CN1992544A - Low complexity data detection using fast fourier transform of channel correlation matrix - Google Patents

Abstract

A combined signal is received over a shared spectrum in a time slot in a time division duplex communication system using code division multiple access. Each data signal experiences a similar channel response. The similar channel response is estimated. A matrix representing a channel of the data signals based on in part the estimated channel response is constructed. A spread data vector is determined based on in part a fast fourier transform (FFT) decomposition of a circulant version of the channel matrix. The spread data vector is despread to recover data from the received combined signal.

Description

Use the low-complexity Data Detection of channel correlation matrix fast fourier transform

The application is the dividing an application of No. 02804604.8 application for a patent for invention of the above-mentioned title submitted on August 6th, 2003.

(1) technical field

The relevant wireless telecommunication system of the present invention, particularly, the Data Detection in the relevant wireless telecommunication system of the present invention.

(2) background technology

Fig. 1 describes wireless telecommunication system 10.Communication system 10 has and subscriber equipment (Ues) 14 ₁To 14 ₃The base station 12 of communication ₁To 12 ₅Each base station 121 has with its operating space and subscriber equipment (UEs) 14 ₁To 14 ₃The associative operation district of communication.

Cut apart multiple access (CDMA) and use sign indicating number to cut apart the time by-pass (TDD/CDMA) of multiple access as sign indicating number, multiplex communication is to be transmitted in same frequency spectrum.These communications are by its channelization code difference.For more effective use frequency spectrum, sign indicating number is cut apart multiple access/time by-pass (TDD/CDMA) communication system and is used the repeating frame of being distinguished the time slot that is used for communication.The communication that is transmitted with this system will have one or multichannel associated code and be designated to its time slot.One yard use in one time slot is to be called as Resource Unit.

Because multiline message can be transmitted in same frequency spectrum and identical time, so the receiver in this system must be distinguished multiline message.A kind of method that detects this signal is to detect for the multichannel user.During multichannel user detects, with all UEs14 ₁To 14 ₃The signal that the user produces connection is simultaneously detected.The method of carrying out multichannel user detection is to comprise that the change basis bindings such as piece linearity of using Cholesky or approximate Cholesky to decompose detect (BLE-JD).

Another kind method is to detect for single user.During single user detected, data were only at single user (UE14 ₁) be resumed.Based on application, single user detect data can use one or many yards be transmitted.The method that fill order user detects comprises the changes such as piece linearity of using Cholesky or approximate Cholesky to decompose.These methods have high complexity.High complexity causes UE14 ₁The power consumption that the place increases progressively, thereby the minimizing that produces battery life.So expection has can detect the alternative method that is received data.

(3) summary of the invention

Composite signal is to use sign indicating number to cut apart multiple access and is received in public frequency spectrum in the time slot in the time-division two-way communication system.Each data-signal is the similar channel response of experience.Similar channel response is estimated.Expression partly based on the matrix of the channel of the data-signal that is responded by estimated channel by construction.Spread data vector is that part is decomposed based on the fast fourier conversion (FFT) of the conversion approximation of channel matrix and determined.Spread data vector is restrained from received composite signal comes restore data.

(4) description of drawings

Fig. 1 is a wireless telecommunication system.

Fig. 2 is that a simplification transmitter and a single user detect receiver.

Fig. 3 is the schematic diagram of communication burst.

Fig. 4 is the flow chart of low-complexity Data Detection.

Fig. 5-Figure 15 is the curve chart of low-complexity Data Detection usefulness.

(5) embodiment

Fig. 2 describes to use sign indicating number to cut apart one of low-complexity Data Detection in the multiple access/time-division two-way communication system to simplify transmitter 26 and receiver 28.In the canonical system, transmitter 26 is to be positioned at each UE14 ₁To 14 ₃In, and the multiplex circuit 26 of transmission multiplex communication is to be positioned at each base station 12 ₁To 12 ₅In.Low-complexity Data Detection receiver 28 can be positioned at base station 12 ₁, UEs14 ₁To 14 ₃Or both places.Receiver 28 can be used in UE14 ₁, at the multichannel user of medium or single user to detection as the high data rate service of per second 2 megabits (Mbs).As a single UE14 only ₁When being transmitted in the time slot, receiver 28 also can be used in base station 12 ₁

Transmitter 26 can transmit data in less radio-frequency channel 30.Data producer 32 in the transmitter 26 can produce the data that are transferred into receiver 28.Launch and changing device 34 deployable data, and can make launch reference data with suitably assign in time slot and the sign indicating number in the slow row sequence that connects do the time multitask, to produce communication burst or burst.

