&gamma;}{\left({\widetilde{\mathrm{\&Lambda;}}}^1\right)}_{i,i})}_{+}$

Horizontal plane parameter c wherein ₁Selection must satisfy the Power Limitation condition

Upgrade power division matrix element value

Step 4:

Obtaining the power division matrix does

Retransmitting the linear predictive coding matrix does

Transmitting terminal uses F ₁The s re-transmitted signal.Receiving terminal at first receiving the data that

multiply by on the left side before to have been received each code word high specific merging respectively then of signal, is decoded then.

Fig. 3 is that two code words are all used Adaptive Modulation and Coding grade MCS5 and MCS1 respectively, under the non-ideal communication channel information state, retransmits the Packet Error Ratio performance comparison that once later multi-input multi-output system of the present invention retransmits linear predictive coding processing method and conventional retransmission linear predictive coding processing method.Wherein solid line E is the Packet Error Ratio performance curve of conventional retransmission linear predictive coding processing method, and the dotted line F of solid line E below retransmits the Packet Error Ratio performance curve of linear predictive coding processing method for multi-input multi-output system of the present invention.Multi-input multi-output system re-transmission linear predictive coding processing method of the present invention can reach littler Packet Error Ratio performance under the as can be seen from Figure 3 identical state of signal-to-noise.

The adaptive coding and modulating rank difference that two code words adopt in the present embodiment is apart from bigger, uses the code word of MCS1 to need the lower protective capability just can correct decoding, uses the code word of MCS5 to need higher protective capability ability correct decoding.Conventional retransmission linear predictive coding processing method is always oppositely matched channel when retransmitting; Promptly; Channel conditions subchannel preferably when the pairing of the relatively poor subchannel of initial transmission channel conditions retransmits, the relatively poor subchannel of channel conditions when the subchannel pairing preferably of initial transmission channel conditions retransmits; And conventional method can be distributed to when retransmitting subchannel more power preferably according to formula (11); Then conventional retransmission linear predictive coding processing method can the protective capability of the code word of using Adaptive Modulation and Coding grade MCS1 carry higher; Can the waste part energy, and cause using the not enough correct decoding of protective capability of the code word of Adaptive Modulation and Coding grade MCS5.And multi-input multi-output system re-transmission linear predictive coding processing method of the present invention can be matched and power distribution strategies according to the require adaptive adjusting subchannel of each code word to protective capability, can reach littler Packet Error Ratio performance.

Claims (1)

1. a multi-input multi-output system retransmits the linear predictive coding processing method; Comprise: adopt a plurality of code words to transmit at transmitting terminal; When a packet transmitted for the first time, each code word was independently added CRC, and independently carried out Adaptive Modulation and Coding; When packet retransmitted, the first information transmitted of information and date bag that transmitting terminal retransmits was consistent; The transmitting terminal elder generation premultiplication linear predictive coding matrix data of redispatching, the method that the linear predictive coding matrix during the packet initial transmission is poured water through channel singular value decomposition and power obtains; At receiving terminal, elder generation is the associate matrix of the left unitary matrix of premultiplication channel singular value decomposition to received signal, then each code word is carried out the demodulating and decoding operation of receiving terminal: if the transmission first time of a packet is then directly decoded to each code word; If retransmit, then earlier the data of each code word are carried out high specific and merge, then decoding; Have only when all code words all separate to the time, receiving terminal is just successfully indicated to the decoding that transmitting terminal feeds back a bit, otherwise the decoding failure indication of feeding back a bit, transmitting terminal carries out retransmission operation when receiving decoding failure indication;

If mimo system transmitting terminal antenna number is M _T, receiving terminal antenna number M _RMore than or equal to the transmitting terminal antenna number, the information symbol behind the Adaptive Modulation and Coding is designated as

S wherein _i(i=1 ..., M _T) i code word of expression, the ratio of signal power and noise power is γ;

Error code word rate performance curve with each Adaptive Modulation and Coding grade in the following error code word rate model formation match system:

Wherein SINR is for receiving the Signal Interference and Noise Ratio of code word, a _iAnd g _iBe respectively multiplier constant and exponential constant that the corresponding error code word rate match of Adaptive Modulation and Coding grade that i code word select obtains,

It is the corresponding decoding threshold of Adaptive Modulation and Coding grade that i code word selected;

The channel that system estimation was come out when if packet the m time retransmitted for

then channel estimation errors modeling formula do

H wherein _mBe the mimo channel of reality, m equals the initial transmission of 0 expression packet, and m is greater than the re-transmission of 0 expression packet; Error matrix E _mEach element obey that average is 0, variance does Multiple Gaussian distribution, if desirable channel information, then variance

Be 0;

The channel

that system estimation went out when packet was retransmitted for the m time carries out singular value decomposition and gets

Equal 0 for m, that is, packet transmits for the first time, and the linear predictive coding matrix does

Wherein the power division matrix does

() _{I, i}The element value of the capable i row of the i of the matrix in the expression bracket, () ₊That bigger value in value and 0 two numerical value in the bracket, the algorithm horizontal plane parameter c of pouring water are got in expression ₀Selection need satisfy the normalized power restrictive condition

It is characterized in that: greater than 0, when promptly retransmitting for the m time, the linear predictive coding expression matrix does for m

Wherein permutation matrix P is a subchannel pairing matrix, and power division matrix

satisfies normalized Power Limitation condition

The j retransmission when the i-th codeword received signal to interference noise ratio is denoted by

is the m-th retransmission after the merger of the i-th codeword total received SINR is

define the i-th codeword previously received an error probability

Support the mimo system re-transmission linear predictive coding matrix of many code words to obtain in order to following method:

Initialization: the optimum permutation matrix P of initialization _GBe unit matrix, that is, and P _G=I, overall minimum Packet Error Ratio The set of all permutation matrixes of mark is S;

Step 1:

If the permutation matrix set then jumps to step 2 for empty set; Otherwise, from the permutation matrix S set, select a permutation matrix to give permutation matrix P its assignment arbitrarily; Diagonal matrix after order is replaced through permutation matrix P does

Give each code word preassignment power

If

then from the permutation matrix S set, get rid of permutation matrix P jumps to step 1;

If you

, the remaining power

power allocation formulas based on Lagrange

The Power Limitation condition is satisfied in searching

Parameter c _m, according to minimum Packet Error Ratio formula

Obtain the minimum Packet Error Ratio

of corresponding permutation matrix P

If the minimum Packet Error Ratio of corresponding permutation matrix P is less than the minimum Packet Error Ratio of the overall situation, that is, Then upgrading overall minimum Packet Error Ratio is the corresponding minimum Packet Error Ratio of permutation matrix P, that is,

Upgrade optimum permutation matrix P _GBe permutation matrix P, that is, and P _G=P gets rid of permutation matrix P from the permutation matrix S set, jump to step 1 then;

Step 2:

Permutation matrix P according to optimum _G, carry out power division;

Order is through optimum permutation matrix P _GDiagonal matrix after the displacement does

Give each code word preassignment power

If you

is the i-th subchannel allocated power

jump to step three;

If you

, the remaining power

assign each codeword remaining power

Horizontal plane parameter c wherein _mSelection must satisfy the Power Limitation condition

Upgrade power division matrix element value

Step 3:

Get the power allocation matrix is

retransmission linear precoding matrix is

When retransmitting for the m time, the information symbol s of transmitting terminal after to Adaptive Modulation and Coding carries out the precoding processing operation: pre-coding matrix is taken advantage of on the information symbol s left side behind Adaptive Modulation and Coding, uses F _mS sends signal.

Images