United States Patent
`
`[19]
`
`[11] Patent Number:
`
`5,742,635
`
`Sanderford, Jr.
`
`[45] Date of Patent:
`
`Apr. 21, 1998
`
`USOO5742635A
`
`[54] ENHANCED TIME OF ARRIVAL METHOD
`
`[75]
`
`Inventor: H. Britton Sanderford, J:2, New
`Orleans, La.
`
`Primary Examiner—DaVid C. Cain
`Attomey, Agent, or Firm—Oblon, Spivak, McClel1and,
`Maier & Neustadt, P.C.
`
`[73] Assignee: Sanconix, Inc.. New Orleans. La.
`
`[57]
`
`ABSTRACT
`
`[21] Appl. No.: 238,326
`
`[22] Filed:
`
`May 5, 1994
`
`Int. Cl.‘ ............................... H04B 15/00; GO1S 3/02
`[51]
`[52] US. Cl.
`......................... .. 375/200; 342/450; 342/463
`[58] Field of Search ............................. 375/200; 342/450,
`342/463
`
`[56]
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`A method for improving a radio location system based on
`time—of—arriva1. Time-of-arrival radio location systems are
`
`limited in ultimate accuracy by signal-to—noise ratio and by
`the time varying multipath environment in which they must
`
`operate. These systems time stamp some feature of an
`
`arriving signal, from a transmitter, in order to calculate a
`range or a hyperbolic line-of—position. Some feature of the
`received signal must be identified which can provide repeat-
`able readings, in order for a time-of-arrival system to be
`reliable. The present invention teaches techniques which
`maintain a high signal—to—noise ratio while identifying a
`feature of the receivedsignal which is least aifected by
`multipath. The technique uses correlation peak/envelope
`information to estimate the leading edge of the correlation
`function, then enhances discrete samples at the leading edge
`of the correlation function to yield high signal—to-noise ratio
`readings. The present invention can reduce required trans-
`mitted bandwidth, increase system resolution and accuracy
`by twenty to one, and maintain high message traflic through-
`
`Pu‘-
`
`41 Claims, 5 Drawing Sheets
`
`....................... 343/6.5
`7/1976 Couvillon et a1.
`3,969,725
`.. 343/103
`8/1979 Michaels et al.
`4,166,275
`343/12
`11/1932 Kingston et al.
`4,357,610
`343/7.3
`6/1984 Koshio et al.
`.
`4,455,556
`364/450
`5/1985 Keearns ........
`4,520,445
`364/450
`4,558,418 12/1985 Keearns
`.... .. 375/1
`4,644,560
`2/1987 Torre et al.
`.. 342/132
`4,758,839
`7/1988 Goebel et al.
`...... 375/1
`4,972,431
`11/1990 Keegan .........
`375/95
`4,972,441
`11/1990 Roberts et al.
`.. 342/465
`5,017,930
`5/1991 Stoltz et al.
`.......
`...... 375/1
`5,093,841
`3/1992 Vancraeynest
`.. 342/450
`5,119,104
`6/1992 Heller ................
`375/96
`5,148,452
`9/1992 Kennedy et al.
`..
`...... 375/1
`5,179,573
`1/1993 Paradise ........
`5,266,953
`11/1993 Kelly et al.
`............................... 342/47
`OTHER PUBLICATIONS
`
`
`
`.
`.
`
`.
`
`Enge, Per K., Bandwidth Selection for Urban Radionaviga-
`tion Systems Worcester Polytechnic Institute, Mar. 25. 1988.
`Enge. Per K., Bandwidth Selectionfor Urban Radionaviga-
`tion System (Addendum), Worcester Polytechnic Institution,
`Jul. 29, 1988.
`
`NORMAL BW
`
`5"‘
`
`NORMAL
`INTEGRATE
`TIME
`
`~/~50?
`
`
`
`RECEIVED SIGNAL
`CORRELATION
`STRENGTH INDICATION
`
`
`INCREASED MEASUREMENT
`OR
`QUIETING DETECTION
`
`OR
`PLL DETECTION
`
`
`
`
`
`
`OPTIONAL BANDWIDTH
`COMPRESSION
`
`
`
`ADVANCE
`CHIP CODE
`
`PROCESSOR
`Y
`
`
`
`GENERATOR
`READIWRITE CHI3 CODE POSITION
`CPU OR LOGIC
`
`
`
`
`
`RECEIVED
`I.F.
`
`PETITIONERS 1003-0001
`
`