Show that as Fig. 3 typical communication burst 16 has slow 20, one guard periods 18 and two data fields 22,24 in one.In slow 20 separate two data fields 22,24, and guard period 18 can be separated the communication burst and impels the arrival time that is transmitted from the burst of different transmitters 26 to produce difference.Two data fields 22,24 comprise the communication bursty data.

The communication burst by modulator 36 modulations to radio frequency (RF).Antenna 38 can be via the antenna 40 of less radio-frequency channel 30 emitting radio frequency signals to receiver 28.It is one of any that the type that is used to be transmitted the modulation of communication can be known to the skilled artisan person, as Quadrature Phase Shift Keying (QPSK) or quadrature amplitude modulation modulation (QAM).

The antenna 40 of receiver 28 can receive various radiofrequency signals.Received signal by demodulator 42 demodulation to produce baseband signal.Baseband signal is to be assigned to the suitable sign indicating number that is received burst by being processed over time slot as channel estimation device 44 and low-complexity data detection device 46 and having.Channel estimation device 44 be to use in the baseband signal in the slow row sequence that connects form channel information be provided.Channel information is transmitted data as hard symbol (hardsymbols) by what data detection device 46 was used for estimating received communication burst.

Data detection device 46 is to use the data of the communication burst that channel information that channel estimation device 44 provided and transmitter 26 employed known spreading codes estimate to estimate that received communication burst receives.The low-complexity Data Detection is explained with the flow chart of Fig. 4.Though the low-complexity Data Detection is to use third generation partner plan (3GPP) universal terrestrial communication radio access (UTRA) sign indicating number to cut apart multiaccess system and explains that as the target communication system it still is applicable to other system.This system is by-pass (W-CDMA) system during for the direct sequence wideband, and wherein to transmit be to be restricted to the mutual exclusion time slot for up link and down link.

Receiver 28 is to use its antenna 40 to receive whole K bursts 48 of arriving at simultaneously.But the K burst is above being stacked and placed on each other with an observation interval.Some or all K bursts be derived from or to same subscriber to obtain higher data rate service.

The K of K burst ^ThBurst is to use length Q chip C ^(k)Sign indicating number launches N _sIn the symbol each is to draw length QN _sChip sequence.K ^ThBurst can length W chip the response of known or estimated channel h ^(k)Come by channel to form the chip sequence of length, N _c=(SFN _s+ W-1).SF is for launching the factor.Because uplink signal can be led from multichannel UEs14 ₁To 14 ₃So, each in the up link h ^(k)Can be distinguished.Because can't disperse the down link that transmits, so all bursts are all by identical channel and have identical h ^(k)Single vector that receives is all regarded from all users' burst by receiver 28 places rStacked and arrive at.Some or all K bursts can be many yards transmission.Because many yards are originated from identical transmitter 26, so that it has is identical h ^(k)

Multichannel subscriber signal model is to comprise N _cIndividual known receiving chip and KN _sThe information-bearing symbol of individual the unknown.K ^ThThe symbol response of burst s ^(k)Be for C ^(k)With h ^(k)Convolution.So, s ^(k)Be length for (SF+W-1) chip.W is impulse response, the chip vestige that its expression union of symbol stays.From the hurdle vector d ^(k) r ^(k)K ^ThThe N of burst _sUnknown symbols is to be K ^ThBurst is received the contribution of chip vector to integral body, r d ^(k)Be K ^ThThe data vector of burst. d ^(k)And r ^(k)Linked to each other by equation 1.

r ^(k)=A ^(k) d ^(k), wherein the k=1...K equation 1

A ^(k)Be K ^ThThe channel response matrix of burst, it is j ^ThThe hurdle is d ^(k)The N of the symbol response of element _cXN _sMatrix.Suppose time constant sign response, A ^(k)Each hurdle has identical support, s ^(k)And the hurdle that continues is the conversion version for the zero filling and first hurdle.The vector that is received of whole chip rate is according to equation 2.

