`Raitola et al.
`
`USOO6445757B1
`(10) Patent No.:
`US 6,445,757 B1
`(45) Date of Patent:
`*Sep. 3, 2002
`
`(54) DIVERSITY COMBINING METHOD, AND
`RECEIVER
`
`(*) Notice:
`
`(75) Inventors: Mika Raitola, Masala; Harri Jokinen,
`Hiisi, Pekka Ranta, Nummela, all of
`(FI)
`(73) Assignee: Nokia Telecommunications Oy, Espoo
`(FI)
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 0 days.
`This patent is Subject to a terminal dis
`claimer.
`(21) Appl. No.:
`09/242,405
`(22) PCT Filed:
`Aug. 15, 1997
`(86) PCT No.:
`PCT/FI97/00479
`S371 (c)(1),
`(2), (4) Date: Jun. 3, 1999
`(87) PCT Pub. No.: WO98/07243
`PCT Pub. Date: Feb. 19, 1998
`Foreign Application Priority Data
`(30)
`Aug. 15, 1996
`(FI) ................................................. 9632O3
`(51) Int. Cl. .................................................. H04B 7/12
`(52) U.S. Cl. ....................... 375/347; 375/348; 375/349;
`375/341; 375/262; 375/267; 455/504; 455/517;
`455/63
`(58) Field of Search ................................. 375/259, 262,
`375/267, 147, 152, 148, 340, 341, 346,
`347, 348,349; 455/67.1, 67.3, 63, 422,
`513, 504,506, 517
`
`(56)
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`5,199,047 A 3/1993 Koch
`5,297,171. A
`3/1994 Koch
`
`
`
`5,553,102 A 9/1996 Jasper et al.
`6,192.238 B1
`2/2001 Piirainen .................... 455/422
`FOREIGN PATENT DOCUMENTS
`
`EP
`EP
`WO
`WO
`
`O 490 427 A3
`0 716513 A1
`WO 92/08298
`WO97/08841
`
`6/1992
`6/1996
`5/1992
`3/1997
`
`OTHER PUBLICATIONS
`ASztély, D., “On Antenna Arrays in Mobile Communication
`Systems, Fast Fading and GSM Base Station Receiver
`Algorithms, IR-S3-SB-9611, Royal Institute of Technol
`ogy, pp. 1-73 (Mar. 1996).
`Bottomley, et al., “Adaptive Arrays and MLSE Equaliza
`tion", IEEE, pp. 50–54, (1995).
`International Search Report for PCT/FI97/00479.
`“Dallas Globecom 89-IEEE Global Telecommunications
`Conference & Exhibition' Nov. 27–30, 1989 “Communica
`tions Technology for the 1990s and Beyond”.
`“Combining Technology,” Chapter 10. Mobile Communica
`tions Engineering. pp. 291-336.
`* cited by examiner
`Primary Examiner-Chi Pham
`Assistant Examiner Khanh Cong Tran
`(74) Attorney, Agent, or Firm-Altera Law Group, LLC
`(57)
`ABSTRACT
`The invention relates to a diversity combining method and
`to a receiver. In diversity combining, outputs (33) of a
`matched filter (25) of each branch are weighted with a
`quality estimate (32) which is generated in quality means
`(28b) in such a matter that the quality estimate (30) is
`proportional to the inverse of the interference strength (31)
`of the Signal. The strength (31) of Signal interference is
`generated in interference means (28c) e.g. in variance-like
`fashion from the differences between a reference signal (30)
`generated in reference signal means (28a) as the convolution
`of the estimated channel impulse response and the prede
`termined Sequence and a signal received from the channel.
