United States Patent (19)
`Atkinson et al.
`
`w
`
`54 METHOD AND APPARATUS FOR
`DIVERSITY RECEPTION OF
`TIME-DISPERSED SIGNALS
`75 Inventors:
`Frederick G. Atkinson, Winfeld;
`Gerald P. Labedz, Chicago; Duane C.
`Rabe, Rolling Meadows; Joseph J.
`Schuler, Roselle; Alton P. Werronen,
`Palatine, all of Ill.
`73) Assignee: Motorola, Inc., Schaumburg, Ill.
`21) Appl. No.: 435,650
`22 Filed:
`Nov. 13, 1989
`51) Int. C. ........................ H04B 7/005; H04B 7/08
`52 U.S. Cl. ........................................ 375/13; 375/14;
`375/96; 37.5/100; 455/138
`58) Field of Search ..................... 375/13, 14, 96, 100,
`375/102, 103, 106; 455/137, 138, 139
`References Cited
`U.S. PATENT DOCUMENTS
`3,633,107 1/1972 Brady .................................. 455/37
`4,112,370 9/1978 Monsen ............................... 375/100
`4,271,525 6/1981 Watanabe .............................. 375/14
`4,281,411 7/1981 Bonn et al. .......................... 375/100
`4,328,585 5/1982 Monsen ............................... 375/100
`4,731,801 3/1988 Henriksson ......................... 375/100
`4,733,402 3/1988 Monsen ............................... 375/100
`4,829,543 5/1989 Borth et al. ........................... 375/83
`OTHER PUBLICATIONS
`G. Ungerboeck, "Adaptive Maximum Likelihood Re
`ceiver for Carrier-Modulated Data-Transmission Sys
`tems,' IEEE Transactions on Communications, vol.
`COM-22, No. 5, May 1974, pp. 624–636.
`G. D. Forney, “Maximum Likelihood Sequence Esti
`mation of Digital Sequences in the Presence of Inter
`
`(56)
`
`11
`45
`
`Patent Number:
`Date of Patent:
`
`5,031,193
`Jul. 9, 1991
`
`symbol Interference', IEEE Transactions on Informa
`tion Theory, vol. IT-18, No. 3, May, 1972, pp. 363-377.
`John G.
`Proakis,
`"Digital Communications',
`McGraw-Hill Book Company, 1983, pp. 357-386.
`Primary Examiner-Benedict V. Safourek
`Attorney, Agent, or Firm-Shawn B. Dempster; F. John
`Motsinger
`
`ABSTRACT
`57
`A method and appartus for diversity reception in a
`communication system wherein at least a dual branch
`receiver is provided with a stored replica of expected
`reference information that is correlated with the re
`ceived time-dispersed signals to obtain an estimate of
`the transmission channel's impulse response as seen by
`each branch, and determine, among other things, phase
`error between the branch local oscillators and the time
`dispersed signals. Matched filters are constructed which
`then coherently align the time-dispersed signals from
`each branch with that branch's local oscillator, also
`constituting the first part of the equalization. The diver
`sity processing stage may perform bit by bit selection on
`the re-aligned signals, maximal ratio combining of the
`re-aligned signals, or equal gain combining of the re
`aligned signals, following each by a sequence estimation
`which uses similarly selected or combined channel dis
`tortion compensation parameters to complete the equal
`ization process on the new signal. In digital modulated
`carrier systems, providing expected reference informa
`tion eliminates the neeed for carrier recovery feedback
`for each branch while performing part of the equaliza
`tion process.
`
`27 Claims, 3 Drawing Sheets
`
`
`
`600
`SIGNA QuaDRATURE
`DEMODULATOR
`
`
`
`A/D
`CONVERTER
`
`625
`
`630
`
`EOUALIZER
`
`
`
`SIGNA
`WEIGHTING
`
`635
`
`DECISION
`CIRCUIT
`
`
`
`645
`
`
`
`
`
`SYNC
`CORRELATION
`(STORED REFERENCE)
`
`
`
`RECEIVER
`GAN
`NFORMATION
`
`SIGNAL
`AWPLITUDE
`ESTATION
`
`
`
`620
`
`NTIAL
`TAP GAIN
`CACULATION
`
`TO
`DECISION
`DEVICE
`
`M
`S(n)
`TAP GAIN
`ADJUSTMENT 640
`ALGORTH
`615 BRANCH 2
`BRANCH N
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 1 of 10
`
`