$\underset{-}{r}=\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{\underset{-}{r}}^{\left(k\right)}+\underset{-}{n}$ Equation 2

N is for having variance σ ²Independent equity distribute the zero average noise vector of (i.i.d.) composition.When being write when doing single matrix equation formula, equation 2 becomes equation 3.

r=A d+ nEquation 3

A is whole channel response matrix, and it is big or small N _c* KN _sMatrix. dBe data vector, it is length K N _sThe hurdle vector.Equation 2 and equation 3 can be made and be received vector rIn the symbol border disturb (ISI) and multiple access to disturb (MAI).

Equation 1,2 and 3 signal model are that sampling is customized at chip rate, as per second 3.84 megachips (Mcps) in third generation partner plan (3GPP) universal terrestrial communication radio access system.At by the statistical accuracy that increases, receiver 28 sampling that can overuse is sampled as the multichannel chip rate.Typical multichannel chip rate sampling is the twice for chip rate, and received signal burst can be the excessive sampling that produces the multichannel sampling sequence.Each sequence is to be sampled each other with the chip rate that different time is setovered.At m ^ThSample sequence, K ^ThBurst is with known or estimated channel response h _m ^(k)Pass through channel. r _m ^(k)Be to be K ^ThBurst is to m ^ThWhole-sample chip vector r _mConvolution.Data symbol vector d ^(k)And m ^ThThe sampling A vector r _m ^(k)Linked to each other by equation 4.

r _m ^(k)=A _m ^(k) d ^(k), k=1...K wherein, m=1...M equation 4

A _m ^(k)Be m ^ThThe symbol response matrix of sequence.It is j ^ThThe hurdle is d ^(k)J ^ThThe N of the sampling symbol response of element _c* N _sMatrix.

Equation 5 is m ^ThThe whole chip rate of sample sequence be received vector.

${\underset{-}{r}}_m=\underset{i=1}{\overset{K}{\mathrm{\Σ}}}{{\underset{-}{r}}_m}^{\left(k\right)}+\underset{-}{n}\·\·\·m=1\·\·\·M$ Equation 5

At the chip rate sampling of M multichannel, single matrix notation formula becomes equation 6.

r'=A ' d + nEquation 6

Wherein r' for received signals vector and be defined as equation 7.

$\underset{-}{r^{\′}}=\left[\begin{array}{c}{\underset{-}{r}}_1\\ {\underset{-}{r}}_2\\ \·\\ \·\\ \·\\ {\underset{-}{r}}_M\end{array}\right]$ Equation 7

A ' is defined as equation 8.

$A^{\&prime;}=\left[\begin{array}{c}{A}_1\\ {\mathrm{<figure-callout\; id="2"\; label="A"\; filenames="A20071000448700241.png"\; state="\{\{state\}\}">A</figure-callout>}}_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ A_M\end{array}\right]$ Equation 8

Equation 9 is the totalling pattern of the rewriting of equation 6 as the K burst.

${\underset{-}{r}}^{\&prime;}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{A}^{\&prime;\left(k\right)}{\underset{-}{d}}^{\left(k\right)}+\underset{-}{n}$ Equation 9

Equation 9 can be rewritten as equation 10

${\underset{-}{r}}^{\&prime;}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{H}^{\&prime;\left(k\right)}{C}^{\left(k\right)}{\underset{-}{d}}^{\left(k\right)}+\underset{-}{n}$ Equation 10

C ^(k)Be K ^ThThe sign indicating number sequence of burst.H ' ^(k)Be K ^ThThe channel response of sequence, sampling is defined as equation 11 at M multichannel chip rate for it.

$H^{\&prime;\left(k\right)}=\left[\begin{array}{c}{H_1}^{\left(k\right)}\\ {H_2}^{\left(k\right)}\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ {H_M}^{\left(k\right)}\end{array}\right]$ Equation 11

When the signal burst in the time slot was derived from the same subscriber in the up link or comes same subscriber in the down link, this burst can be by identical transfer path and identical decline channel.As a result, H ' ^(k)All identical for all bursts (to all k and j, H ' ^(k)=H ' ^(j)=H ' _c), and in equation 10, be substituted H ' _cBe equation 12.

${\underset{-}{r}}^{\&prime;}=H_c^{\&prime;}\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}{C}^{\left(k\right)}{\underset{-}{d}}^{\left(k\right)}+\underset{-}{n}$ Equation 12

Equation 13 is rewritten as single matrix notation formula for equation 12.

r'=H _c' C d+ nEquation 13

C is the sign indicating number matrix.At the sampling of M chip rate, H ' _cCan be write as equation 14.