`
`10 Claims, 2 Drawing Sheets
`
`IPR2020-00038
`MM EX1030, Page 1
`
`
`
`U.S. Patent
`US. Patent
`
`Sep. 3, 2002
`Sep. 3, 2002
`
`Sheet 1 of 2
`Sheet 1 0f 2
`
`US 6,445,757 B1
`US 6,445,757 B1
`
`21
`
`
`
`Fig. 3
`
`
`
`A/D
`
`|PR2020-00038
`
`MM EX1030, Page 2
`
`IPR2020-00038
`MM EX1030, Page 2
`
`
`
`U.S. Patent
`US. Patent
`
`Sep. 3, 2002
`Sep. 3, 2002
`
`Sheet 2 of 2
`Sheet 2 0f 2
`
`US 6,445,757 B1
`US 6,445,757 B1
`
`
`
`
`
`
`51
`
`41
`
`50
`
`Fig.
`
`5
`
`|PR2020-00038
`
`MM EX1030, Page 3
`
`IPR2020-00038
`MM EX1030, Page 3
`
`
`
`1
`DIVERSITY COMBINING METHOD, AND
`RECEIVER
`
`US 6,445,757 B1
`
`15
`
`2
`from the estimated channel impulse response and the pre
`determined Sequence by convolution; an interference
`Strength connected to the desired Signal is generated using
`the differences of the reference Signal and the predetermined
`Sequence received from the channel; a Strength value of the
`desired signal is generated, whereby a quality estimate is
`generated by dividing the Strength value of the desired signal
`by the interference Strength of the desired Signal; and
`diversity combining is performed in Such a manner that the
`symbols of the different branches corresponding with each
`other in the time domain are combined, and the outputs of
`the matched filters of each branch and the autocorrelation
`taps of the impulse response are weighted with the quality
`estimate of each branch.
`The receiver of the invention is characterized in that the
`receiver comprises reference Signal means for generating a
`reference Signal from the estimated channel impulse
`response and the predetermined Sequence by convolution;
`interference means for generating the interference Strength
`asSociated with the desired signal using the differences of the
`reference Signal and the predetermined Sequence received
`from the channel; the receiver is arranged to generate a
`Strength value of the desired Signal and quality means are
`arranged to generate a quality estimate by dividing the
`Strength value of the desired signal by the interference
`Strength of the desired signal; and combining means of the
`diversity branches combine the symbols of the different
`branches corresponding to each other in the time domain,
`and that the receiver is arranged to weight the matched filter
`outputs of each branch and the autocorrelation taps of the
`impulse response with the quality estimate of each branch.
`Great advantages are achieved with the invention. With
`the method of the invention the interference strength can be
`estimated directly from the received signal without perform
`ing a Viterbi detection. By avoiding the use of the Viterbi
`algorithm usually applied to the ML method memory and
`time used for calculating are Saved. The generated interfer
`ence Strength can be utilized for estimating the Status of the
`channel, as help in methods of estimating bad frames and for
`scaling the ML metric. Furthermore, the interference
`Strength can be utilized for diversity combining and it is
`particularly useful when multipath Signals are combined
`before detection.
`In the following, the invention will be described in
`greater detail with reference to examples in the accompa
`nying drawings, in which
`FIG. 1 shows a radio system,
`FIG. 2 shows a normal burst of the GSM system,
`FIG. 3 shows a block diagram of the receiver,
`FIG. 4 shows a block diagram of the receiver and
`FIG. 5 shows a receiver using a diversity combining
`technique.
`The method and the receiver of the invention can be
`applied to the GSM radio system (Global System for Mobile
`communication) without restricting it thereto. In FIG. 1 the
`radio System comprises base Station 1, and a number of
`generally moving Subscriber terminals 2-4 having
`bi-directional connections 6-8 with the base station. Base
`Station 1 transmits the connections of terminals 2–4 to base
`station controller 5 which transmits them further to other
`parts of the System and if necessary to a fixed network. Base
`station controller 5 controls the function of one or several
`base stations 1. In the GSM system both base station 1 and
`terminals 2–4 constantly measure connection quality.