`
`FIG. 1
`
`RECEIVED
`I SIGNAL
`
`AMPLITUDE
`
`104
`
`A
`/=I..g».\
`
`101
`
`\
`
`/
`
`/
`
`/
`
`J
`
`\
`
`\
`
`V» 102
`
`\__W__,
`1 CHIP TIME
`
`—-—— TYPICAL DELAY SPREAD OF RECEIVED SIGNAL
`— — — CORRELATION FUNCTION
`.
`-------- --
`EFFECTS OF BANDWIDTH COMPRESSION
`
`23‘
`En
`
`"'6
`E,
`
`g(‘F
`
`3661‘IzJdv
`
`U)
`E
`I;
`
`9:
`U!
`
`U,
`‘L1
`3
`‘ca
`us
`U1
`
`TIME
`
`PETITIONERS 1003-0002
`
`

`
`
`
`111919d'S'I1
`
`:9
`-S
`7‘
`g
`I;
`:3
`
`g;(D
`(‘D
`
`N.
`
`9»
`U-
`
`U,
`\l
`
`A 3
`
`°@0
`
`)
`U1
`
`PETITIONERS 1003-0003
`
`/\
`
`FIG. 2
`
`205
`
`203
`
`CORRELATION
`
`PEAK
`DETECTED AT
`NORMAL BW
`(-_- DATA Bw)
`
`201
`NOISE FLOOR
`PRIOR To
`CORRELATION
`AT NORMAL
`B_VI{3ATA B
`I‘
`W)
`
`’
`
`‘
`
`SIGNAL
`AMPLITUDE
`
`\___/
`202
`FIRST
`
`ARRIVAL
`
`SIGNAL
`(LOW SNR)
`
`FIRST ARRIVING SIGNAL
`
`AT REDUCED BW AND/OR
`
`INTEGRATED OVER A
`
`LONGER TIME INTERVAL
`
`TO INCREASE SNR
`
`/———~—-—-—-—\
`
`204
`EXTENDED
`INTEGRATE’
`_Fr‘(I§[iI(J)‘\‘;\I]3!ERBW
`NOISE FLOOR
`/-—\
`
`/
`
`203 SLOPE VECTOR
`
`THIRD
`SECOND
`FIRST
`READING READING READING
`205
`207
`
`
`
`

`
`US, Patent
`
`Apr. 21, 1998 A
`
`Sheet 3 of 5
`
`5,742,635
`
`FIG. 3
`
`FIRST READING SHOULD ESTABLISH AMBIENT NOISE
`
`FLOOR. (301)
`
`SECOND READING (dB) (302)
`
`———————————
`
`LOOK UP EMPIRICAL DATA TO
`
`YIELD MOST ACCURATE RESULT. (304)
`
`/ .
`
`PETITIONERS 1003-0004
`
`

`
`US; Patent
`
`Apr. 21, 1998
`
`Sheet 4 of 5
`
`5,742,635
`
`
`
` SEARCH FOR
`
`
`CORRELATIO
`
`401
`
`USE CORRELATION
`PEAK TO ESTIMATE
`
`POSITION OF LEADING EDGE
`
`402
`
`REDUCE BW AND/OR
`INCREASE INTEGRATION TIME TO
`ENHANCE SNR
`
`FIG. 4
`
`404
`
`MOVE CHIP CODE POSITION
`CORRESPONDING TO THE ESTIMATED
`LEADING EDGE OF CORRELATION FUNCTION
`
`(MOVE CHIP CODE POSITION BY ~ ONE CHIP TIME)
`
`409
`
`
`
`VERIFY CHIP POSITION IS
`PRIOR TO LEADING EDGE OF
`
`
`
`SAMPLE AND STORE
`
`CORRELATION
`RESULTS
`
`MOVE CHIP CODE
`POSTION BY A PORTION
`1/8 OF A CHIP TIME
`
`REPEAT STEPS 405 AND
`
`406 FOR A
`MULTIPLICITY OF SAMPLES
`
`407
`
`408
`
`USE THESE SAMPLES TO FORM
`
`A SLOPE OR FOR A TABLE LOOK-UP
`TO ENHANCE THE
`RESOLUTION OF THE T.O.A.
`MEASUREMENTS
`
`
`
`
`
`
`PETITIONERS 1003-0005
`
`CORRELATION FUNCTION
`
`
`
`410
`I
`
`MOVE CHIP POSITION
`
`FORWARD BY 1/8
`OF A CHIP TIME
`I
`'
`
`
`
`
`
`
`
`
`
`
`
`VERIFY
`
`THAT CORRELATION
`
`
`IMPROVEMENT IS > AUTO
`CORRELATION
`NOISE AND/OR
`IMPULSE
`NOISE
`
`412
`
`
`
`TIME STAMP
`CHIP CODE
`POSITION
`
`
`
`
`
`
`
`

