CN101197603A - Low complexity step-by-step detecting system and method of multi-antenna system based on spherical decoding - Google Patents

Abstract

The invention relates to a low complication fractional step detection system for an MIMO system based on sphere decoding and the detection method thereof, a signal received by the receiving terminal of the MIMO system is followed to estimate a hard-decision value of the signal emitted by the emitting terminal of the MIMO system, the hard-decision value is divided into a plurality of sub-values according to a constellation division principle, then corresponding sub-signals of the received signal can be calculated according to the plurality of sub-values acquired, external information of the sphere decoding system can be calculated after the plurality of sub-signals are calibrated. Therefore, the invention decreases the complication of the sphere decoding of the MIMO system.

Description

Multiaerial system is based on the low complexity step-by-step detection system and the detection method of globular decoding

Technical field

The present invention relates to low complexity step-by-step detection system and the detection method of a kind of multiaerial system based on globular decoding.

Background technology

In multiaerial system, the iterative detection decoding technology can make transmission rate reach the channel capacity value, and this moment, optimal detector was the maximum a posteriori probability detector.But the complexity of maximum a posteriori probability is with the bit number exponent function relation of transmit/receive antenna number and each modulation symbol.The detection architecture that has now had some suboptimum low complex degrees as technology such as Turbo-layering, iteration tree search, sequence globular decodings, but all relates to the calculating of soft information, and complexity is all very big.

Moreover existing detection technique all is that performance is followed a kind of the trading off between the detection complexity.For example, the Turbo-layering detects and to be based on hierarchical space-time code and iterative receiver structure at random, and each iteration of iterative receiver can both improve performance of BER fast, but can not guarantee best performance; The iteration tree searching and detecting is to use the M algorithm to search for optimal path in a structure tree, seeks the transmission symbolic vector with maximum a posteriori, in this algorithm, can reach the performance of optimum detection when surviving path is enough big; The sequence globular decoding is M the phasor of seeking near receiving symbol, this set is called a candidate sequence, produce the output of detector with the value of these sequences, during enough big and sequence length M long enough, the sequence globular decoding can reach the performance that the optimal detector maximum a posteriori probability detects fully when the initial radium of globular decoding.

For instance, for the multiaerial system of iterative detection decoding, it has N _TIndividual transmitting antenna N _RIndividual reception antenna, at transmitting terminal, data flow u interweaves to the constellation mapping module through after the error correction coding, and promptly every M coded-bit is mapped to a plural QAM signal, again through a string and modular converter, symbolic vector s=[s ₁..., s _NT] ^TFrom N _TLaunch simultaneously on the individual transmitting antenna.And at receiving terminal, receiving vector can be by many antenna arrays model representation of following formula plural form:

y＝HS+n (1)

Y=[y wherein ₁..., y _NR] ^TIt is the complex symbol vector that receives; H is known N _R* N ^TDimension ideal communication channel matrix, its each element all is to add up the independently multiple Gaussian random variable of zero-mean unit variance; Vector n=[n ₁..., n _NR] be that the zero-mean variance is σ ²Multiple Gaussian noise.Detect for the sequence globular decoding, it is based on the maximum a posteriori probability detection algorithm, and in maximum a posteriori probability detects, establishes each emission symbol s _i, i=1 ..., N _TBe by M bit x _i ^m, m=1 ..., the mapping of M obtains, x _i ^m, m=1 ..., the value of M is positive and negative 1, then corresponding to each coded-bit x of emission symbolic vector s _k, k=1 ..., N _TThe outside log-likelihood ratio of M can be expressed as:

$L_E\left(x_k\right|y)\≈\frac{1}{2}\underset{X\∈X_{k,+1}}{\max }\{\frac{-1}{{\mathrm{\σ}}^2}{\left|\right|y-\mathrm{HS}\left|\right|}^2+X^T{L}_A\}$ (2)

$-\frac{1}{2}\underset{X\∈X_{k,-1}}{\max }\{\frac{-1}{{\mathrm{\σ}}^2}{\left|\right|y-\mathrm{HS}\left|\right|}^2+X^T{L}_A\}-L_A\left(x_k\right)$