`Let us now examine in more detail the Solution of the
`invention in the GSM system. A normal burst of the GSM
`system is shown in FIG. 2, the burst comprising 148 symbols
`
`The invention relates to a diversity combining method in
`a digital radio System receiver, in which receiver matched
`filtering and maximum likelihood detection are used and an
`estimated channel impulse response and autocorrelation taps
`of the impulse response are generated, and in which radio
`System Substantially all Signal processing occurs as Symbols
`and a desired signal comprises a predetermined Sequence.
`The invention also relates to a receiver in a digital radio
`System, the receiver comprising a matched filter, diversity
`branches and a maximum likelihood detector, the receiver
`being arranged to generate an estimated channel impulse
`response and autocorrelation taps of the impulse response,
`and in which radio System a desired signal comprises a
`predetermined Sequence and in which radio System signal
`processing is arranged to occur as Symbols.
`In a radio System the quality of the connection between
`a base Station and a Subscriber terminal varies continuously.
`This variation is due to interfering factors on the radio path
`and to attenuation of the radio waves as a function of
`distance and time in a fading channel. Connection quality
`can be measured for example by observing the received
`power. Variance in connection quality can partly be com
`25
`pensated by power regulation.
`However, in a digital radio System a more precise method
`than power measuring is needed for estimating connection
`quality. Then the known quality parameters are for example
`the bit error rate (BER) and the signal-to-noise ratio.
`It is previously known to utilize decisions of the ML
`(Maximum Likelihood) type detection for estimating the
`Signal-to-noise ratio of a received signal. Thus a Viterbi
`detector usually functions as the ML detector and a base
`Station or Subscriber terminal can be the receiver. In known
`solutions the Viterbi detection is performed on the received
`burst in full before determining the Signal-to-noise ratio.
`However, as a Viterbialgorithm is often a too demanding
`measure for a digital Signal processing program to perform
`during the processing time allowed by the receiver, Separate
`Viterbi hardware has to be used. This has been described in
`greater detail in J. Hagenauer, P. Hoeher: A Viterbi Algo
`rithm with Soft-decision Outputs and its Applications, IEEE
`GLOBECOM 1989, Dallas, Tex., November 1989, which is
`incorporated herein by reference.
`It is known that a signal quality estimate, often the
`Signal-to-noise ratio, is needed when using different diver
`sity receivers. In diversity reception the most common
`diversity receivers combine the Signals before or after detec
`tion and comprise e.g. Selective combining, maximal-ratio
`combining and equal-gain combining. The diversity signals
`are usually detected using a Viterbi detector, the Signals
`being combined after detection. However, it is preferable to
`combine the Signals before detection, thus achieving a
`greater amplification of the Signal. Diversity receivers have
`been described in greater detail for example in the book
`William C. Y. Lee: Mobile Communications Engineering,
`chapter 10, Combining technology, pages 291-336,
`McGraw-Hill, USA, 1982, which is incorporated herein by
`reference.
`An object of the present invention is to implement a
`method for estimating the interference Strength directly from
`a received signal without the help of ML detection and
`Simultaneously enabling the combination of diversity Sig
`nals before detection when using diversity receivers.
`This is achieved with the method set forth in the pre
`amble characterized in that a reference Signal is generated
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`IPR2020-00038
`MM EX1030, Page 4
`
`
`
`3
`in all. The symbols comprise bits and bit combinations. The
`Symbols of the burst are arranged in Sequences comprising
`3 start symbols (TS) 10, 58 data symbols (Data) 11, 26
`training symbols (TRS) 12, 58 data symbols (Data) 13 and
`3 end symbols (TS) 14. In the solution of the invention the
`Symbol Sequence of the reference Signal is calculated as the
`function of training Sequence 12 and of the estimated
`channel impulse response, preferably being the convolution
`of Said Sequences. The generation of convolution function
`h(t) can be shown in the following way between functions
`f(t) and g(t) in its general form:
`
`15
`
`25
`
`In the following, one method of the invention is
`described when applying it to the GSM System in particular.
