WO2008023296A1 - M-qam scaling factor estimation method and receiver using this method - Google Patents

M-qam scaling factor estimation method and receiver using this method Download PDF

Info

Publication number
WO2008023296A1
WO2008023296A1 PCT/IB2007/053045 IB2007053045W WO2008023296A1 WO 2008023296 A1 WO2008023296 A1 WO 2008023296A1 IB 2007053045 W IB2007053045 W IB 2007053045W WO 2008023296 A1 WO2008023296 A1 WO 2008023296A1
Authority
WO
WIPO (PCT)
Prior art keywords
real
estimation
scaling factor
imaginary components
bin
Prior art date
Application number
PCT/IB2007/053045
Other languages
French (fr)
Inventor
Stefania Sesia
Ahmet Bastug
Original Assignee
Nxp B.V.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nxp B.V. filed Critical Nxp B.V.
Publication of WO2008023296A1 publication Critical patent/WO2008023296A1/en

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B1/00Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
    • H04B1/69Spread spectrum techniques
    • H04B1/707Spread spectrum techniques using direct sequence modulation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L27/00Modulated-carrier systems
    • H04L27/32Carrier systems characterised by combinations of two or more of the types covered by groups H04L27/02, H04L27/10, H04L27/18 or H04L27/26
    • H04L27/34Amplitude- and phase-modulated carrier systems, e.g. quadrature-amplitude modulated carrier systems
    • H04L27/38Demodulator circuits; Receiver circuits