Wherein, M is an order of modulation, and M=2 is that quarternary phase-shift keying (QPSK) (QPSK) modulation, M=4 are that 16 quadrature amplitudes (QAM) modulation, M=6 are the 64-QAM modulation, and formula (2) has adopted max log to be similar to criterion, L _ABe known prior probability, and X _{K ,+1}={ X|x _k=+1}, X _{K ,-1}={ X|x _k=-1}.For making the search volume be reduced to the level of an appropriateness, sequence globular decoding module only need find the N that approaches maximum likelihood estimator most _CandiIndividual candidate item can be taken into account the accuracy and the detection speed of detection.According to the disclosed empirical value of existing document, work as MN _T, can do Maximum Likelihood Detection at≤8 o'clock.As 8＜MN _T≤ 32 o'clock N _Candi=512,32＜MN _T≤ 48 o'clock N _Candi=1024.Its complexity was unacceptable when yet sequence globular decoding method was used in high modulation exponent number system, only when emission symbolic vector S be binary form, the soft or hard globular decoding based on $2H^T\tilde{y}={\mathrm{\σ}}^2{L}_A$ Conversion after can calculate external information with following formula:

$L_E\left(x_k\right|y)\&ap;\frac{{\widehat{-x}}_{k,\mathrm{<figure-callout\; id="2"\; label="map"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">map</figure-callout>}}}{{2\mathrm{\&sigma;}}^2}{\left|\right|y+\tilde{y}-H{\hat{X}}_{\mathrm{map}}\left|\right|}^2$ (3)

$+\frac{{\hat{x}}_{k,\mathrm{<figure-callout\; id="2"\; label="map"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">map</figure-callout>}}}{{2\mathrm{\&sigma;}}^2}\underset{X\&Element;X_{k,-{\hat{x}}_{k.\mathrm{map}}}}{\min }{\left|\right|y+\tilde{y}\mathrm{HX}\left|\right|}^2-L_A\left(x_k\right)$

Wherein ${\hat{X}}_{\mathrm{map}}=\underset{X\&Element;X}{\arg \min }{\left|\right|y+\tilde{y}-\mathrm{HX}\left|\right|}^2$ Can be calculated and be got by hard sphere shape decoding module, the soft or hard globular decoding only need produce MN _T+ 1 candidate item is accurately calculated posteriority bit probabilities value.

As from the foregoing, if will arrive or near the performance of optimal detector, the complexity of detector is all very big usually.And when the emission symbol be binary, adopt soft or hard globular decoding method (soft to hard sphere decoder), its performance is equal to the sequence globular decoding fully and its computation complexity has obvious decline, the average complexity under high s/n ratio is O (M ⁴) (wherein M is a data block length).Because the soft or hard globular decoding can only be at binary amplitude shift keying (2ASK) and quaternary phase shift keyed (4PSK, also be QPSK) use in the modulating system, therefore, how to reduce the decoding complexity problem that has High Order Modulation System now and become the technical task that those skilled in the art need to be resolved hurrily in fact.

Summary of the invention

The object of the present invention is to provide low complexity step-by-step detection system and the detection method of a kind of multiaerial system, to reduce the complexity of described multiaerial system globular decoding based on globular decoding.

In order to achieve the above object, the invention provides the low complexity step-by-step detection system of a kind of multiaerial system based on globular decoding, it comprises: be used for estimating according to the signal that described multiaerial system receiving terminal receives value of declaring firmly of the signal that described multiaerial system transmitting terminal is launched, and declare detection and constellation fractionation module firmly according to what constellation split that principle will described value of declaring firmly be split as a plurality of components, be respectively applied for according to each self-corresponding described a plurality of soft or hard globular decoding module of declaring detection and the described corresponding sub-signal of signal that receives of the constellation fractionation component that module obtained calculating firmly, be used for calculating the external information computing module of corresponding external information according to the prior information of resulting each sub-signal of described a plurality of soft or hard globular decoding modules and acquisition in advance.

Wherein, described a plurality of soft or hard globular decoding module comprises the adjustment unit that is used for splitting according to described constellation described each sub-signal of principle correction.

Further, the present invention also provides the low complexity step-by-step detection method of a kind of multiaerial system based on globular decoding, it comprises step: 1) signal that receives according to described multiaerial system receiving terminal is estimated value of declaring firmly of the signal that described multiaerial system transmitting terminal is launched, and according to constellation fractionation principle described value of declaring firmly is split as a plurality of components; 2) calculate the corresponding sub-signal of the described signal that receives according to a plurality of components that obtained; 3) calculate corresponding external information according to the prior information of each sub-signal that is obtained and acquisition in advance.