`The calculation of the channel's momentary quality estimate
`QE comprises two essential Steps: firstly the generation of
`reference Signal YR from estimated channel impulse
`response H and training sequence TRS (training Symbols 12
`in FIG. 2) preferably as a convolution and secondly the
`generation of the interference Strength for example as inter
`ference energy VAR from reference Signal YR and training
`Sequence Y received from the channel in a variance-like
`fashion. By calculating the convolution Such an advantage is
`achieved that reference Signal YR is generated in the same
`way as the actual signal on the channel and by comparing
`this reSuit with the desired signal received from the channel
`the interference strength can be estimated. Variance VAR is
`generally calculated for the discreet distribution in the
`following way:
`
`o° = VAR = X(x, -pi) f(x,),
`
`i
`
`(2)
`
`35
`
`where u is an expected value. The interference Strength can
`also be determined for example in a Standard deviation-like
`fashion. Standard deviation 6 is according to its definition
`the positive Square root of variance of. Furthermore, qua
`dratic difference (x-1) can in the method of the invention
`be replaced by any exponent x-ul of the absolute value of
`the difference where Z is any real number. When interference
`strength VAR is calculated from reference signal YR and
`from the Signal received from the channel in a variance type
`fashion, Such an advantage is achieved that the result
`obtained is directly the effective value of the interference.
`AS training Sequence TRS is predetermined it is possible
`to determine momentary estimated channel impulse
`response H. Usually estimated impulse response H has 5
`symbols i.e. N=5 is valid for the number of symbols N. In
`the first Step of the method of the invention reference Signal
`YR, which is the expected value of received training
`Sequence TRS with Said estimated impulse response H, is
`calculated e.g. according to formula (3) as the convolution
`of estimated channel impulse response H and training
`sequence TRS.
`
`40
`
`45
`
`50
`
`55
`
`N
`YR(i) = X. HU (i). (1 - 2. TRS(j - i))
`i=0
`
`(3)
`
`60
`
`where N is the number of symbols in estimated impulse
`response HandjeN is valid for symbol indexj which shows
`the symbol to be calculated. Entire reference signal YR is
`obtained by going through the symbols between N and 26
`
`65
`
`US 6,445,757 B1
`
`4
`or the number of Symbols in a predetermined Sequence.
`Using obtained reference Signal YR and received signal Y
`comprising the training Sequence, their variance-type inter
`ference Strength VAR is calculated for example by using
`formula (4).
`
`26
`X. Re(Y(i+ offset) - YR(i)) +lm(Y(i+ offset) - YR(i))
`VAR = Y
`
`K
`
`(4)
`
`The maximum number of Symbols taken into account in
`formula (4) is the number of symbols of predetermined
`Sequence 12 less the number of Symbols in the estimated
`channel impulse response. Then the number of Symbols
`considered in the calculation can be freely chosen. Interfer
`ence Strength VAR is thus calculated as variance, but number
`Kin its nominator is not significant, as the nominator simply
`has to be formed and it only functions as the Scalar of the
`interference Strength. This is easy to observe and to correct
`in any step of generating the quality estimate. In formula (4)
`the value of variance-type result VAR is the same as the
`interference energy per Sample if the number of Symbols
`used in Summing is Set as the value of divisor K, or the
`energy per entire Sequence if the value of divisor K is one.
`In formula (4) I/O modulation markings are used the Sym
`bols being shown in their complex mode. An offset is also
`observed in formula (4) i.e. it is preferable to transfer the
`Symbols of the received signal in Such a manner that the
`Symbol of the received signal corresponds with the Symbol
`of the reference Signal.
`The strength value E of the received desired signal, which
`value can be the amplitudinal strength of the symbols to be
`considered in the Summing or the effective value or another
`corresponding exponent of the Symbols amplitude, can be
`calculated either using the estimated channel impulse
`response H, using reference Signal YR or using the desired
`Signal received from the channel. The advantage of calcu
`lating the effective value of the taps of estimated channel
`impulse response H is that energy E of the Signal is obtained
`per Symbol. When energy E is calculated using the complex
`Symbols of the I/O modulation of the reference Signal e.g.