Definitions

  • the present invention relates to a M-QAM scaling factor estimation method and a receiver using this method.
  • M-QAM Multilevel Quadrature Amplitude Modulation
  • WCDMA Wide Band Code Division Multiple Access
  • HPDA High Speed Downlink Packet Access
  • M-QAM constellations scaling factor is estimated from a received block of N modulated symbols, each symbol having been modulated using the M-QAM constellation and having a real and an imaginary components, where N is an integer greater than or equal to eight.
  • Estimating the M-QAM constellation scaling factor is useful to accurately determine decision boundaries to discriminate between different received symbols.
  • the channel instantaneous power and the instantaneous channel quality may rapidly vary. This makes this scaling factor estimation for a M-QAM constellation quite a difficult task.
  • the invention provides a method to estimate this scaling factor that comprises the steps of: a) splitting the real and imaginary components of the N symbols into at least two bins using an initial decision boundary Thr , the real and imaginary components in the first bin having modules inferior to boundary Thr and the real and imaginary components in the second bin having modules superior to boundary Thr , b) calculating the value of two parameters X 1 , X 2 , the value of parameter X 1 being computed using the real and the imaginary components in the first bin and without using the real and imaginary components in the second bin, the value of parameter X 2 using the real and imaginary components in the second bin and without using the real and imaginary components in the first bin, the values of parameters X 1 and X 2 being a function of the scaling factor, and c) computing an estimation A of the scaling factor according to the values of parameters X 1 and X 2 .
  • the above method is very fast because it does work with symbol block size as small as eight.
  • the embodiments of the above method may comprise one or several of the following features: - d) computing a new decision boundary from estimation A obtained at the issue of step c), and e) iterating steps a) to c) using the new boundary instead of the initial boundary to converge to a better estimation of the scaling factor, steps a) to e) forming an expectation maximization algorithm.
  • - the values of parameters X 1 and X 2 are linked through known relations to estimation A and to an estimation ⁇ ⁇ of a noise variance, and during step c) both estimations A and C n are computed according to the values of parameters X 1 and X 2 and using the known relations.
  • V X 7 ⁇ -
  • V 1 is the module of either a real or an imaginary component
  • N 1 and N 2 are the numbers of real and imaginary components in the first and second bins, respectively.
  • B 1 and B 2 represent respectively the first and second bins so that the value of parameters X 1 and X 2 are only computed using every modules of the first and second bins, respectively.
  • C n is the noise variance estimation obtained by using only pilot symbols
  • Z is defined by the following relation:
  • CL 1 R and CL 1 1 are the square modules, respectively, of the real and imaginary components of each received symbol
  • N is the block size
  • i is an index
  • step a) through c) further improves the estimation of the scaling factor, - obtaining at the same time an estimation of the noise variance is useful for further processing like signal to interference-plus-noise ratio ("SINR"), performing a coarse estimation of the scaling factor using the assumption that the N symbols are uniformly distributed in the M-QAM constellation makes the method even faster.
  • SINR signal to interference-plus-noise ratio
  • the invention also relates to a receiver adapted to execute the above method to estimate the scaling factor of the M-QAM constellation.
  • Fig.l is a schematic diagram of a communication system using a M-QAM constellation
  • Fig.2 is a schematic diagram of a 16-QAM constellation used in the system ofFig.l
  • Fig.3 is a flowchart of a method to estimate the scaling factor of the M-QAM constellation of Fig.2
  • Fig.4 is a graph illustrating vectors used during the execution of the method ofFig.3,
  • Fig.5 is a part of a flowchart of another embodiment of a method to estimate the scaling factor of the M-QAM constellation of Fig.2.
  • Fig.l shows an orthogonal CDMA (Code Division Multiple Access) communication system 2.
  • CDMA Code Division Multiple Access
  • system 2 is a wireless telecommunication network like a UMTS (Universal Mobile Telecommunication System) system.
  • UMTS Universal Mobile Telecommunication System
  • the user equipment is a radio receiver 6 like a mobile phone.
  • Base station 4 has a radio orthogonal CDMA emitter 10 to transmit data to many user equipments within a cell.
  • emitter 10 is conform to the specification of standards 3GPP
  • Emitter 10 has a M-QAM modulator 12 able to modulate a data stream, like a bit stream, in order to transform this data stream into a succession of complex symbols Y D .
  • the M-QAM modulation used in this embodiment is a 16-QAM modulation corresponding to a 16-QAM constellation illustrated in Fig.2.
  • the constellation includes sixteen constellation points Pi to Pie and uses two amplitude levels .
  • the amplitude of the real or imaginary component of each symbol may only be equal to A or 3A .
  • A is known as the scaling factor of the constellation.
  • - ⁇ ⁇ and ⁇ is an integer value chosen in the set ⁇ -3; -1; 1; 3 ⁇ , - A is the scaling factor.
  • R A and Q 1 A represent the inphase (or real) and quadrature (or imaginary) components of symbol Y D , respectively,
  • HSDSCH High Speed Downlink Shared Channel
  • pilot symbols Y p are transmitted by emitter 10 through a dedicated pilot channel known as PCPICH (Primary Command Pilot Channel).
  • PCPICH Primary Control Pilot Channel