`

`U.S. Patent
`
`July 9, 1991
`
`Sheet 1 of 3
`
`5,031,193
`
`
`
`
`
`
`
`
`
`
`
`
`
`SYMBOL
`ARRAY
`
`105
`
`
`
`SEOUENCE
`ESTMATION
`STAGE
`
`PRESING
`
`
`
`
`
`101
`
`GAN
`CONTROLLER
`
`
`
`
`
`
`
`STORED
`REFERENCE
`
`DEMODULATION/
`PROCESSING
`
`DVERSITY
`PROCESSOR
`
`
`
`TO
`DECISION
`
`103
`
`F. G. 4
`
`a
`
`515
`
`FI G. 1
`
`520
`
`COMPLEX
`SIGNAL
`
`
`
`105
`
`SPARA2
`
`
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 2 of 10
`
`

`

`U.S. Patent
`
`July 9
`, 1991
`
`Sheet 2 of 3
`
`,031,193
`5
`
`
`
`
`
`|JÄÄV?g DWD
`
`? WNWdS
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`NETTO?I NOO
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 3 of 10
`
`

`

`US. Patent
`
`m
`
`e
`
`3
`
`5,031,193
`
`2
`
`mfinwmfi.‘
`
`
` 3253::go;3;:2.32:A2225mm”.v22.22:.oz;
`
`
`m:3552:28«25:83:..9,m;
`
`
`
`2229.35333a)55231:.w335handcom
`
`
`
`02:12;_52:5385®.QNBamum
`
`
`
`
`
`f2.3“332:532.3uN52:3m;.35...3655232.
`
`210253
`
`
`
`
`
`zo:<._:3<o22.3233zo_._.<x~_omz_
`
`cum
`
`ERICSSON V. UNILOC
`
`EX. 1017 / Page 4 of 10
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 4 of 10
`
`
`
`
`
`