`
`FIG‘ 5
`
`506
`
`
`
`RECEIVED SIGNAL
`STRENGTH INDICATION
`OR
`QUIETING DETECTION
`0R
`PLL DETECTION
`
`~
`
`NORMAL
`INTEGRATE
`TIME
`
`507
`
`CORRELATION
`INCREASED MEASUREMENT
`INTEGRATE |
`TIME
`I
`
`I
`,
`
`L _ _
`
`._ _J
`508
`
`9?
`Cl)
`
`,1
`9,
`1-!-
`g
`H
`
`.
`
`>
`3
`N
`,I-*
`._.
`:2
`00
`
`m
`
`5 U
`
`I
`
`RECFFIVED
`
`NORMAL BW
`504
`
`%
`
`501
`
`?
`.& | REDUCED |
`I
`BW
`I
`I
`I
`I_ % _ I
`’\z
`505
`
`.502
`_ _ _. ._ _.
`
`I‘ “
`
`7
`
`' 9'4
`'
`I A
`'
`I
`I
`I_ _ __ 9"_T'£"iAh
`
`510
`
`ob OPTIONAL BANDWIDTH
`f\/ COMPRESSION
`
`9..
`0-
`
`
`
`CHIPCODE
`GENERATOR
`
`ADvAN E
`_, TA
`PROCESSOR
`U]
`
`
`E toREADNVRITE CHIP CODE POSITION CPU OR LOGIC
`
`
`
`509
`
`“ex
`I.»
`U:
`
`PETITIONERS 1003-0006
`
`