  • Other symbols a ⁇ , n may also be transmitted in parallel in other channels defined in the UMTS standards like in other traffic channels, for example.
  • Pilot symbol Y p is a symbol which is known by receiver symbol before this symbol is received.
  • symbol Y p is always equal to l+j-
  • Each symbols Y p , Y D is transmitted to a respective module S pi that performs a channelisation operation. More precisely, each module S pi multiplies the received symbol by a channelisation code C 1 which is orthogonal to any other channelisation codes simultaneously used for other channels.
  • Channelisation code C 1 is chosen in a OVSF (Orthogonal Variable Spreading Factor) code tree, for example.
  • OVSF Orthogonal Variable Spreading Factor
  • channellisation codes C 1 and C 2 are used to spread symbols Y p and Y D , respectively.
  • Channelisation codes C 1 and C 2 have spreading factors equal to 256 and 16, respectively.
  • the output of each module S pi is connected to an adder 50. Adder 50 adds the chip sequences corresponding to each spreaded symbol.
  • Adder 50 outputs the resulting global chip sequence to a scrambler 52.
  • Scrambler 52 scrambles the global chip sequence. More precisely, scrambler 52 multiplies the global chip sequence by a scrambling code S[I] to obtain a scrambled global chip sequence b[l].
  • Sequence b[l] is transmitted through different modules not shown before to be radiated in the air by an antenna 56 as a radio signal 58.
  • Radio signal 58 is known as a "downlink signal" in CDMA communication systems.
  • Receiver 6 has an antenna 60 to receive radio signal 58 and a radio frequency receiver 62 to convert the received radio signal into a baseband scrambled global chip sequence y[l].
  • Sequence y[l] enters an adaptative equalizer 64 that outputs an estimated scrambled global chip sequence b [Hd]-
  • the equalization of sequence y[l] introduces a delay equal to I d .
  • the estimated global chip sequence t> [Hd] is received by a descrambler 66 that descrambles the estimated chip sequence.
  • descrambler 66 multiplies sequence b [1-ld] by a complex conjugate S* [1-ld] of the scrambling code S[I-Id] used in emitter 10 at instant 1-l d .
  • the descrambled global chip sequence is then transmitted to despreader D s i as well as despreader D s2 .
  • despreader D s i despreads the descrambled global chip sequence to obtain a pilot symbol estimation YP of pilot symbol Y p . To do so, despreader D s i multiplies the descrambled global chip sequence by the channelisation code C 1 .
  • Received s2 despreads the descrambled global chip sequence using channelisation code C 2 to obtain a symbol estimation YD of symbol YD .
  • Receiver 6 also includes a noise variance estimator 70 that outputs an estimation ⁇ ⁇ of the variance of the noise that disrupts the transmission through the air of symbol Y D .
  • Estimation ⁇ ⁇ is obtained from estimation YP outputted from D sl .
  • Receiver 6 has also a scaling factor estimator 72 as well as a 16-QAM demodulator 74.
  • Estimator 72 is able to output an estimation A of scaling factor A , computed from estimations ⁇ ⁇ and Y D .
  • Demodulator 74 is able to demodulate each symbol estimation Y D in order to build a bit stream corresponding to the bit stream inputted in modulator 12. To this end, demodulator 74 uses estimation A .
  • estimators 70 and 72 are further details on estimators 70 and 72.
  • receiver 6 receives a block of N symbols transmitted through HSDSCH and, in parallel, at least one pilot symbol Y p received through PCPICH. During the reception of one pilot symbol, receiver 6 receives sixteen symbols Y D because the spreading factor C 1 is sixteen times longer than spreading factor C 2 . As a result, in this embodiment, block size N must be equal to or greater than sixteen.
  • step 92 each symbol received in step 90 is descrambled and despreaded as well as equalized in order to obtain estimations Y D and YP .
  • estimator 70 estimates from the received pilot symbol the noise variance ⁇ ⁇ .
  • the module of a vector n u is computed.
  • vector n u corresponds to the orthogonal projection of a vector Y p over a direction u .
  • Vector Y p is a vector having coordinates in a plan which are equal to the real and imaginary components of Y p , respectively.
  • Direction u is a direction which is orthogonal to a vector Y p .
  • Vector Y p is a vector having coordinates equal to the real and imaginary components of pilot symbol Y p respectively, i.e. "1" and "j", respectively.
  • estimation ⁇ ⁇ p is calculated from the following relation:
  • - n u ⁇ is the module of vector n u
  • - N p is the number of pilot symbols received during the reception of the block of N symbol Y D .
  • Relation (2) assumes that the noise is circularly symmetric Gaussian so that the noise variance in direction u is equal to half of the noise variance G n .
  • estimation ⁇ n p is normalized to obtain estimation ⁇ ⁇ of the noise variance that disrupts HSDSCH. More precisely, estimation ⁇ ⁇ is obtained from the following relation:
  • C is a normalization factor
  • the normalization factor is computed from the following relation:
  • - SF data and SF Pllot are the spreading factors of the HSDSCH channel and the CCPICH channel, respectively.
  • C is equal to C 2 IC 1 , i.e. 1/16.
  • estimator 72 proceeds to a coarse estimation step 102.
  • estimator 72 calculates a coarse estimation A 0 of scaling factor A .
  • a value Z is calculated according to the following relation:
  • estimation A 0 is obtained from the following relation:
  • Relation (6) has been obtained in the following way.
  • the squared module of the real or imaginary component of estimation Y D can be defined as follows:
  • - ⁇ is the number chosen in the following set ⁇ -3; -1; 1; 3 ⁇ , and
  • Relation (8) takes into consideration that the term E[ ⁇ n]is null because the value of ⁇ is independent from the value of n and the average value of n is null.