`using formula (5)
`
`the entire energy of the reference Signal is obtained directly.
`The energy of the Signal received from the channel can be
`calculated Similarly. If the normalized average energy of the
`Signal corresponding to energy E, is formed by preproceSS
`ing means 24 according to prior art, it does not have to be
`Separately calculated. The Strength value of the desired
`Signal is directly calculated from the desired signal as in
`formula (5), but the symbols of reference signal YR are
`replaced with the symbols of desired signal Y.
`Momentary channel quality estimate QE is preferably
`obtained by forming the inverse of the interference Strength
`as shown in formula (6)
`
`1
`
`(6)
`
`Quality estimate QE can also be formed as shown in formula
`(7) by way of principle by dividing signal energy per Symbol
`
`IPR2020-00038
`MM EX1030, Page 5
`
`
`
`US 6,445,757 B1
`
`S
`E by noise energy per symbol VAR when the number of
`symbols is used as the value of divisor K in formula (4).
`
`QE = ?.
`WAR
`
`(7)
`
`6
`25 are weighted by quality estimate QE. The detected
`symbols are the output of ML detection means 29.
`Let us now examine in greater detail a Second radio
`System receiver of the invention as an alternative to the first,
`the block diagram of which is shown with essential parts in
`FIG. 4. The receiver is, to a very large extent, similar to the
`receiver in FIG. 3. In this receiver Solution quality estimate
`32 (QE) is transferred from quality means 28c to means 26,
`which generates an impulse response weighted by quality
`estimate QE, whereby quality estimate QE affects output 33
`of matched filter 25 and weights the desired Signal being
`processed. The detected symbols are the output of ML
`detection means 29.
`The solutions shown in FIGS. 3 and 4 can preferably be
`utilized in multipath reception, FIG. 5 showing such an
`arrangement, when the receiver uses diversity combining.
`The receiver in FIG. 5 comprises two diversity branches 50,
`51, both of which comprising antenna 41 and 42, means 43
`and 44, which in turn comprise e.g. radio frequency parts 22,
`conversion means 23, preprocessing means 24, matched
`filters 25, channel impulse response estimation means 26,
`calculation means 28 of the quality estimate, as do the
`receivers in FIGS. 3 and 4. Although FIG. 5 shows only two
`diversity branches, or channels, 50, 51, similar diversity
`combining can also be applied to Several channels. Esti
`mated impulse response autocorrelation taps 34 of the
`different channels are generated by means 27, which repre
`sent the same function as means 27 of FIGS. 3 and 4. The
`Signal components arriving from different channels, and that
`are outputs 33 of matched filter 25, are combined by means
`45 where the combining is performed e.g. by Summing or
`averaging and when So desired by multiplying the Signals by
`a Suitable constant. In the Solution of the invention the Signal
`component of each diversity branch is weighted by quality
`estimate QE of the particular branch 50, 51. In a steep
`weighting only the best Signal component or the best Signal
`components are selected for detector 29 on the basis of
`quality estimate QE. After combining the Signal is conveyed
`to ML detection means 29. Also outputs 34 of generating
`means 27 of the impulse response autocorrelation taps are
`combined by means 46 e.g. by Summing or averaging and
`when So desired by multiplying the Signals by a Suitable
`constant. In combining diversity branches 50, 51 and auto
`correlation taps 34 it is preferable to combine only the
`symbols or the bits that correspond with one another in time.
`The output of means 46 is also conveyed to ML detection
`means 29. Such a Solution is particularly useful because a
`greater amplification of the Signal is achieved when the
`Signal components are combined before detection.
`The Solutions of the invention can be implemented par
`ticularly regarding digital Signal processing with e.g. ASIC
`or VLSI circuits. The functions to be performed are prefer
`ably implemented as programs based on microprocessor
`technology.
`Even though the invention has been described above with
`reference to the example of the accompanying drawings, it
`is obvious that the invention is not restricted to it but can be
`modified in various ways within the scope of the inventive
`idea disclosed in the attached claims.