$H_c^{\&prime;}=\left[\begin{array}{c}{H}_{c1}\\ {\mathrm{<figure-callout\; id="2"\; label="H\; c"\; filenames="A20071000448700241.png"\; state="\{\{state\}\}">H</figure-callout>}}_c\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ H_{\mathrm{cM}}\end{array}\right]$ Equation 14

At m ^ThThe chip rate sampling, H _CmBe m ^ThThe channel response of sample sequence.Each H _Cm, m=1....M is by channel estimation device 44,50 decisions.Each H _CmMatrix structure be to be represented as equation 15,52.

Equation 15

The overall signal model of data detection is to be represented as equation 16 and 17.

${\underset{-}{r}}^{\&prime;}=H_c^{\&prime;}\underset{-}{s}+\underset{-}{n}$

Equation 16

s=C dEquation 17

sBe expanding data chip vector.C is a code vector.One method of decision is to use forces separating as equation 18 that equation 16 makes zero.

$\underset{-}{s}={\left(H_c^{\&prime;H}{H}_c^{\&prime;}\right)}^{-1}{H}_c^{\&prime;H}\underset{-}{r}$

Equation 18

H _c ^{' H}Be H _c' determinant.Other method is to use least mean-square error (MMSE) to separate as equation 19.

$\underset{-}{s}={(H_c^{\&prime;H}{H}_c^{\&prime;}+{\mathrm{\&sigma;}}^{2}I)}^{-1}{H}_c^{\&prime;H}\underset{-}{r}$

Equation 19

σ ²Be noise variance.I is a unit matrix.At sSolve an equation after 17 or 18, separating of equation 17 can be represented as equation 20,56 by convergence.

d=C ^H sEquation 20

Below at s18 and 19 the method for solving an equation is to use channel correlation matrix R or channel response matrix H _c', the approximate fast fourier conversion (FFT) of 54 conversion is decomposed.Use arbitrary matrix to be similar to; Yet, use channel response matrix H _c' also need the intercepting of the last W-1 row of matrix to come square.So in order to eliminate the depression of order that is produced because of intercepting, channel correlation matrix R is preferable being used.

The conversion of the fast fourier of channel correlation matrix R decompose be presented as follows.At forcing the method for making zero, R is defined as equation 21.

$R=H_c^{\&prime;H}{H}_c^{\&prime;}=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{H_{\mathrm{cm}}}^H{H}_{\mathrm{cm}}$ Equation 21

At the least mean-square error method, R is defined as equation 22.

$R=H_c^{\&prime;H}{H}_c^{\&prime;}+{\mathrm{\&sigma;}}^{2}I$

Equation 22

The structure of channel correlation matrix R is to be represented as equation 23.

Equation 18 and 19 is to be written as equation 24 and 25 respectively with the R pattern.

$\underset{-}{s}=R^{-1}{H_c}^{\&prime;H}\underset{-}{r}$

Equation 24

$R_{\underset{-}{S}}={H_c}^{\&prime;H}\underset{-}{r}$

Equation 25

Matrix-vector multiplier R _s Can be regarded as the linear combination of the hurdle vector of channel correlation matrix R, its corresponding element by data chip vector S is weighted to equation 26.

R _s =s ₁ g ₁+ s ₂ g ₂+ ...+s _W g _W+ s _W+1 g _W+1+ ...+s _NSF g _NSFEquation 26

g _iI hurdle for channel correlation matrix R.s _iI element for expanding data chip vector s.

By the structure of revising matrix R, channel correlation matrix R _CirThe approximate available equation 27 of the suitableeest circular matrix decide.

First hurdle qFull nonzero element with no any intercepting.Circular matrix R _CirBy its first hurdle qDefinition.Circular matrix R _CirFirst hurdle qBe to use as equation 28 defined in-place computation or pointer vectors, by the W hurdle of displacement channel correlation matrix R g _wAnd draw.

P=[W:NQ, 1:W-1] equation 28

Alternately, circular matrix is also by the W hurdle of channel correlation matrix R g _wDefinition.Usually, any hurdle greater than the W hurdle all suitably the pointer vector be used (permutation vector).