  • estimator 72 establishes a fine estimation A of scaling factor A using an EM (Expectation Maximization) algorithm. More precisely, in an operation 112, estimation A is initialized with the value of A 0 computed in step 102. Then, in an operation 114, a decision boundary Thr is calculated according to the following relation:
  • modules r t and r ⁇ + ⁇ are equal to the modules of $ R A + n R and $ t A + H 1 , respectively, where n R and n t are the noise that disrupts the real and imaginary components of the transmitted symbol Y D , respectively.
  • ⁇ ⁇ and ⁇ are defined by relation (1). Thereafter, for simplicity, only module r t is used irrespective of the fact that it corresponds to the module of a real or imaginary component.
  • module r t is inferior to boundary Thr , then in step 118, module ⁇ is put in a first bin B 1 . Otherwise, if module r t is superior to or equal to boundary Thr then, in operation 120, module r t is put in a second bin B 2 .
  • bin Bi is intended to receive each module r t corresponding to the real or imaginary component of a transmitted symbol Y D for which ⁇ ⁇ or ⁇ ; is equal to "1".
  • bin B 2 is intended to receive each module r t corresponding to the real or imaginary component of a transmitted symbol Y D for which ⁇ ⁇ or ⁇ ; is equal to "3".
  • each bin B 1 or B 2 may contain modules corresponding to real and imaginary components.
  • estimator 72 calculates the two following parameters X 1 and X 2 according to the following relations:
  • V 2 ⁇ i ifeLB£2 0 _ (H)
  • - N 1 and N 2 are the numbers of modules ⁇ stored in bin Bi and B 2 , respectively.
  • - B 1 and B 2 indicate that parameters X 1 and X 2 are computed using only the values stored in bin B 1 and bin B 2 , respectively.
  • the values of parameters X 1 and X 2 are linked to the value of the scaling factor estimation A by the following relations:
  • System (14) can be solved using a least square method if the value of C n is considered as already known from step 94. Otherwise, the value of C n may be considered as an unknown variable and system (14) is solved as a normal system having two equations and two unknown variables. This second solution provides the advantage to obtain a new estimation of C n which is more accurate than the one provided by step 94.
  • estimator 72 determines if step 112 to 124 have to be iterated in order to converge to a more accurate scaling factor estimation. For example, in operation 126, estimator 72 checks if a maximum number of iterations has been reached or if the value of N 1 obtained in two consecutive iterations is the same. This later case means that the method has already converged and there is no more advantage in iterating once time more operations 112 through 126.
  • step 128 the new value of the scaling factor calculated in operation 124 replaces the former one and the method returns to operation 114.
  • estimator 72 determines that iteration has to be stopped, the last obtained value of the scaling factor estimation is outputted toward demodulator 74.
  • Steps 112 to 128 form an EM algorithm.
  • demodulator 74 demodulates received symbol Y D using the estimation A to calculate the decision boundary and uses the calculated decision boundary to discriminate between different points of the constellation.
  • Fig.5 is an other method to estimate scaling factor A .
  • the method of Fig.5 is identical to the method of Fig.3 except that step 110 is replaced by a step 140. Therefore, for simplicity, already described steps have been omitted.
  • estimator 72 computes a more accurate scaling factor estimation from the initial scaling factor estimation obtained at the issue of step 102.
  • Step 140 begins by an operation 142 which is identical to operation 112. Then, in an operation 144, the value of the module r t of each real and imaginary component of Y D are classified into three bins B 1 , B 2 and B 3 , respectively.
  • the idea of classifying the value of module ⁇ into three bins comes from the fact that numbers N 1 and N 2 cannot differ too much. Therefore, it is possible to calculate two static thresholds Thr ⁇ ow and Thr ⁇ .
  • the modules r t below Thr ⁇ ow always stay in bin B 1 and the modules ⁇ above threshold Thr ⁇ always stay in bin B 2 . Only the modules r t between Thr low and Thr high can change from one bin to the other during the EM algorithm iterations. This speeds up the processing.
  • the modules r t are classified from the smallest to the highest one.
  • the N m smallest modules r t which represent thirty-five per cent of the total number 2N of modules ⁇ are put in bin B 1 in an operation 148.
  • the N B2 greatest modules ⁇ which represent thirty-five per cent of the total number 2N of modules r t are put in bin B 2 .
  • the other modules ⁇ are put in bin B 3 .
  • decision boundary Thr is calculated according to relation (9).
  • an operation 152 only modules r t contained in bin B 3 are compared to boundary Thr . If module r t is inferior to Thr then module r t is put in a bin B 1 ' , in an operation 154. Otherwise, module ⁇ is put in a bin B 2 ' , in an operation 156.
  • estimator 72 proceeds to operations 162, 164 and 166 which are identical to operation 124, 126 and 128, respectively.
  • step 94 can be replaced by any other well-known methods to calculate estimation ⁇ ⁇ .
  • This estimation ⁇ ⁇ can be computed differently from pilot symbols or can also be computed from data symbols.
  • the method has been described in the particular case of a 16-QAM constellation. However, this method may be adapted to any M-QAM constellation greater than 8-QAM constellation, like 64-QAM constellation provided that these constellations use at least two amplitude levels.
  • the method has also been described in the particular case where it is implemented to demodulate HSDSCH.
  • this method can be adapted to estimate the scaling factor of a constellation used in any other kind of communication systems like WIFI systems or other cellular telecommunication systems.