`

`10
`
`15
`
`45
`
`1.
`
`METHOD AND APPARATUS FOR DIVERSITY
`RECEPTION OF TIME-DSPERSED SIGNALS
`
`5
`
`TECHNICAL FIELD OF INVENTION
`This invention relates generally to diversity receivers
`in communication systems and more specifically to
`receivers providing diversity reception for time-dis
`persed signals in communication systems.
`BACKGROUND OF THE INVENTION
`Enhanced signal detection in a time-dispersive me
`dium generally requires a receiver to perform some type
`of echo signal equalization on the received time-dis
`persed signals to produce an output which has a better
`output than would result from allowing the echoes to
`interfere with one another. One such equalization tech
`nique used in a digital radio Time Division Multiple
`Access (TDMA) system is described in instant assign
`ee's U.S. Pat. No. 4,829,543 entitled “Phase-Coherent
`TDMA Quadrature Receiver for Multipath Fading
`Channels' filed on behalf of Borth et al.
`The Borth et al. invention describes a phase coherent
`method for demodulating a Quadrature Phase Shift
`25
`Keyed (QPSK) radio signal that is subjected to multi
`path fading. Equalization is facilitated by correlating a
`stored training sequence, known to the receiver, against
`the incoming signal, and using the resulting correlation
`to remove the phase difference between the incoming
`signal and the receiver's local oscillator, effecting co
`herent detection. Equalization can then proceed.
`Other techniques have been proposed for dealing
`with the intersymbol interference which can be gener
`ated in a transmitted signal by a time-dispersive trans
`35
`mission channel. Such receivers are described in
`"Adaptive Maximum Likelihood Receiver for Carrier
`Modulated Data-Transmission Systems', authored by
`G. Ungerboeck, IEEE Transactions on Communica
`tions, Vol. COM-22, No. 5, May 1974, pp. 624–636, and
`"Maximum Likelihood Sequence Estimation of Digital
`Sequences in the Presence of Intersymbol Interfer
`ence', authored by G. D. Forney, IEEE Transactions
`on Information Theory, Vol IT-18, No. 3, May, 1972,
`pp. 363-377.
`However, in high data rate systems where transmis
`sion is through a severely delay-spread radio channel,
`single branch-single receiver equalization may fail to
`provide adequate time-dispersed distortion (multi-ray
`fading) correction. For example, practical implementa
`50
`tions of equalizing receivers may have imperfect esti
`mates of the critical error signal in the case of decision
`feedback equalization, or imperfect estimates of the
`transmission channel's impulse response in some other
`equalization schemes.
`55
`Therefore, diversity reception (the same signal re
`ceived on multiple branches-which may be on differ
`ent antennas, or on a single antenna at different times, or
`made in other ways, as is well known in the art) is typi
`cally necessary to sufficiently reduce the effect of multi
`ray fading. One such receiver is described in U.S. Pat.
`No. 4,271,525 entitled, "Adaptive Diversity Receiver
`For Digital Communications'. This patent describes an
`adaptive diversity receiver using an adaptive transver
`sal filter for each receiver branch, followed by a deci
`65
`sion feedback equalizer. The tapgains of the transversal
`filters are updated via feedback from the output of the
`equalizer, and other points in the receiver.
`
`5,031, 193
`2
`U.S. Pat. No. 4,731,801 entitled "Method For The
`Reception And Detection Of Digital Signals' discloses
`an improvement over U.S. Pat. No. 4,271,525 and other
`prior art by improving reception in highly dispersive
`transmission paths using coherent demodulation. This
`invention uses a technique wherein the output of the bit
`decision circuitry becomes a basis for calculating a
`correction signal. A reference carrier, resulting from
`summing the quadrature baseband signals and the in
`phase baseband signals, is fed back to the local oscillator
`of quadrature demodulators which in turn compensates
`the phase difference between the received signals and
`the receiver's local oscillator to facilitate coherent de
`modulation.
`However, inventions such as described in U.S. Pat.
`No. 4,271,525 require a set of adaptive transversal fil
`ters, one for each receiver branch, in addition to the
`equalization circuitry. Inventions such as U.S. Pat. No.
`4,731,801 require complex circuitry to phase shift the
`signal in each diversity branch, and, more importantly,
`cannot arrive at the correct phase adjustments quickly
`enough to be useful in, for example, TDMA systems
`characterized by information which is received, and
`must be corrected, in short bursts separated by rela
`tively long periods of time. During these long periods,
`signal phases in multi-ray fading channels can change
`radically relative to the receiver's local oscillator.
`Accordingly, there exists a need for a reduced com
`plexity receiver that performs diversity reception on
`continuous, or non-continuous, high speed digital sig
`nals and is capable of substantially reducing effects of
`both flat fading and multi-ray, dispersive fading due to
`time-dispersive transmission mediums.
`SUMMARY OF THE INVENTION
`These needs and others are substantially met by the
`method and apparatus for diversity reception of time
`dispersed signals in communication systems disclosed
`below. The described method comprises correlating a
`first time-dispersed signal received on a first receiver
`branch against a known reference, resulting in a first
`correlation signal, and correlating a second time-dis
`persed signal received on at least a second receiver
`branch, against the known reference, resulting in a sec
`ond correlation signal, then, using the correlation sig
`nals, re-aligning the first time dispersed signal and the
`second time dispersed signal to the known receiver
`reference signal and the branch's local oscillator, result
`ing in a first aligned signal and a second aligned signal,
`and generating a resulting signal in view of the first
`aligned signal and the second aligned signal.
`The known reference signal is located in a stored
`look-up table (containing multiple synchronizing se
`quences as is appropriate in the case of either a Time
`Division Multiple Access (TDMA) system or a Fre
`quency Domain Multiple Access (FDMA) system with
`embedded reference signals). The correlation deter
`mines, among other things, an estimate of the radio
`transmission channel's impulse response. After correla
`tion is complete, a matched filter, usually a transversal
`filter having taps derived from the estimated channel
`impulse response, is used to perform a convolution on
`the time-dispersed received signals, thereby performing
`a phase equalization. The phase equalization substan
`tially compensates for the phase difference between the
`received time-dispersed signal and the local oscillator in
`each receiver branch.
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 5 of 10
`
`