This alternative approximation cycle channel correlation matrix R ' _CirR can link to each other _CirBe equation 29.

R ' _Cir=R _Cir(:, p) equation 29

The advantage of the method is g _wCan not need replace and directly be used.Yet, the expanding data chip vector that this is separated sNeed by the pointer vector p ^-Backward-substitition is an equation 30.

First row by in the displacement previous mode then do not need backward-substitition s

P_=[NSF-W+2:NSF, 1:NSF-W+1] equation 30

Equation 31 is matrix R _CirFast fourier conversion decompose.

R _Cir=D _p ^-1Λ _RD _PEquation 31

D _PBe P point fast fourier transition matrix, and Λ _RBe diagonal matrix, its diagonal angle is matrix R _CirThe fast fourier conversion on first hurdle.Λ _RBe defined as Λ _R=diag (D _P q)

Use channel response matrix H _C' fast fourier conversion decompose be presented as follows.Coupling is filtered H _C' ^H r' by equation 32.

$H_C^{\&prime;H}\underset{-}{r^{\&prime;}}=\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{H}_{\mathrm{cm}}^H{\underset{-}{r}}_{\mathrm{<figure-callout\; id="32"\; label="m\; Equation"\; filenames="A20071000448700241.png"\; state="\{\{state\}\}">m</figure-callout>}}$ Equation 32

Corresponding each sample sequence H _Cm, m=1,2 ... the channel response matrix of .M is to be circular matrix.Each matrix can be broken down into three fast fourier transition matrix multipliers such as equation 33.

$H_{\mathrm{cm}}=D_P^{-1}{\mathrm{\&Lambda;}}_{H_{\mathrm{cm}}}{D}_P,m=1\&CenterDot;\&CenterDot;\&CenterDot;\&CenterDot;M$ Equation 33

As a result, the decomposition of channel response matrix becomes equation 34.

$H_C^{\&prime;H}\underset{-}{r^{\&prime;}}=D_P^{-1}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&Lambda;}}_{H_{\mathrm{cm}}}^{*}{D}_p{\underset{-}{r}}_{\mathrm{<figure-callout\; id="34"\; label="m\; Equation"\; filenames="A20071000448700241.png"\; state="\{\{state\}\}">m</figure-callout>}}$ Equation 34

For restore data chip vector s, equation 35 is to be used.

$\underset{-}{s}=R_{\mathrm{cir}}^{-1}{H}_c^{\&prime;H}{\underset{-}{r}}^{\&prime;}=D_P^{-1}{\mathrm{\&Lambda;}}_R^{-1}\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&Lambda;}}_{H_{\mathrm{cm}}}^{*}{D}_P{\underset{-}{r}}_m$ Equation 35

In the frequency domain, equation 35 becomes equation 36.

$F\left(\underset{-}{s}\right)=\frac{\underset{m=1}{\overset{M}{\mathrm{\&Sigma;}}}F{\left({\underset{-}{h}}_m\right)}^{*}\&CircleTimes;F\left({\underset{-}{r}}_m\right)}{F\underset{-}{\left(q\right)}}$ Equation 36

 represents the operand that the element multiplier multiplies each other.Utilize equation 36, F ( s) determined.By adopt F ( s) the phase transoid, spread data vector sDetermined.If the multichannel user who is used in the down link detects, or the single channel user only uses the time slot in the up link, then sCan be by using all sign indicating numbers to restrain to recover to be transmitted data dAs soft symbol.If the single channel user who is used in the down link detects, then sCan be by using user code to be restrained with restoring user data as soft symbol.

Two kinds of methods of carrying out fast fourier conversion decomposition are to be main gene algorithm (PFA) and the truth of a matter-2 algorithm.Though when two non-exponential of fast fourier transfer point was used, the main gene algorithm was regarded as more efficient than the truth of a matter-2 algorithm, so for simplicity, following complexity analyzing is based on the conversion of the truth of a matter-2 fast fourier and implements.Complexity based on the truth of a matter-2 algorithm can be regarded as the poorest case.When the main gene algorithm is used, can obtain the extra improvement of complexity.The zero tracking truth of a matter-2 fast fourier conversion implements to stay zero tracking, H _CmFirst hurdle, m=1...M, vector r _m, m=1...M and q.Zero tracking can make its length equal the nearest truth of a matter-2 integer, and it is more than or equal to the length of data field.For example, being planned direct sequence wideband time-division two-rail code by third generation partner, to cut apart the length of the data field of the burst type 1 in the burst of multiple access standard criterion be 976 chips.976 the nearest truth of a matter-2 integer is to be 1024 (P=1024).P is the truth of a matter-2 integer.