Abstract

The method to estimate a scaling factor of a M-QAM (Multilevel Quadrature Amplitude Modulation) constellation comprises the steps of: a) splitting (in 116) the real and imaginary components of N symbols into at least two bins, the real and imaginary components in the first bin having modules inferior to a boundary Thr and the real and imaginary components in the second bin having modules superior to boundary Thr, b) calculating (in 122) the value of two parameters X1, X2 from the components in the first and second bins, respectively, and c) computing (in 124) an estimation A of the scaling factor according to the values of parameters X1 and X2.

Description

M-QAM scaling factor estimation method and receiver using this method
FIELD OF THE INVENTION
The present invention relates to a M-QAM scaling factor estimation method and a receiver using this method.
BACKGROUND OF THE INVENTION
Typically, M-QAM (Multilevel Quadrature Amplitude Modulation) is used in many telecommunication systems to modulate data. For example M-QAM is used to modulate symbols transmitted through High Speed Downlink Shared Channels (HSDSCH) in Universal Mobile Telecommunication System ("UMTS"). HSDSCH were introduced as an extension of the Wide Band Code Division Multiple Access (WCDMA) standards known as High Speed Downlink Packet Access (HSPDA).
Generally, M-QAM constellations scaling factor is estimated from a received block of N modulated symbols, each symbol having been modulated using the M-QAM constellation and having a real and an imaginary components, where N is an integer greater than or equal to eight.
For example, such a method is known from patent application US 2004/0264591 in the name of MALM et al.
Estimating the M-QAM constellation scaling factor is useful to accurately determine decision boundaries to discriminate between different received symbols. However, in some situations like with HSDSCH the channel instantaneous power and the instantaneous channel quality may rapidly vary. This makes this scaling factor estimation for a M-QAM constellation quite a difficult task.
OBJECT AND SUMMARY OF THE INVENTION Accordingly, it is an object of the invention to provide a method to rapidly estimate a scaling factor of a M-QAM constellation.
The invention provides a method to estimate this scaling factor that comprises the steps of: a) splitting the real and imaginary components of the N symbols into at least two bins using an initial decision boundary Thr , the real and imaginary components in the first bin having modules inferior to boundary Thr and the real and imaginary components in the second bin having modules superior to boundary Thr , b) calculating the value of two parameters X1 , X2 , the value of parameter X1 being computed using the real and the imaginary components in the first bin and without using the real and imaginary components in the second bin, the value of parameter X2 using the real and imaginary components in the second bin and without using the real and imaginary components in the first bin, the values of parameters X1 and X2 being a function of the scaling factor, and c) computing an estimation A of the scaling factor according to the values of parameters X1 and X2.
The above method is very fast because it does work with symbol block size as small as eight.
The embodiments of the above method may comprise one or several of the following features: - d) computing a new decision boundary from estimation A obtained at the issue of step c), and e) iterating steps a) to c) using the new boundary instead of the initial boundary to converge to a better estimation of the scaling factor, steps a) to e) forming an expectation maximization algorithm. - the values of parameters X1 and X2 are linked through known relations to estimation A and to an estimation σκ of a noise variance, and during step c) both estimations A and Cn are computed according to the values of parameters X1 and X2 and using the known relations. - the known relations used during step c) are: A2 +^I = X1
2 1
9A2 + ^ = X2
2
where:
V 2
*1 =
V X7 = ≤^-
N,
where:
• V1 is the module of either a real or an imaginary component,
• N1 and N2 are the numbers of real and imaginary components in the first and second bins, respectively, and
• B1 and B2 represent respectively the first and second bins so that the value of parameters X1 and X2 are only computed using every modules of the first and second bins, respectively. performing a coarse estimation A0 of the scaling factor of the constellation using the assumption that the N symbols are uniformly distributed in the M-QAM constellation, and calculating the value of the initial decision boundary Thr from the coarse estimation A0 before executing for the first time step a). the estimation A0 is obtained from the following relation:
Figure imgf000006_0001
where Cn is the noise variance estimation obtained by using only pilot symbols, and Z is defined by the following relation:
Figure imgf000006_0002
where CL1 R and CL1 1 are the square modules, respectively, of the real and imaginary components of each received symbol, N is the block size, and i is an index.
The above embodiments of the method present the following advantages: iterating step a) through c) further improves the estimation of the scaling factor, - obtaining at the same time an estimation of the noise variance is useful for further processing like signal to interference-plus-noise ratio ("SINR"), performing a coarse estimation of the scaling factor using the assumption that the N symbols are uniformly distributed in the M-QAM constellation makes the method even faster.
The invention also relates to a receiver adapted to execute the above method to estimate the scaling factor of the M-QAM constellation.
These and other aspects of the invention will be apparent from the following description, drawings and claims.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig.l is a schematic diagram of a communication system using a M-QAM constellation, Fig.2 is a schematic diagram of a 16-QAM constellation used in the system ofFig.l,
Fig.3 is a flowchart of a method to estimate the scaling factor of the M-QAM constellation of Fig.2, Fig.4 is a graph illustrating vectors used during the execution of the method ofFig.3,
Fig.5 is a part of a flowchart of another embodiment of a method to estimate the scaling factor of the M-QAM constellation of Fig.2.
DESCRIPTION OF EMBODIMENTS
Fig.l shows an orthogonal CDMA (Code Division Multiple Access) communication system 2. In the following description, well-known functions or constructions by a person of ordinary skill in the art are not described in details.
For example, system 2 is a wireless telecommunication network like a UMTS (Universal Mobile Telecommunication System) system.
For simplicity, only one base station 4 and only one user equipment are shown. For example, the user equipment is a radio receiver 6 like a mobile phone.
Base station 4 has a radio orthogonal CDMA emitter 10 to transmit data to many user equipments within a cell. For example, emitter 10 is conform to the specification of standards 3GPP
TS 25.213 concerning the spreading and modulation. Therefore, only the details of emitter 10 necessary to understand the invention are described here.
Emitter 10 has a M-QAM modulator 12 able to modulate a data stream, like a bit stream, in order to transform this data stream into a succession of complex symbols YD . For instance, the M-QAM modulation used in this embodiment is a 16-QAM modulation corresponding to a 16-QAM constellation illustrated in Fig.2.
As illustrated in Fig.2, the constellation includes sixteen constellation points Pi to Pie and uses two amplitude levels .
In the 16-QAM constellation, the amplitude of the real or imaginary component of each symbol may only be equal to A or 3A . A is known as the scaling factor of the constellation.