`

`15
`
`5,031, 193
`3
`4.
`The resulting processed signal samples from each
`either TDMA or FDMA systems, it may be necessary
`branch are chosen, along with additional channel im
`to adjust the receiver branch's local oscillator, or pro
`pulse response-related parameters (s-parameters), in a
`cess the received data by some other means, if this is not
`sample selection technique and then all fed into a se
`the case.
`quence estimator to complete the equalization process
`A more detailed block diagram of the invention is
`on the newly-created signal. Various signal-combining
`shown in FIG. 2. The first diversity receiver branch
`techniques such as the well-known maximal ratio com
`(100) and the second diversity receiver branch (101)
`bining or equal gain combining techniques may also
`receive signals transmitted from the same point, but
`serve as acceptable diversity techniques to combine the
`which have become time-dispersed, in different ways, in
`phase-compensated signals and channel impulse-related
`traveling from a transmitter to each branch of the re
`parameters useful for equalization.
`ceiver. These signals are processed by techniques well
`known in the art by being amplified and mixed in the
`BRIEF DESCRIPTION OF THE DRAWINGS
`intermediate frequency stages (200 and 201). Thereafter
`FIG. 1 is a block diagram generally depicting the
`the signals are demodulated (transformed to baseband)
`invention.
`into in-phase signals, I1 and I2, and quadrature phase
`FIG. 2 is a more detailed block diagram of the inven
`signals Q1 and Q2 by quadrature demodulators (202 and
`tion as it applies to quadrature reception of digital sig
`203) as understood in the art, whose injection comes
`nals.
`from local oscillators (208 and 209). As already known
`FIG. 3 is a block diagram generally depicting the
`in the art, the same local oscillator may be used for
`diversity processor using bit by bit selection diversity in
`multiple branches.
`accordance with the preferred embodiment of the in
`Each of the in-phase signals and quadrature signals of
`vention.
`each branch are passed through their respective low
`FIG. 4 is a block diagram generally depicting the
`pass filters (220), (230), (240), (250) and sampled by
`diversity processor employing a method of maximal
`analog to digital (A/D) converters (204), (205), (206),
`ratio combing in accordance with the invention.
`25
`(207). Each of these signal samples are brought into
`FIG. 5 is a block diagram generally depicting a sec
`correlators so that sampled I1 is fed into correlator no. 1
`ond method of maximal ratio combing in accordance
`(210), sampled I2 is fed into correlator no. 2 (212), sam
`with the invention.
`pled Q1 is also fed into correlator no. 1 (210), and san
`FIG. 6 is a block diagram generally depicting the
`pled Q2 is also fed into correlator no. 2 (212). Stored in
`invention using adaptive linear equalization.
`the correlators' memory, or any other suitable location,
`is a known reference (104), such as a replica of an ex
`DESCRIPTION OF THE PREFERRED
`pected bit or pattern, that comprises amplitude and
`EMBODIMENT
`phase information substantially similar to an ideal signal
`FIG. 1 generally depicts the invention as adapted to
`that has not undergone time-dispersive fading.
`an equalizing receiver described by G. Ungerboeck and
`35
`The sampled signals are correlated to this stored
`referenced above. Quadrature Phase Shift Keying
`reference information resulting in several metrics.
`(QPSK) is employed. However, other digital multi
`These include an estimate of the impulse response of the
`dimensional signaling, such as GMSK, may equiva
`transmission path (or a measure of the time-dispersed
`lently be employed. As shown, the invention comprises
`distortion of the environment), an estimate of the fre
`a first receiver branch (100) and at least a second re
`quency/phase offsets relative to the known reference,
`ceiver branch (101), each branch comprising a radio
`and some measurement of signal strength. The signal
`frequency demodulation stage and only part of the pro
`strength measurements are then input to a gain control
`cessing necessary to equalize a time-dispersed received
`ler (106), which in turn individually adjusts the gain of
`signal (102 and 103). Although the description of the
`each branch in accordance with a predetermined gain
`invention will refer to a receiver having two branches,
`45
`equation suitable to the application. As is already
`the disclosed invention readily applies to receivers hav
`known in the art, the gain of each IF stage (200 and 201)
`ing N branches.
`of each branch (100 and 101) may be simultaneously
`Both receiver branches are provided with a stored
`adjusted to produce the same gain on each branch.
`expected reference sequence (104) to allow post demod
`The other metrics are used to construct a channel
`ulation correlation between received time-dispersed
`50
`matched filter for each branch, as well known in the art.
`signals and the reference sequence using known correla
`After passing the signal samples through each branch's
`tion techniques. The correlation information provides
`matched filter (214 and 215), the output signals have
`synchronization information, and parameters from
`had the effects of time-dispersed distortion and phase
`which a channel matched filter can be made. The re
`error with each branch's local oscillator substantially
`ceived time-dispersed signal is processed through this
`55
`matched filter and the resulting phase re-aligned signal
`removed.
`For example, in a radio TDMA communication sys
`from each branch is then used by the diversity decision
`tem utilizing a synchronization sequence in an assigned
`block (105) to generate an output signal. The correla
`timeslot of a transmitted burst, a stored replica of a
`tion further provides information which can be used in
`synchronization sequence expected by the receiver
`an appropriate gain controller (106) to keep the re
`would be the data stored in the look-up table. The re
`ceived signal within a certain voltage range. No carrier
`recovery circuit (phase compensation feedback) is nec
`ceiver correlates received signals from both branches
`(100 and 101) against the stored expected synchroniza
`essary to correct the modulated time-dispersed signals
`tion pattern to determine the time-dispersed profile
`in relation to the local oscillator in each branch, pro
`vided that the transmission channel's impulse response,
`models (channel impulse response) of each branch and
`65
`hence the correlation and hence the matched filter taps,
`then calculates the matched filter coefficients based on
`do not change appreciably during the period of time
`samples of the resulting correlation, or channel impulse
`response estimate.
`during which the data to be processed is received. In
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 6 of 10
`
`