`What is claimed is:
`1. A diversity combining method in a digital radio System
`receiver (1-4), in which receiver matched filtering and
`maximum likelihood detection are used and an estimated
`channel impulse response and autocorrelation taps of the
`impulse response are generated, and in which radio System
`Substantially all Signal processing occurs as Symbols and a
`desired signal comprises a predetermined sequence (12),
`characterized in that
`
`Another preferable way of calculating quality estimate QE
`according to formula (7) is to divide entire signal energy E
`by entire noise energy VAR, when the value of divisor Kin
`formula (4) is one, and to avoid unnecessary dividing as
`formulas (4) and (5) comprise in this case a Substantially
`equal amount of elements to be summed. When the receiver
`comprises Several diversity branches, the Signal components
`of the different diversity branches and the autocorrelation
`taps of the estimated impulse response are weighted with
`quality estimate QE of each branch. The weighting prefer
`ably occurs by multiplying the Signal components in the
`matched filter by the autocorrelation taps of the estimated
`impulse response as they are being generated.
`Quality estimate QE is preferably calculated Separately
`for each burst as the connection quality differs greatly even
`during a short time.
`Let us now examine in greater detail the receiver of the
`cellular radio System of the invention, the block diagram of
`which is shown with essential parts in FIG. 3. Both a base
`Station and a Subscriber terminal can function as the receiver
`of the invention. The receiver comprises antenna 21 for
`conveying a received desired signal to radio frequency parts
`22 in which the Signal is converted into an intermediate
`frequency. From the radio frequency parts the Signal is
`conveyed into conversion means 23 in which the Signal is
`converted from analog into digital form. The digital Signal
`propagates to preprocessing means 24 where the Signal can
`e.g. be filtered, a DC offset can be removed from it, an
`automatic amplification of the digital Signal can be con
`trolled and the signal can be demodulated. Filter 25 matched
`to the channel restores the Signal distorted on the channel to
`the original data flow with a low symbol error probability.
`The channel impulse response estimate and its effective
`value are generated by means 26. Autocorrelation taps 34 of
`the estimated channel impulse response are generated from
`the impulse response information using means 27.
`In the digital radio System the channel impulse response
`is described with a number comprising N symbols. The
`channel impulse response usually comprises five Symbols,
`i.e. N obtains the value 5. Quality estimate 32 (QE) is
`calculated with the method of the invention using means 28
`comprising means 28a, 28b and 28c. Reference signal
`means 28a are used for generating reference signal 30 (YR)
`from the estimated channel impulse response and the pre
`determined Sequence comprised in the Signal. Interference
`means 28b generate interference strength 31 from the dif
`ference between reference Signal YR and the received pre
`determined Sequence. Quality means 28c generate quality
`estimate QE of the connection of the invention in Such a
`manner that the quality estimate is reversed proportional to
`the interference connected to the desired signal. The receiver
`is also arranged to correct the offset, or the Signal, of
`reference signal 30 and the predetermined Sequence, i.e. the
`transfer in time of the Symbols in relation to one another
`caused by propagation delay. Finally ML detection means 29
`of the receiver, preferably a Viterbi detector, receive output
`33 of matched filter 25 i.e. the different sequences of the
`received burst shown in FIG. 2 and autocorrelation taps 34
`of the channel impulse response from means 27. In this
`receiver solution both autocorrelation taps 34 of the impulse
`response in means 27 and the desired signal in matched filter
`
`15
`
`25
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`IPR2020-00038
`MM EX1030, Page 6
`
`
`
`7
`a reference Signal (30) is generated from the estimated
`channel impulse response and the predetermined
`Sequence (12) by convolution;
`an interference strength (31) connected to the desired
`Signal is generated using the differences of the refer
`ence Signal (30) and the predetermined sequence (12)
`received from the channel;
`a strength value (35) of the desired signal is generated,
`whereby a quality estimate (32) is generated by divid
`ing the strength value (35) of the desired signal by the
`interference strength (31) of the desired signal; and
`diversity combining is performed in Such a manner that
`the symbols of the different branches (50, 51) corre
`sponding with each other in the time domain are
`combined, and the outputs (33) of the matched filters
`(25) of each branch (50, 51) and the autocorrelation
`taps of the impulse response are weighted with the
`quality estimate (32) of each branch (50, 51).