According to the above constellation, symbol YD can be written as follows: YD = VRA + jVtA (1)
where:
- βΛ and β; is an integer value chosen in the set {-3; -1; 1; 3}, - A is the scaling factor.
- $RA and Q1A represent the inphase (or real) and quadrature (or imaginary) components of symbol YD, respectively,
- "j" discriminates the real component from the imaginary component of symbol YD and is defined by the following relation: : / = -1 . We assume that symbol YD is the symbol to be transmitted through a
HSDSCH (High Speed Downlink Shared Channel), for example.
In parallel to data symbol YD , pilot symbols Yp are transmitted by emitter 10 through a dedicated pilot channel known as PCPICH (Primary Command Pilot Channel). Other symbols aκ,n may also be transmitted in parallel in other channels defined in the UMTS standards like in other traffic channels, for example.
Pilot symbol Yp is a symbol which is known by receiver symbol before this symbol is received. For example, in this embodiment, symbol Yp is always equal to l+j- Each symbols Yp , YD is transmitted to a respective module Spi that performs a channelisation operation. More precisely, each module Spi multiplies the received symbol by a channelisation code C1 which is orthogonal to any other channelisation codes simultaneously used for other channels.
Channelisation code C1 is chosen in a OVSF (Orthogonal Variable Spreading Factor) code tree, for example. Such a code tree and the allocation of each code to respective channels is described in standard 3GPP TS 25.213.
Here, channellisation codes C1 and C2 are used to spread symbols Yp and YD , respectively. Channelisation codes C1 and C2 have spreading factors equal to 256 and 16, respectively. The output of each module Spi is connected to an adder 50. Adder 50 adds the chip sequences corresponding to each spreaded symbol.
Adder 50 outputs the resulting global chip sequence to a scrambler 52. Scrambler 52 scrambles the global chip sequence. More precisely, scrambler 52 multiplies the global chip sequence by a scrambling code S[I] to obtain a scrambled global chip sequence b[l]. Sequence b[l] is transmitted through different modules not shown before to be radiated in the air by an antenna 56 as a radio signal 58. Radio signal 58 is known as a "downlink signal" in CDMA communication systems.
Receiver 6 has an antenna 60 to receive radio signal 58 and a radio frequency receiver 62 to convert the received radio signal into a baseband scrambled global chip sequence y[l].
Sequence y[l] enters an adaptative equalizer 64 that outputs an estimated scrambled global chip sequence b [Hd]- The equalization of sequence y[l] introduces a delay equal to Id. The estimated global chip sequence t> [Hd] is received by a descrambler 66 that descrambles the estimated chip sequence. In fact, descrambler 66 multiplies sequence b [1-ld] by a complex conjugate S* [1-ld] of the scrambling code S[I-Id] used in emitter 10 at instant 1-ld.
The descrambled global chip sequence is then transmitted to despreader Dsi as well as despreader Ds2.
Despreader Dsi despreads the descrambled global chip sequence to obtain a pilot symbol estimation YP of pilot symbol Yp . To do so, despreader Dsi multiplies the descrambled global chip sequence by the channelisation code C1 .
Despreader Ds2 despreads the descrambled global chip sequence using channelisation code C2 to obtain a symbol estimation YD of symbol YD .
Receiver 6 also includes a noise variance estimator 70 that outputs an estimation σκ of the variance of the noise that disrupts the transmission through the air of symbol YD . Estimation σκ is obtained from estimation YP outputted from Dsl.
Receiver 6 has also a scaling factor estimator 72 as well as a 16-QAM demodulator 74. Estimator 72 is able to output an estimation A of scaling factor A , computed from estimations σκ and YD .
Demodulator 74 is able to demodulate each symbol estimation YD in order to build a bit stream corresponding to the bit stream inputted in modulator 12. To this end, demodulator 74 uses estimation A .
Further details on estimators 70 and 72 will be apparent from the description of the methods of figures 3 and 5.
The operation of receiver 6 will now be described with reference to figures 3 and 4. Initially, in step 90, receiver 6 receives a block of N symbols transmitted through HSDSCH and, in parallel, at least one pilot symbol Yp received through PCPICH. During the reception of one pilot symbol, receiver 6 receives sixteen symbols YD because the spreading factor C1 is sixteen times longer than spreading factor C2. As a result, in this embodiment, block size N must be equal to or greater than sixteen.
Then, in step 92, each symbol received in step 90 is descrambled and despreaded as well as equalized in order to obtain estimations YD and YP .
Subsequently, in step 94, estimator 70 estimates from the received pilot symbol the noise variance σκ . For example, in the operation 96, for each received pilot symbol, the module of a vector nu is computed. As illustrated in Fig.4, vector nu corresponds to the orthogonal projection of a vector Yp over a direction u . Vector Yp is a vector having coordinates in a plan which are equal to the real and imaginary components of Yp , respectively. Direction u is a direction which is orthogonal to a vector Yp . Vector Yp is a vector having coordinates equal to the real and imaginary components of pilot symbol Yp respectively, i.e. "1" and "j", respectively.
Then, in the operation 98, estimation σκ p is calculated from the following relation:
Figure imgf000011_0001
where:
- nu\ is the module of vector nu , and - Np is the number of pilot symbols received during the reception of the block of N symbol YD .
Relation (2) assumes that the noise is circularly symmetric Gaussian so that the noise variance in direction u is equal to half of the noise variance Gn .
Subsequently, in an operation 100, estimation σn p is normalized to obtain estimation σκ of the noise variance that disrupts HSDSCH. More precisely, estimation σκ is obtained from the following relation:
Figure imgf000011_0002
where C is a normalization factor.
The normalization factor is computed from the following relation:
Q _ ^ data
(4)
SF P1 ilot
where:
- SFdata and SFPllot are the spreading factors of the HSDSCH channel and the CCPICH channel, respectively.
In this embodiment, C is equal to C2IC1 , i.e. 1/16.
Subsequently, estimator 72 proceeds to a coarse estimation step 102. During step 102, estimator 72 calculates a coarse estimation A0 of scaling factor A .
To do so, it is assumed that real or imaginary components of YD are uniformly distributed. Thus, the probability that the module of the real or imaginary component is equal to A is equal to the probability that the real or imaginary component is equal to 3A .
According to this assumption, in operation 104, a value Z is calculated according to the following relation:
Figure imgf000012_0001
where:
- (X1 R and CL1 1 are the square modules, respectively, of the real and
imaginary components of each estimation YD of symbol YD received during step 90,
- N is the block size of symbols YD received in step 90, and
- i is an index.
Then, in an operation 106, estimation A0 is obtained from the following relation:
Figure imgf000012_0002
Relation (6) has been obtained in the following way.
The squared module of the real or imaginary component of estimation YD can be defined as follows:
α = {A$ + nf = A2$ 2 +n2 + 2A$ n (7)
where: - A is the scaling factor,
- β is the number chosen in the following set {-3; -1; 1; 3}, and
- n is the noise amplitude that disrupts either the real or the imaginary component of the received symbol YD. Then, the following relation can be laid down from relation (7):
E[a ] = A2E[$ 2]+E[n2] = ^ + 5A2 (8)
where:
- E[...] represents the expected value.
Relation (8) takes into consideration that the term E[βn]is null because the value of β is independent from the value of n and the average value of n is null.