Four types of the truth of a matter-2 fast fourier conversion Calculation is to be necessity: D _P r _m, D _P h _m, D _P g ₁And Two calculating are calculated M time at the sequence of all samplings: D _P r _m, m=1....M and D _P h _m, m=1....M.Both are only calculated 1 time at sampled sequence in addition: D _P h _m, m=1....M and D _P g ₁Every time slot is calculated once.D _P r _m, m=1....M, Every time slot is calculated secondary.As a result, need the truth of a matter of 3 (M+1)-2 fast fourier conversion Calculation altogether, each needs Plog ₂The P complex calculation.Needing four actual operation by each complex calculation of hypothesis, is to become equation 37 with the complexity of the truth of a matter-2 fast fourier conversion Calculation of per second 1,000,000 actual operation (MROPS) pattern.

C ₁=3 (M+1) Plog ₂P410010 ^-6MROPS equation 37

At the complexity of vectorial multiplier, have the M element to element vector multiplier and an element to the element vector divisor, its every time slot is performed secondary.As a result, the complexity with the vector operation of per second 1,000,000 actual operation (MROPS) pattern is to become equation 38.

C ₂=2 (M+1) P410010 ^-6MROPS equation 38

At compute vector qComplexity, it needs MW ²Individual complex calculation, its every time slot is performed once.Complexity with per second 1,000,000 actual operation (MROPS) pattern is to become equation 39.

C ₃=MW ²410010 ^-6MROPS equation 39

Total complexity the convergence in per second 1,000,000 actual operation is to be presented in equation 40.

C _Fft=C ₁+ C ₂+ C ₃MROPS equation 40

Restraining every time slot is performed twice.Complexity with the convergence of per second 1,000,000 actual operation patterns is to be presented in equation 41.

C _Desp=2KNQ410010 ^-6MROPS equation 41

As a result, the total complexity that comprises the Data Detection of convergence is to be presented in equation 42 or 43.

C _Total=C _Fft+ C _DespMROPS equation 42

C _total＝[3(M+1)Plog ₂P+2(M+1)P+MW ²+2KNQ]·4·100·10 ^-6 MROPS

Equation 43

Complexity in per second 1,000,000 actual operation that following table 1024 truth of a matter-2 of demonstration (P=1024) are calculated.Complexity is to be shown in table 1 and to be shown in table 2 with the sampling of twice chip rate with chip rate.The complexity that changes such as piece linearity basis links between detection and the low-complexity Data Detection relatively is to use approximate Cholesky to decompose.Table 5 be show to use approximate Cholesky decompose as piece linearity etc. change the basis link the complexity ratio that detects the low-complexity Data Detection complexity complexity relatively.As show that it is low complexity that the low-complexity Data Detection has far beyond linking the approximate Cholesky that detects based on changes such as piece linearity basis.For most of cases, to decide on the quantity that is transmitted and launches the factor that links the complexity of the approximate Cholesky that detects based on changes such as piece linearity basis, the low-complexity Data Detection is to be 25% in chip rate, the twice chip rate is to be 30%.

Table 1. uses per second 1,000,000 actual operation of the full burst of low-complexity Data Detection to burst type 1 with the chip rate sampling needle

	Linear function is carried out in every burst					Function is carried out in per half burst twice
											The # of burst, K		Construction q	Via the truth of a matter-2	Via the truth of a matter-2	Via the truth of a matter-2		Via the truth of a matter-2	Convergence C (k)H s	Sum total

Table 2. uses per second 1,000,000 actual operation of the full burst of low-complexity Data Detection to burst type 1 with twice chip rate sampling needle

1	2.6	8.2	8.2	16.4	16.4	0.78	52.6
								8	2.6	8.2	8.2	16.4	16.4	6.25	58.1
12	2.6	8.2	8.2	16.4	16.4	9.4	61.2
								13	2.6	8.2	8.2	16.4	16.4	10.1	61.9
14	2.6	8.2	8.2	16.4	16.4	10.9	62.7
								16	2.6	8.2	8.2	16.4	16.4	12.5	64.3