Wherein, adopt in the described step 1) and declare detection method firmly and estimate described value of declaring firmly, describedly declare detection method firmly and comprise ZF algorithm or M algorithm, when described multiaerial system is the system of 16 quadrature amplitudes (QAM) modulation, if described value of declaring firmly is S, it can be split as component S to split principle according to constellation ₁And S ₂, wherein, $S=2/\sqrt{5}{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+1/\sqrt{5}{\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">S</figure-callout>}}_2,$ When the described signal that receives is y, its corresponding sub-signal is respectively y ₁, y ₂, satisfy: ${\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">y</figure-callout>}}_1=y-H\&CenterDot;1/\sqrt{5}{\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">S</figure-callout>}}_2,$ ${\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">y</figure-callout>}}_2=y-H\&CenterDot;2/\sqrt{5}{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">S</figure-callout>}}_1,$ Wherein, H is the ideal communication channel matrix, described step 2) also comprise an aligning step according to described each sub-signal of described constellation fractionation principle correction, promptly split principle with described sub-signal y according to described constellation ₁Proofread and correct and be y ₁/ 4.

In sum, multiaerial system of the present invention is based on the low complexity step-by-step detection system and the detection method of globular decoding, splits principle according to constellation High Order Modulation System is split as a plurality of QPSK system, can reduce the complexity of described multiaerial system globular decoding.

Description of drawings

Fig. 1 is a 16-QAM constellation point gray mappings structural representation.

Fig. 2 is the low complexity step-by-step detection system structural representation of multiaerial system of the present invention based on globular decoding.

Fig. 3 is four four and receives many antennas 16-QAM modulating system substep globular decoding detection performance of BER schematic diagram.

Fig. 4 is the complexity comparison schematic diagram based on real multiplications.

Embodiment

See also Fig. 1, multiaerial system of the present invention is based on the low complexity step-by-step detection method of globular decoding, and it may further comprise the steps:

1) signal that receives according to described multiaerial system receiving terminal is estimated value of declaring firmly of the signal that described multiaerial system transmitting terminal is launched, and split principle according to constellation described value of declaring firmly is split as a plurality of components, usually employing is declared detection method firmly and is estimated described value of declaring firmly, wherein, describedly declare detection method firmly and comprise ZF algorithm and M algorithm etc., existing constellation fractionation principle has disclosed a 16-QAM constellation point and can be split into two sub-constellation point of QPSK, therefore, multiaerial system for 16 quadrature amplitudes (QAM) modulation, setting described value of declaring firmly is S, splits principle according to existing constellation it is split as S ₁, S ₂, the three satisfies following formula:

$S=2/\sqrt{5}{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+1/\sqrt{5}{S}_2---\left(4\right)$

Wherein ${\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">S</figure-callout>}}_1={\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">S</figure-callout>}}_2=\&PlusMinus;\sqrt{2}/2\&PlusMinus;\sqrt{2}/2\&CenterDot;j,$ And S, S ₁, S ₂All be normalized form, by 16-QAM constellation point gray mappings structural representation shown in Figure 1 as can be known, the big square frame among Fig. 1 has been represented S ₁The size of planisphere, little square frame has been represented S ₂The size of planisphere, S ₁Relevant with preceding two bits of 16-QAM constellation point, S ₂Relevant with latter two bit of 16-QAM constellation point.In addition, the multiaerial system for the 64QAM modulation also can be split as three QPSK constellation point with the 64QAM constellation point by corresponding fractionation formula, does not repeat them here.

2) according to the corresponding sub-signal of the described signal that receives of a plurality of components calculating that is obtained, setting the received signal of described multiaerial system receiving terminal is y, and its corresponding sub-signal is respectively y ₁, y ₂, should satisfy: ${\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">y</figure-callout>}}_1=y-H\&CenterDot;1/\sqrt{5}{\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">S</figure-callout>}}_2,$ ${\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="A20061011931400121.png"\; state="\{\{state\}\}">y</figure-callout>}}_2=y-H\&CenterDot;2/\sqrt{5}{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="A20061011931400112.png"\; state="\{\{state\}\}">S</figure-callout>}}_1,$ Wherein, H is the ideal communication channel matrix, in described multiaerial system, and y ₁, y ₂Can obtain by two soft or hard globular decoding algorithms respectively.As shown in Figure 1, S ₁The size of planisphere is S ₂The twice of the big or small submodule 2 of planisphere for offsetting the influence of various constellations size, need be proofreaied and correct sub-signal, because that the twice of planisphere size is equivalent to energy is big 4 times, therefore, and in the present embodiment, with the sub-signal y that obtains ₁Proofread and correct and be y ₁/ 4.