It should also be understood that, due to the assumption that the amplitude of real and imaginary components are uniformly distributed, β 2 takes with a probability of 50 per cent the value "1" and with a probability of 50 per cent the value "9". Thus, the expected value of £|β 2J is equal to five.
The expected value
Figure imgf000013_0001
is approximately equal to Z as defined by relation (5). Subsequently, in step 110, estimator 72 establishes a fine estimation A of scaling factor A using an EM (Expectation Maximization) algorithm. More precisely, in an operation 112, estimation A is initialized with the value of A0 computed in step 102. Then, in an operation 114, a decision boundary Thr is calculated according to the following relation:
Thr = 2 A (9)
Then, in an operation 116, for each received symbol of the block of size N, modules rt , rl+l of, respectively, the real and the imaginary components of estimation ΫD is compared to boundary Thr .
In fact, modules rt and rι+ι are equal to the modules of $RA + nR and $tA + H1 , respectively, where nR and nt are the noise that disrupts the real and imaginary components of the transmitted symbol YD , respectively. βΛ and β; are defined by relation (1). Thereafter, for simplicity, only module rt is used irrespective of the fact that it corresponds to the module of a real or imaginary component.
If module rt is inferior to boundary Thr , then in step 118, module η is put in a first bin B1. Otherwise, if module rt is superior to or equal to boundary Thr then, in operation 120, module rt is put in a second bin B2.
So, bin Bi is intended to receive each module rt corresponding to the real or imaginary component of a transmitted symbol YD for which βΛ or β; is equal to "1". Similarly, bin B2 is intended to receive each module rt corresponding to the real or imaginary component of a transmitted symbol YD for which βΛ or β; is equal to "3".
It is also to be noticed that, during operation 116, the same treatment is applied to the real or imaginary component. As a result, each bin B1 or B2 may contain modules corresponding to real and imaginary components.
Once every module η has been classified in either bin B1 or bin B2 , in an operation 122, estimator 72 calculates the two following parameters X1 and X2 according to the following relations:
ιeB,
Xr = (10)
N
V 2 κ = i ifeLB£20_ (H)
2 N2
where:
- N1 and N2 are the numbers of modules η stored in bin Bi and B2, respectively. - B1 and B2 indicate that parameters X1 and X2 are computed using only the values stored in bin B1 and bin B2 , respectively. The values of parameters X1 and X2 are linked to the value of the scaling factor estimation A by the following relations:
A* + JL = X, (12)
9A2 + ^ = X2 (13)
2
Explanation to obtain relations (12) and (13) are similar to the ones given in view of relation (6), except that relation (12) is obtained for β = 1 and relation (13) is obtained for β = 3.
Thereafter, in an operation 124, the following system is solved to obtained the new estimations of A and c ,, :
(14)
Figure imgf000015_0001
System (14) can be solved using a least square method if the value of Cn is considered as already known from step 94. Otherwise, the value of Cn may be considered as an unknown variable and system (14) is solved as a normal system having two equations and two unknown variables. This second solution provides the advantage to obtain a new estimation of Cn which is more accurate than the one provided by step 94.
Once new estimations A and Cn have been obtained, in an operation 126, estimator 72 determines if step 112 to 124 have to be iterated in order to converge to a more accurate scaling factor estimation. For example, in operation 126, estimator 72 checks if a maximum number of iterations has been reached or if the value of N1 obtained in two consecutive iterations is the same. This later case means that the method has already converged and there is no more advantage in iterating once time more operations 112 through 126.
If estimator 72 determines that operation 112 through 124 have to be iterated, in step 128 the new value of the scaling factor calculated in operation 124 replaces the former one and the method returns to operation 114.
Otherwise, if estimator 72 has determined that iteration has to be stopped, the last obtained value of the scaling factor estimation is outputted toward demodulator 74.
Steps 112 to 128 form an EM algorithm. In step 134, demodulator 74 demodulates received symbol YD using the estimation A to calculate the decision boundary and uses the calculated decision boundary to discriminate between different points of the constellation.
Fig.5 is an other method to estimate scaling factor A . The method of Fig.5 is identical to the method of Fig.3 except that step 110 is replaced by a step 140. Therefore, for simplicity, already described steps have been omitted.
During step 140, estimator 72 computes a more accurate scaling factor estimation from the initial scaling factor estimation obtained at the issue of step 102.
Step 140 begins by an operation 142 which is identical to operation 112. Then, in an operation 144, the value of the module rt of each real and imaginary component of YD are classified into three bins B1 , B2 and B3 , respectively. The idea of classifying the value of module η into three bins comes from the fact that numbers N1 and N2 cannot differ too much. Therefore, it is possible to calculate two static thresholds Thr \ow and Thr ^φ. The modules rt below Thr \ow always stay in bin B1 and the modules η above threshold Thr ^φ always stay in bin B2. Only the modules rt between Thr low and Thr high can change from one bin to the other during the EM algorithm iterations. This speeds up the processing.
To this end, in a sub-operation 146, the modules rt are classified from the smallest to the highest one. The Nm smallest modules rt which represent thirty-five per cent of the total number 2N of modules η are put in bin B1 in an operation 148. During operation 148, the NB2 greatest modules η which represent thirty-five per cent of the total number 2N of modules rt are put in bin B2. The other modules η are put in bin B3 .
Subsequently, in an operation 150, decision boundary Thr is calculated according to relation (9). Thereafter, in an operation 152 only modules rt contained in bin B3 are compared to boundary Thr . If module rt is inferior to Thr then module rt is put in a bin B1' , in an operation 154. Otherwise, module η is put in a bin B2' , in an operation 156.
Once every module of bin B3 has been put in either bin B1' or B2' , the modules η of bin B1 are incorporated into bin B1' and the modules η of bin B2 are incorporated into bin B2 ' . Then, parameters Xi and X2 are calculated in an operation 160. Operation 160 is identical to the one described in view of step 122 except that bins B1 and B2 are replaced by bins B1 ' and B2 ' , respectively.
Then, estimator 72 proceeds to operations 162, 164 and 166 which are identical to operation 124, 126 and 128, respectively.
Many other embodiments are possible. For example, step 94 can be replaced by any other well-known methods to calculate estimation σκ . This estimation σκ can be computed differently from pilot symbols or can also be computed from data symbols. The method has been described in the particular case of a 16-QAM constellation. However, this method may be adapted to any M-QAM constellation greater than 8-QAM constellation, like 64-QAM constellation provided that these constellations use at least two amplitude levels.
The method has also been described in the particular case where it is implemented to demodulate HSDSCH. However, this method can be adapted to estimate the scaling factor of a constellation used in any other kind of communication systems like WIFI systems or other cellular telecommunication systems.