`

`O
`
`15
`
`5,031, 193
`6
`5
`FIG. 4 depicts another diversity processing tech
`The phase-aligned signals out of the matched filter
`nique, instead using combining analogous to maximal
`(214) in the first branch (100) are represented by IM1 and
`ratio combining of the aligned signals, whereby the
`QM while the phase-aligned signals out of the matched
`weighting factor is determined by received signal
`filter (215) in the second branch (101) are represented
`by IM2 and QM2. The diversity processor (105) then uses
`strengths in accordance with a technique described in
`instant assignee's patent application Ser. No.
`at least a portion of at least one of these aligned signals
`from both branches (and other pertinent information
`07/358,325, filed May 26, 1989, entitled “Rapid Re
`ceived Signal Strength Indication' invented by Labedz.
`such as a measure of signal strength) to best determine
`the data of the originally transmitted signal.
`et al. The weighting factor is best derived from a sum
`mation of the squares of the quadrature components of
`The diversity processor (105) may use various tech
`niques to effectuate an optimum representation of the
`the energies at relative maxima of correlations between
`original transmitted signal. FIG. 3 depicts the diversity
`received echoes and a stored reference sequence. These
`processor using bit-by-bit selection diversity, wherein
`correlated energy measurements are integrated to deter
`mine the energy present among the multiple time-dis
`selected branch correlation parameters, called s-param
`persed echoes, and the resulting weighting factor is
`eters, and selected samples of the aligned signals are fed
`to a sequence estimator which generates a representa
`termed “received signal strength indicator'. However,
`a sample of or integration of several samples of the
`tion of the original transmitted signal.
`A more detailed explanation of s-parameter genera
`received signal's envelope could also be used.
`tion in receivers is given in the paper by Ungerboeck,
`The received signal strength indicator for the first
`branch (RSSI1) (400) is multiplied with the aligned
`referenced above, equation 17. Each branch derives its
`20
`signal from the first branch (AS1) using the multiplier
`own s-parameters from the correlation, based on the
`(410) and forming a weighted aligned signal for the first
`convolution of the channel's estimated impulse response
`and the impulse response of its respective matched fil
`branch. The received signal strength indicator for the
`second branch (RSSI2) (405) is multiplied with the
`te.
`The aligned signal for the first branch (AS1) and the
`aligned signal from the second branch (AS2) using the
`25
`multiplier (415) and forming a weighted aligned signal
`aligned signal for the second branch (AS2) are input
`into a processing stage (300). One sample per transmit
`for the second branch. These weighted signals are then
`summed (420) resulting in a signal comprised of
`ted data symbol of each of the aligned signals is com
`pared to its appropriate transmitted data symbol sample
`weighted signals from both branches.
`The s-parameters, as described above, are processed
`of the other branch. The absolute values of the samples
`30
`are compared, and the actual sample with the greatest
`in a similar fashion. The received signal strength indica
`absolute value is put into a symbol array which will
`tor for the first branch (RSSI1) (400) is multiplied with
`the s-parameters from the first branch (s-para1) using
`later be passed to a sequence estimation stage (305),
`the multiplier (430) and forming a weighted set of s
`which comprises a sequence estimator as known in the