3) reach the external information that the prior information that is obtained is in advance calculated each sub-signal correspondence according to each sub-signal that is obtained, usually, described each sub-signal is deducted pairing prior information promptly obtain each external information, described external information inputs to the channel decoding module of described multiaerial system and decodes accordingly for it, wherein said prior information is provided by described channel decoding module, its known to those skilled in the art knowing does not repeat them here.

See also Fig. 2, multiaerial system of the present invention comprises based on the low complexity step-by-step detection system of globular decoding: one declares detection and constellation fractionation module, a plurality of soft or hard globular decoding module and an external information computing module firmly.

Described declare firmly detect and constellation splits value of declaring firmly that module is used for estimating according to the signal that described multiaerial system receiving terminal receives the signal that described multiaerial system transmitting terminal is launched, and split principle according to constellation described value of declaring firmly is split as a plurality of components, for example, multiaerial system for 16 quadrature amplitudes (QAM) modulation, split principle according to constellation and it can be split as two sub-constellation point of QPSK, multiaerial system for the 64QAM modulation also can be split as three QPSK constellation point with the 64QAM constellation point by corresponding fractionation formula.

Described a plurality of soft or hard globular decoding module is respectively applied for according to each self-corresponding described declare firmly detection and the corresponding sub-signal of the constellation fractionation component that module the obtained described signal that receives of calculating, in the present embodiment, described a plurality of soft or hard globular decoding module i.e. the first soft or hard globular decoding module and the second soft or hard globular decoding module, when described value of declaring firmly is split as two component S ₁And S ₂, the first soft or hard globular decoding module is according to component S ₂Calculate the corresponding sub-signal y of the described signal that receives ₁, the second soft or hard globular decoding module is according to component S ₁Calculate the corresponding sub-signal y of the described signal that receives ₂, split principle as can be known according to constellation, S ₁The size of planisphere is S ₂The twice of the big or small submodule 2 of planisphere, for offsetting the influence of various constellations size, need proofread and correct sub-signal, because it is big 4 times that the twice of planisphere size is equivalent to energy, therefore, in the present embodiment, the described first soft or hard globular decoding module also is provided with adjustment unit, is used for described sub-signal y ₁Proofread and correct and be y ₁/ 4.Be noted that, the number of soft or hard globular decoding module is not to exceed with present embodiment, those skilled in the art can be set according to the actual needs, have again, the setting of adjustment unit also can be set according to the actual needs, for example, the multiaerial system for the 64QAM modulation is provided with three soft or hard globular decoding modules accordingly and is used for calculating respectively each corresponding sub-signal.

Described external information computing module is used for calculating corresponding external information according to the prior information of resulting each sub-signal of described a plurality of soft or hard globular decoding modules and acquisition in advance, for example, and with described sub-signal y ₁The prior information that deducts the one or two promptly obtains the one or two external information, with described sub-signal y ₂The prior information that deducts the three or four promptly obtains the three or four external information, can obtain the external information of every bit thus.

Below be to the present invention being carried out the simulation result of emulation, the system emulation parameter is shown in following table one, Fig. 3 is that the present invention reaches the performance of BER under the different signal to noise ratio conditions under flat fading channel, the performance of existing sequence globular decoding also illustrates in the drawings as a reference, and its simulation parameter is also as shown in the table:

Table one simulated conditions is set

Antenna configurations	4×4
		Debud mode	16QAM
Channel	The flat fading Rayleigh channel
		The bit number of each data block	4004
Emulation frame number under each signal to noise ratio condition	2500
		Chnnel coding	1/2Turbo(5，7)
Turbo iterative decoding inner iteration number of times	8
		The globular decoding initial search radius	1e4
The iterations of iterative receiver	4

Initial globular decoding search radius is set at an enough big value and can finds suitable point to calculate external information to guarantee the QPSK submodule, considered that prior information obtains by the search of hard sphere shape decoding module, has considered the point that priori can search out later so need an enough big radius to be included in because satisfy the candidate item of aforementioned formula (3).

In bit error rate is 10 ^-4The time scheme hard sphere shape decoding during with the cascade of substep globular decoding the performance of (SD/MSD) 2dB has descended than LSD (sequence globular decoding), the performance of (MMSE/MSD) when also having provided simultaneously the scheme least mean-square error with the cascade of substep globular decoding, its 1dB that descended again.

It is suitable with the complexity of MMSE/MSD to get SD/MSD from Fig. 4, but considers that from performance SD/MSD is more superior.