Table 3 links per second 1,000,000 actual operation that detect between (BLE-JD) and the low-complexity Data Detection with change bases such as chip rate sampling comparison block linearities

Launch the factor, Q	The # of burst, K	Suppose logarithm	Piece linearities etc. are changed the basis and are linked detection
				1	1	26.7	318.2
16	8	32.2	81.1
					12	35.3	174.6
	13	36	205.5
					14	36.8	239.4
	16	38.4	318.2

Table 4 links per second 1,000,000 actual operation that detect between (BLE-JD) and the low-complexity Data Detection with change bases such as chip rate sampling comparison block linearities

Launch the factor, Q	The # of burst, K	Suppose logarithm	Piece linearities etc. are changed the basis and are linked detection
				1	1	52.6	427.6
16	8	58.1	124.8
					12	61.2	248.3
	13	61.9	287.7
					14	62.7	330.4
	16	64.3	427.6

Table 5 is as the complexity of the fast fourier conversion of the channel correlation matrix of the ratio of the complexity that links the approximate Cholesky that detects based on changes such as piece linearity basis.The complexity that links the approximate Cholesky that detects based on changes such as piece linearity basis is to be transmitted with 100% complexity.

Launch the factor, Q	The # of burst, K	The chip rate sampling	The sampling of twice chip rate
				1	1	8％	12％
16	8	39％	47％
					12	20％	25％
	13	18％	22％
					14	15％	19％
	16	12％	15％

Fig. 5-Figure 15 is a curve chart of carrying out the low-complexity Data Detection.Two high data rate service are by emulation.One is that the solid size with SF=1 transmits, and another one is many yards transmission that have 12 yards and the expansion factor 16 is respectively arranged.The low-complexity Data Detection is to launch tested down in the various delays that comprise four (WG4) of third generation partner planning work group of being defined the delay frequent case 1,2 of expansion and 3.Emulation is to be set at chip rate sampling and the sampling of twice chip rate.The length that postpones to launch is assumed to be W=57.Zero time error is supposed via full emulation.Channel impulse response supposed known accurately.Usually in emulation, it is good carrying out the many yards more corresponding solid size copies of bit error rate (BER) that transmit case.At employed special case in the emulation, solid size transmits every time slot and uses 16 Resource Units, and the many yards every time slots of transmission only use 12 Resource Units.Only use 12 sign indicating numbers can produce less interference and preferable bit error rate.Link to detect with change such as piece linearity basis and compare, in solid size and the many yards transmission, decompose the hypothesis logarithm of (FFT-R) based on the fast fourier conversion of channel correlation matrix, only a little or limited execution depression of order are observed.Solid size transmits in the case, in the mode that (FFT-R) decomposed in the fast fourier conversion of channel correlation matrix, reaches the mode of decomposing (FFT-H) based on the fast fourier conversion of channel response matrix, is mutually the same under the chip rate sampling.

To decompose (FFT-R) serve as the basis and be changes bases such as single user's bindings of ideal, the poorest case matched filtering, piece linearity to be linked single user's detection that detection and use be similar to the changes such as piece linearity of Cholesky decomposition compare based on the execution that the low-complexity Data Detection of (FFT-H) is decomposed in the fast fourier conversion of channel response matrix with the conversion of the fast fourier of channel correlation matrix in use.At key work, the scope of bit error rate is usually between 1% and 10%.Link with change such as piece linearity basis and to detect and clear signal to be done in matched filtering (MF) noise proportional (SNR) is carried out wild phase down,, only have a little or constrained signal that noise proportional is observed at the low-complexity Data Detection.Also performance is good down in addition additive white Gaussian (AWGN) channel environment for the low-complexity Data Detection.Fig. 5-Figure 15 shows that the low-complexity Data Detection links compared to the linear basis of the piece that uses approximate Cholesky to decompose to detect, and it provides the bit error rate of outstanding performance and signal to noise proportional and power consumption with low-down complexity.

Claims (15)

1. subscriber equipment that is used for branch groove multi-user communication system, described subscriber equipment comprises a receiver, described receiver is used for receiving at a time slot composite signal of a shared frequency spectrum, wherein said composite signal is included in a plurality of data-signals that receive in the described time slot, each data-signal experiences similar channel response, and described subscriber equipment comprises:

Be used to estimate the device of described similar channel response;

Being used for part is the device that construction one channel correlation matrix is come on the basis with estimated channel response;

Be used for part and be decomposed into the device that a spread data vector is decided on the basis with the fast fourier conversion of the conversion approximation of described channel correlation matrix; And

Be used to restrain described spread data vector with device from the composite signal restore data that received.