Claims

CLAIMS:
1. A method to estimate a scaling factor of a M-QAM (Multilevel Quadrature Amplitude Modulation) constellation from a received block of N modulated symbols, each symbol having been modulated using the M-QAM constellation and having a real and an imaginary components, where N is an integer greater than or equal to eight, wherein the method comprises the steps of: a) splitting (in 116) the real and imaginary components of the N symbols into at least two bins using an initial decision boundary Thr , the real and imaginary components in the first bin having modules inferior to boundary Thr and the real and imaginary components in the second bin having modules superior to boundary Thr , b) calculating (in 122) the value of two parameters X1 , X2 , the value of parameter X1 being computed using the real and the imaginary components in the first bin and without using the real and imaginary components in the second bin, the value of parameter X2 using the real and imaginary components in the second bin and without using the real and imaginary components in the first bin, the values of parameters X1 and X2 being a function of the scaling factor, and c) computing (in 124) an estimation A of the scaling factor according to the values of parameters X1 and X2.
2. The method according to claim 1, wherein the method further comprises the steps of: d) computing (in 128, 114) a new decision boundary from estimation
A obtained at the issue of step c), and e) iterating steps a) to c) using the new boundary instead of the initial boundary to converge to a better estimation of the scaling factor, steps a) to e) forming an expectation maximization algorithm.
3. The method according to any one of the preceding claims, wherein the values of parameters X1 and X2 are linked through known relations to estimation A and to an estimation σκ of a noise variance, and wherein during step c) both estimations A and σκ are computed according to the values of parameters X1 and X2 and using the known relations.
4. The method according to claim 3 for a 16-QAM constellation, wherein the known relations used during step c) are:
X +^ = X1
9A2 + ^- = X2
2
where:
Figure imgf000019_0001
ιeB?
*2 =
N,
where:
- V1 is the module of either a real or an imaginary component, - N1 and N2 are the numbers of real and imaginary components in the first and second bins, respectively, and
- B1 and B2 represent respectively the first and second bins so that the value of parameters X1 and X2 are only computed using every modules of the first and second bins, respectively.
5. The method according to any one of the preceding claims, wherein the method comprises the steps of:
- performing (in 102) a coarse estimation A0 of the scaling factor of the constellation using the assumption that the N symbols are uniformly distributed in the M-QAM constellation, and
- calculating (in 114) the value of the initial decision boundary Thr from the coarse estimation A0 before executing for the first time step a).
6. The method according to claim 5, wherein for a 16-QAM constellation, the estimation A0 is obtained from the following relation:
Figure imgf000020_0001
where:
- σκ is the noise variance estimation obtained by using only pilot symbols, and - Z is defined by the following relation:
Figure imgf000020_0002
where:
- (X1 R and CL1 1 are the square modules, respectively, of the real and imaginary components of each received symbol,
- N is the block size, and - z is an index.
7. Receiver (6) adapted to execute a method to estimate a scaling factor of a M-QAM constellation according to any one of the preceding claims.
PCT/IB2007/053045 2006-08-23 2007-08-02 M-qam scaling factor estimation method and receiver using this method WO2008023296A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP06300887 2006-08-23
EP06300887.4 2006-08-23

Publications (1)

Publication Number Publication Date
WO2008023296A1 true WO2008023296A1 (en) 2008-02-28

Family

ID=38917641

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IB2007/053045 WO2008023296A1 (en) 2006-08-23 2007-08-02 M-qam scaling factor estimation method and receiver using this method

Country Status (1)

Country Link
WO (1) WO2008023296A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8809982B2 (en) 2008-09-30 2014-08-19 Nxp B.V. Robust high aspect ratio semiconductor device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
AHMET BASTUG: "Advanced Receivers for High Speed Downlink Packet Access in UMTS", INTERNET CITATION, May 2006 (2006-05-01), XP007903841, Retrieved from the Internet <URL:http://pastel.paristech.org/1766/01/ThesisAhmet.pdf> [retrieved on 20080115] *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8809982B2 (en) 2008-09-30 2014-08-19 Nxp B.V. Robust high aspect ratio semiconductor device

Similar Documents

Publication Publication Date Title
KR100842893B1 (en) Iterative and turbo­based method and apparatus for equalization of spread­spectrum downlink channels
US6700923B1 (en) Adaptive multiple access interference suppression
EP1263179B1 (en) Channel estimation for a CDMA system using coded control symbols as additional pilot symbols
US7715487B2 (en) Wireless apparatus employing multi-level QAM and method for estimating threshold value
JP5412657B2 (en) Receiver with chip level equalization
WO2007038039A2 (en) Soft weighted interference cancellation for cdma systems
US8724741B2 (en) Signal quality estimation from coupling matrix
JP2000196522A (en) Device and method for generating control signal
WO2001091396A1 (en) Fading resistant multi-level qam receiver
JP5211305B2 (en) Method and apparatus for estimating noise variance
US20040048619A1 (en) Apparatus and method for detecting a power ratio between a traffic channel and a pilot channel in a mobile communication system
KR100979153B1 (en) Channel gain estimation in a rake receiver using complex weight generation(cwg) algorithms
US8699546B2 (en) Blind detection of modulation configuration for interfering signals
US20120076189A1 (en) Estimation of signal and interference power
Oyerinde Comparative study of Overloaded and Underloaded NOMA schemes with two Multiuser Detectors
US20120230301A1 (en) Cancelling interference in a wireless cellular network
WO2010062405A1 (en) Efficient despread and respread of multi-rate cdma signals
WO2008023296A1 (en) M-qam scaling factor estimation method and receiver using this method
Zhuang et al. Comparison of decision-directed and pilot-aided algorithms for complex channel tap estimation in a downlink WCDMA system
EP1672807B1 (en) Scaling in a receiver for coded digital data symbols
Cavalcante et al. Set-theoretic DS/CDMA receivers for fading channels by adaptive projected subgradient method
Xia et al. Analytical study of QAM with interference cancellation for high-speed multicode CDMA
Oliveira et al. Particle swarm optimization assisted multiuser detector for m-qam ds/cdma systems
He et al. The rate adaptive throughput maximization in pam-modulated overloaded system
Abou-Khousa et al. A multipath detection scheme for CDMA systems with space–time spreading

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 07805286

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

NENP Non-entry into the national phase

Ref country code: RU

122 Ep: pct application non-entry in european phase

Ref document number: 07805286

Country of ref document: EP

Kind code of ref document: A1