`art. Although the greatest absolute value is the basis for
`parameters from the first branch. The received signal
`strength indicator for the second branch (RSSI2) (405)
`selection in this embodiment, the lowest absolute value
`is multiplied with the s-parameters from the second
`or any other suitable basis may also be used.
`branch (s-para2) using the multiplier (440) and forming
`Furthermore, a counter for each branch is available
`a weighted set of s-parameters from the second branch.
`to record the number of samples selected from each
`branch that are put into the symbol array. When the last
`These weighted s-parameters are then summed (450),
`40
`symbol samples of the signal have been compared, the
`resulting in a signal comprised of weighted set of s
`parameters from both branches. This technique may
`counters are compared to determine which branch pro
`vided the most samples to the symbol array. The s
`also be applied where N receiving branches are used.
`Again, combining of the s-parameters may be elimi
`parameters (SS) from the branch providing the most
`samples to the symbol array are sent to the sequence
`nated in the receiver of Ungerboeck, but poorer perfor
`45
`estimator in the form of an s-parameter array. The s
`mance would result. A receiver utilizing a sequence
`parameters provide the sequence estimator with pro
`estimator not of the type described by Ungerboeck
`would not utilize s-parameters, but would still perform
`cessed intersymbol interference information. The se
`quence estimator then completes the equalization pro
`combination on a symbol sample basis following a
`matched filter, and appropriate combination of what
`CSS.
`ever distortion-compensating parameters may be uti
`Selection of the s-parameters may be eliminated, and
`lized by the sequence estimator.
`a set of s-parameters chosen at random from one of the
`In the case where weights from the signal's strength
`branches may be used in the receiver of Ungerboeck,
`but poorer performance would result. A receiver utiliz
`are not used, that is RSSI1 (400) and RSSI2 (415) are
`effectively set equal to 1, a technique analogous to equal
`ing a sequence estimator not of the type described by
`55
`gain combining results, and multipliers (410) and (415),
`Ungerboeck would not utilize s-parameters, but would
`and (430) and (440) are no longer necessary.
`still perform selection on a symbol sample basis follow
`FIG. 5 depicts another method of diversity incorpo
`ing a matched filter, and perform an equivalent selec
`tion of any channel distortion-compensating parameter
`rating a technique analogous to maximal ratio combin
`ing of the aligned signals using a signal strength indica
`which may be utilized by the sequence estimator.
`60
`As stated, the combining of the signals occurs in the
`tor (SSI) resulting from a determination of the signal
`strength measured at the intermediate frequency stages
`middle of the equalizer to facilitate diversity, inasmuch
`as the equalizer of this type in a single branch receiver
`(200 and 201) of each branch. This embodiment com
`bines the comples (in-phase and quadrature phase)
`may be considered to be the combination of the
`aligned signals from each branch before they are passed
`matched filter and the sequence estimator. The instant
`65
`through a complex signal de-multiplexer (520).
`invention need only duplicate the matched filter func
`A signal strength weighting equation (500) deter
`tion, but not the sequence estimator function in a diver
`sity receiver.
`mines the relative weight assigned each branch's
`
`35
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 7 of 10
`
`