`2. A method as claimed in claim 1, characterized in that
`the diversity combining occurs before the maximum likeli
`hood detection.
`3. A method as claimed in claim 1, characterized in that
`the interference strength (31) associated with the desired
`Signal is preferably generated in a variance-like or a Standard
`deviation-like fashion from the differences between the
`reference signal (31) and the predetermined sequence (12) of
`the desired Signal and
`the strength value (35) of the desired signal is generated
`as the Sum, the quadratic Sum or another corresponding
`power Sum of the Symbols Strengths of the reference
`signal (30), the desired signal or the taps (34) of the
`estimated channel impulse response.
`4. A method as claimed in claim 1, characterized in that
`when the transmission occurs in bursts the quality estimate
`(32) is calculated Separately for each received burst.
`5. A method as claimed in claim 1, characterized in that
`when the radio system is the GSM system, the predeter
`mined sequence (12) is the training sequence of a normal
`burst of the GSM system.
`6. A receiver (10–13) in a digital radio system, the
`receiver comprising a matched filter (25), diversity branches
`(50, 51) and a maximum likelihood detector (29), the
`receiver being arranged to generate an estimated channel
`impulse response and autocorrelation taps (34) of the
`impulse response, and in which radio System a desired signal
`comprises a predetermined sequence (12) and in which radio
`System signal processing is arranged to occur as Symbols,
`characterized in that the receiver (10–13) comprises
`
`15
`
`25
`
`35
`
`40
`
`45
`
`US 6,445,757 B1
`
`8
`reference Signal means (28a) for generating a reference
`signal (30) from the estimated channel impulse
`response and the predetermined sequence (12) by con
`Volution;
`interference means (28c) for generating the interference
`Strength (31) associated with the desired signal using
`the differences of the reference signal (30) and the
`predetermined sequence (12) received from the chan
`nel;
`the receiver is arranged to generate a strength value (35)
`of the desired signal and quality means (28b) are
`arranged to generate a quality estimate (32) by dividing
`the strength value (35) of the desired signal by the
`interference strength (31) of the desired signal; and
`combining means (45, 46) of the diversity branches (50,
`51) combine the symbols of the different branches (50,
`51) corresponding to each other in the time domain, and
`that the receiver is arranged to weight the matched filter
`(25) outputs (33) of each branch (50, 51) and the
`autocorrelation taps of the impulse response with the
`quality estimate of each branch (50, 51).
`7. A receiver (10–13) as claimed in claim 6, characterized
`in that the combining means (45 and 46) of the diversity
`branches (50, 51) are located before the maximum likeli
`hood detection means (29).
`8. A receiver as claimed in claim 6, characterized in that
`the interference means (28c) are arranged to generate the
`interference strength (31) associated with the desired signal
`preferably in a variance-like or Standard deviation-like fash
`ion from the differences between the reference signal (30)
`and the predetermined Sequence (12) of the desired signal
`and
`the receiver is arranged to generate the strength value (35)
`of the desired signal using the Sum, the quadratic Sum
`or another corresponding power Sum of the Symbols
`Strengths of the reference signal (30), the desired signal
`or the taps (34) of the estimated channel impulse
`response.
`9. A receiver as claimed in claim 6, characterized in that
`when the transmission occurs in bursts the receiver is
`arranged to generate the quality estimate (32) separately for
`each received burst.
`10. A receiver as claimed in claim 6, characterized in that
`when the radio system is the GSM system the predetermined
`Sequence (12) is the training sequence of a normal burst of
`the GSM system.
`
`IPR2020-00038
`MM EX1030, Page 7
`
`