2. subscriber equipment that is used for branch groove multi-user communication system, described subscriber equipment comprises a receiver, described receiver is used for receiving at a time slot composite signal of a shared frequency spectrum, wherein said composite signal is included in a plurality of data-signals that receive in the described time slot, each data-signal experiences similar channel response, and described subscriber equipment comprises:

Be used for the take a sample device of described composite signal of the multiple of the chip rate of described composite signal;

Be used to estimate the device of described similar channel response;

Be used for part and be decomposed into the device that a spread data vector is decided on the basis with the fast fourier conversion of the conversion approximation of described channel response matrix; And

Be used to restrain described spread data vector with device from described channel response matrix restore data.

3. subscriber equipment as claimed in claim 1 or 2 also comprises: the multiple with the chip rate of described composite signal provides described composite signal, and described composite signal is to be used to estimate described channel response.

4. subscriber equipment as claimed in claim 3, the multiple of wherein said chip rate are the twice chip rate.

5. subscriber equipment as claimed in claim 1 or 2 also comprises: with the multiple of the chip rate of the described composite signal described composite signal of taking a sample, and the composite signal of being taken a sample is to be used for estimating.

6. one kind is used to use sign indicating number to cut apart the equipment (12 of the time-division two-way communication system (10) of multiple access, 14), described equipment has a receiver (28), one demodulator (42) and a channel estimation device (44), described receiver (28) is the radiofrequency signal that is configured to comprise in the time of reception groove a plurality of data-signals, wherein each data-signal experiences similar channel response, described demodulator (42) is to be configured to radiofrequency signal that demodulation receives to produce a fundamental frequency signal, described channel estimation device (44) is to be configured to estimate described similar channel response, and described equipment is characterised in that:

One Data Detection apparatus (46), it is configured to from a channel response matrix construction one channel correlation matrix H, expression is based on the channel of the data-signal of estimated channel response, be decomposed into the basis with part with the conversion of the fast fourier of the conversion approximation of described channel correlation matrix and decide a spread data vector, and restrain described spread data vector with a plurality of data-signal restore data from being received.

7. device as claimed in claim 6, wherein said Data Detection apparatus (46) are to be configured to multiply each other H via conjugation and described channel response matrix with described channel response matrix ^HH, and from the described channel correlation matrix H of described channel response matrix construction.

8. as claim 6 or 7 described devices, it is characterized in that described channel estimation device (44) is to be configured to estimate described similar channel response according to the twice of a plurality of data-signals that received or the sample of signal input of more times of chip rates.

9. device as claimed in claim 8 is characterized in that, described channel estimation device (44) is the sample of signal input that is configured to according to the twice chip rate of a plurality of data-signals that received, and estimates described similar channel response.

10. install as claimed in claim 8 or 9, the described sample of signal input of wherein said channel estimation device (44) also inputs to described Data Detection apparatus (46).

11. as claim 6,7,8,9 or 10 described devices, wherein said Data Detection apparatus (46) is that first row that are configured to use described channel correlation matrix to arrange are carried out fast fourier conversion decomposition.

12. as claim 6,7,8,9 or 10 described devices, wherein said channel estimation device (44) is to be configured to use the definition of described channel correlation matrix to be listed as to carry out the fast fourier conversion decompose.

13. as arbitrary described equipment in the claim 6 to 12, it is configured to a subscriber equipment (14).

14. as arbitrary described equipment in the claim 6 to 12, it is configured to a base station (12).

15. one kind is used for using sign indicating number to cut apart the method for time-division two-way communication system reception one composite signal of multiple access, described composite signal is the shared frequency spectrum in the time slot that is transmitted in, described composite signal comprises a plurality of data-signals, wherein each data-signal experiences similar channel response, the method is characterized in that:

From a channel response matrix construction one channel correlation matrix H, represent a channel of data-signal according to estimated channel response,

Part is decomposed into the basis with the fast fourier conversion of the conversion approximation of described channel correlation matrix and decides a spread data vector; And

Restrain described spread data vector with a plurality of data-signal restore data from being received.

Images