`

`5
`
`10
`
`15
`
`5,031, 193
`7
`8
`aligned in-phase signal (IM1 and IM2) and each
`or combining, may occur after a feed-forward filter (as
`branch's aligned quadrature phase signal (QM1 and
`understood in the art) for each diversity branch, or any
`QM2). This equation weights IM1 and QM1 by
`other appropriate point within the equalizer block itself.
`RSSI 1/(RSSI1--RSSI2) and weights IM2 and QM2 by
`What we claim is:
`RSSI2/(RSSI1--RSSI2). The weighted in-phase signals
`1. A receiver for providing diversity reception com
`(ISSI1 and ISSI2) are summed (510) resulting in a com
`prised of:
`bined in-phase signal for both branches, and the
`(a) correlating means for generating:
`weighted quadrature signals (QM1 and QM2) are
`(i) a first correlation signal by digitally correlating
`summed (505) resulting in a combined quadrature signal
`a first received sampled time-dispersed signal
`for both channels. Equal gain combining of the signals
`against a known reference sequence; and
`or bits therein may also serve as a suitable diversity
`(ii) a second correlation signal by digitally correlat
`decision technique. Such equal gain combining for di
`ing a second received sampled time-dispersed
`versity reception would again involve setting the
`signal against the known reference sequence;
`RSSI and RSSI2 values equal to 1.
`(b) re-aligning means, operably coupled to the corre
`Although the preferred embodiment is suited for use
`lating means, for generating:
`in systems having high speed, noncontinuous signals
`(i) a first aligned signal by coherently re-aligning
`such as TDMA systems having short burst signals, an
`the first time-dispersed signal to the reference
`alternative embodiment of the invention may be appro
`sequence by using at least the first correlation
`priate when receiving sufficiently long data streams
`signal; and
`where the transmission channel impulse response appre
`(ii) a second aligned signal by coherently re-align
`ciably changes during the period of time during which
`ing the second time-dispersed signal to the refer
`the data to be processed is received.
`ence sequence by using at least the second corre
`FIG. 6 shows one branch of the present invention
`lation signal; and
`using an adaptive linear equalizer. In this embodiment,
`(c) signal generation means, operably coupled to
`an initial correlation is performed using the known
`25
`re-aligning means, for generating a digital output
`reference to estimate the channel impulse response
`signal derived from the first and second aligned
`(CIR) and initial equalizer tap gains (Ck(0)). Thereafter,
`signals.
`tap gains of the equalizer are adjusted using typical
`2. The receiver of claim 1 wherein the correlating
`adaptive linear equalization techniques (such as those
`means comprises means for generating:
`described in Digital Communications by John G. Proa
`30
`(a) the first correlation signal by determining a chan
`kis, McGraw-Hill Book Company 1983, on pages
`nel impulse response, using the known reference
`357-386) to continue re-aligning the received signal.
`sequence, for a receiver branch through which the
`Therefore, coherently re-aligning the received signal
`first received time-dispersed signal was received;
`includes at least correlating the signal to the known
`and
`reference (initially), then later adjusting the equalizer's
`35
`(b) the second correlation signal by determining a
`tap gains in accordance with known adaptive linear
`channel impulse response using the known refer