Further, complexity is analyzed as follows:

Under above-mentioned simulated conditions, receive vector for each, sequence globular decoding module need be searched for 512 best candidate item and be guaranteed that its performance is equal to the performance of maximum a posteriori probability detector substantially.Yet method proposed by the invention is divided into the realization of two steps to detection, only needs 2 * 9 candidate item to calculate the posterior probability of each bit.Because, need MN for each QPSK submodule _T+ 1=2 * 4+1=9 candidate item obtains the soft information of each bit.

Fig. 4 provided that method of the present invention and sequence globular decoding detect based on the complexity of real multiplications relatively, the former complexity is the latter's 0.6%.

In second to the 4th iteration of iterative detection, submodule has used with the same fractionation constellation point of the iteration first time.Although least mean-square error detects much simpler than the detection of hard sphere shape decoding, this process of declaring is firmly only carried out once, and SD/MSD is basic identical with the complexity of MMSE/MSD on the whole.Table two has provided concrete real multiplications complexity relatively.

The real multiplications complexity of table two iterative process relatively

Claims (10)

1. a multiaerial system is characterized in that comprising based on the low complexity step-by-step detection system of globular decoding:

Declare firmly and detect and constellation fractionation module, be used for estimating value of declaring firmly of the signal that described multiaerial system transmitting terminal is launched, and described value of declaring firmly is split as a plurality of components according to constellation fractionation principle according to the signal that described multiaerial system receiving terminal receives;

A plurality of soft or hard globular decoding modules are respectively applied for according to each self-corresponding described declare firmly detection and the corresponding sub-signal of the constellation fractionation component that module the obtained described signal that receives of calculating;

The external information computing module is used for calculating corresponding external information according to the prior information of resulting each sub-signal of described a plurality of soft or hard globular decoding modules and acquisition in advance.

2. multiaerial system as claimed in claim 1 is characterized in that based on the low complexity step-by-step detection system of globular decoding: described a plurality of soft or hard globular decoding modules comprise and are used for splitting the adjustment unit that principle is proofreaied and correct described each sub-signal according to described constellation.

3. a multiaerial system is characterized in that may further comprise the steps based on the low complexity step-by-step detection method of globular decoding:

1) signal that receives according to described multiaerial system receiving terminal is estimated value of declaring firmly of the signal that described multiaerial system transmitting terminal is launched, and according to constellation fractionation principle described value of declaring firmly is split as a plurality of components;

2) calculate the corresponding sub-signal of the described signal that receives according to a plurality of components that obtained;

3) calculate corresponding external information according to the prior information of each sub-signal that is obtained and acquisition in advance.

4. multiaerial system as claimed in claim 3 is characterized in that based on the low complexity step-by-step detection method of globular decoding: adopt in the described step 1) and declare the described value of declaring firmly of detection method estimation firmly.

5. multiaerial system as claimed in claim 4 is characterized in that based on the low complexity step-by-step detection method of globular decoding: describedly declare detection method firmly and comprise the ZF algorithm.

6. multiaerial system as claimed in claim 4 is characterized in that based on the low complexity step-by-step detection method of globular decoding: describedly declare detection method firmly and comprise the M algorithm.

7. multiaerial system as claimed in claim 3 is based on the low complexity step-by-step detection method of globular decoding, it is characterized in that: when described multiaerial system is the system of 16 quadrature amplitudes (QAM) modulation, if described value of declaring firmly is S, it can be split as component S to split principle according to constellation ₁And S ₂, wherein, $S=2/\sqrt{5}{S}_1+1/\sqrt{5}{S}_2.$

8. multiaerial system as claimed in claim 7 is characterized in that based on the low complexity step-by-step detection method of globular decoding: the described signal that receives is y, and its corresponding sub-signal is respectively y ₁, y ₂, satisfy: $y_1=y-H\&CenterDot;1/\sqrt{5}{S}_2,$ $y_2=y-H\&CenterDot;2/\sqrt{5}{S}_1,$ Wherein, H is the ideal communication channel matrix.

9. multiaerial system as claimed in claim 3 is characterized in that based on the low complexity step-by-step detection method of globular decoding: described step 2) also comprise an aligning step according to described each sub-signal of described constellation fractionation principle correction.

10. multiaerial system as claimed in claim 9 is characterized in that based on the low complexity step-by-step detection method of globular decoding: split principle with described sub-signal y according to described constellation ₁Proofread and correct and be y ₁/ 4.

Images