`equalization techniques.
`ence sequence for a receiver branch through which
`As shown, the received signal is passed through a
`the second received time-dispersed signal was re
`quadrature demodulator stage (600) and then sampled
`ceived.
`and digitized in the A/D converter stage (605) resulting
`40
`3. The receiver of claim 1 wherein the re-aligning
`in a complex signal (R(n)). This signal is correlated to
`means comprises a linear equalizer.
`the stored reference in the correlation stage (210) result
`4. The receiver of claim 1 wherein the re-aligning
`ing in the channel impulse response (CIR) which is used
`means comprises a decision feedback equalizer.
`in the initial equalizer tap gain (Ck(n)) calculation (615)
`5. The receiver of claim 1 wherein the signal genera
`and signal amplitude estimation (620). Signal amplitude
`tion means generates the digital output signal with bit
`estimation criteria contribute to the weighting factors
`by bit selection diversity signal combining techniques.
`determined in the signal weighting stage (625).
`6. The receiver of claim 1 wherein the signal genera
`As indicated, R(n) is also input to the equalizer (630)
`tion means generates the digital output signal with max
`wherein soft information (S(n)), as understood in the
`art, is generated. The soft information S(n) at the output 50 imal ratio signal combining techniques.
`7. The receiver of claim 6 wherein the maximal ratio
`of the equalizer is routed to a decision circuit (635)
`signal combining techniques comprise determining a
`wherein a tentative decision S(n) is made for the pur
`weighting factor from a Received Signal Strength Indi
`pose of updating the equalizer tap gains (640) as the
`cator, the Received Signal Strength Indicator being
`received signal is processed. Once a suitable S(n) signal
`determined from processed channel sounding tech
`is generated, it is weighted (625) and output to a sum
`55
`niques using a time-dispersal function to determine en
`ming junction (645), which combines some or all of
`ergy levels of received time-dispersed signals.
`both branch (in the case of a dual branch receiver)
`output signals prior to the final bit decision. Each
`8. The receiver of claim 6 wherein the maximal ratio
`signal combining techniques comprise generating a
`branch output signal may be appropriately weighted to
`meet either equal-gain or max-ratio combining criteria 60 weighting factor based on a measurement of signal
`strength measured at an intermediate frequency stage.
`or may be combined using bit by bit selection as previ
`9. The receiver of claim 1 wherein the signal genera
`ously disclosed herein.
`tion means generates the digital output signal by sum
`As appreciated by those skilled in the art, the present
`ming at least a portion of the first aligned signal with at
`invention may also be applied to receivers utilizing
`decision feedback equalizers or any appropriate non-lin
`least a portion of the second aligned signal to form a
`coherent equal gain output signal.
`ear equalizer. For example, S(n) may be soft informa
`tion derived from the decision feedback equalizer prior
`10. The receiver of claim 1 wherein the re-aligning
`to the decision circuit. Also, symbol sample selection,
`means comprises a channel matched filter.
`
`45
`
`65
`
`ERICSSON v. UNILOC
`Ex. 1017 / Page 8 of 10
`
`

`

`5
`
`10
`
`15
`
`5,031, 193
`9
`11. The receiver of claim 1 wherein the correlating
`means further generates:
`(a) first s-par