`
`5,742,635
`
`1
`ENHANCED Tl1\'IE OF ARRIVAL METHOD
`
`BACKGROUND OF THE INVENTION
`
`This invention relates to radio location systems and more
`particularly to an improved time of arrival method using
`spread-spectrum modulation.
`
`DESCRIPTION OF THE RELEVANT ART
`
`The literature describes numerous radio location systems
`which are based on time-of-arrival information. These sys-
`tems are primarily concerned with identifying a correlation
`peak and tracking that peak. The correlation peak is time
`stamped to provide a radio ranging, round trip distance
`measurement. Alternatively. two or more receivers can be
`used to calculate a hyperbolic line-of-position. Additional
`time-of-arrival approaches using direct sequence spread
`spectrum in the prior art include a method for separating one
`satellite signal from other satellite signals through correla-
`tion of a locally generated precise-code sequence with
`encrypted precise-code signals, the use of clock recovery
`circuitry to cycle through M phases of a locally generated
`receiver clock to select the phase yielding the strongest
`signal, and the use of a digital correlator to determine the
`time of arrival of a received pseudonoise sequence. Pulsed
`techniques capable, after compensating for pulse time detec-
`tion errors. of adjusting estimated time-of-arrival as a func-
`tion of the measured pulse slope are also known in the prior
`art. Such systems operate in the presence of background
`noise and multipath reflections created by objects in the
`proximity of the receiver and transmitter. The multipath
`reflections cause errors in the time stamped reading gener-
`ated by such time—of-arrival systems. Existing prior art
`systems do not teach the determination of a correlation peak
`for estimating the leading edge. or the taking of additional
`samples at the leading edge to improve the accuracy of the
`time-of-arrival stamp.
`A long time goal in the prior art has been to reduce the
`effects of multipath on time-of-arrival systems. A typical
`solution is to use a narrower chip time which, in its familiar
`radar equivalent, relates to wider bandwidth. The wider
`bandwidth yields a finer feature in the received signal in
`order to increase accuracy and resolution of the tirne-of-
`arrival measurement. Also, the more narrow chip time signal
`is less aifected by later arriving multipath reflections. The
`wider bandwidth systems. however, make trade-ofls which
`are detrimental to the ultimate performance and/or viability
`of the system. As the correlation function becomes narrower,
`only a portion of the total received energy can be utilized
`This is due to the received signals following a delay spread
`echo envelope pattern which may last one to five micro-
`seconds. If a correlation function were used on the order of
`100 nanoseconds. then undesirable multipath echoes as well
`as the majority of the received signal is rejected in an
`amount proportional to the process gain of the system. This
`reduces the effective signal-to-noise ratio (SNR) of the
`system. Complex, multiple peak, rake detectors have been
`developed to reduce this effect on systems which are pri-
`marily intended for data reception. In a system which is
`primarily intended for radio location, however. the leading
`edge information typically is more important than subse-
`quent information, for producing an accurate time stamp.
`An additional disadvantage of narrow chip times is the
`resulting increased bandwidth. Commercial time-of-arrival
`radio systems must operate under Federal Communications
`Commission (FCC) regulations and guidelines. Systems
`using a great amount of bandwidth suffer a competitive
`
`2
`time-of-arrival, radio
`disadvantage when placed against
`location systems which can successfully use reduced band-
`width and achieve similar accuracies. In addition to the
`regulatory limitations placed by the FCC,
`the hardware
`implementation of very wide bandwidth systems presents
`additional problems. Such systems require receivers which
`have front-ends which are purposely wide enough to receive
`the entire incoming signal. Such front-ends are difficult and
`expensive to design with the desirable filter roll-off charac-
`teristics needed to defend such receivers from out—of-band
`interference. This can lead to intermodulation distortion and
`loss of desired signal. In addition. the wider the front—end
`bandwidth becomes, the easier it is to receive unintentional,
`fundamental or harmonic interference within the front—end
`pass-band. This also reduces effective signal-to-noise ratio
`and measurement accuracy, and leads to the loss of intended
`signals.
`In order for a radio location system to produce the most
`accurate time-of-arrival measurement,
`the radio location
`system identifies a repeatable feature in the received signal
`which is least affected by multipath. Multipath reflections
`travel a greater distance than a direct arriving signal. Since
`multiple signals take a longer time to arrive at a receiver’s
`sight, the leading edge radio energy should be used in order
`to time stamp the time-of-arrival of a received signal. Prior
`art techniques relying on leading edge detection include the
`use of zero crossing detection of the carrier wave for
`improving the accuracy of a tirne-of-flight time stamp;
`pulsed, radar-type systems for measuring distance between
`an interrogator and a transponder based on leading edges of
`an interrogation pulse and a reply pulse, and for estimating
`a leading edge for a navigation system; and the use of
`leading edge as well as late arriving signal information
`collected from a plurality of individual detectors in a radar
`altimeter context The prior art in the area of leading edge
`detection also includes methods for reducing multipath
`errors through the use of Kalman filters, and time-of-arrival
`trigger circuits responsive to the arrival of a direct path
`time-of-flight transmission. These prior art references do not
`teach the use of direct sequence spread spectrum for finding
`a correlation peak, requiring only one detector, and the
`subsequent use of the correlation peak for estimating the
`leading edge of the correlation function.
`The leading edge of the received signal is least aifected by
`multipath error. Unfortunately. the leading edge typically is
`not the highest signal-to-noise ratio portion of the signal.
`The literature teaches a number of methods which can be
`employed to optimize signal-to-noise ratio, including tau
`dither loops, delay lock loops, Costas loops, and the methods
`disclosed in U.S. Pat. No. 4.977,577. Although such meth-
`ods optimize signal-to-noise ratio,
`they generally do not
`track the leading edge of the received signal since the
`leading edge does not represent optimal signal-to-noise
`ratio.
`
`In the case of a weak received signal, the leading edge of
`the correlation function may be well below a minimum
`detectable signal level, whereas the corresponding correla-
`tion function peak could be adequate for decoding data
`alone. In addition to inability to detect the leading edge least
`affected by multipath, conventional tau dither loops, time
`delay lock loops, Costas loops, etc. produce a large amount
`of jitter when the received signal is weak. This jitter results
`in inaccurate tirne—0f—arrival measurements.
`
`SUMMARY OF THE INVENTION
`
`A general object of the invention is using the least amount
`of bandwidth possible to yield accurate time-of-arrival read-
`ings.
`
`10
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`65
`
`PETITIONERS 1003-0007
`
`

`
`5,742,635
`
`3
`Another object of the invention is enhanced accuracy
`time-of-arrival readings with minimal transmitted message
`dwell time to allow the greatest number of simultaneous
`users on a radio location system.
`An additional object of the invention is identifying and
`measuring a feature in a received signal which is least
`affected by multipath errors.
`A further object of the invention is obtaining accurate
`time-of-arrival readings with low signal-to-noise ratio. mini-
`mum detectable signals.
`A still further object of the invention is to provide
`enhanced time-of-arrival readings with resolution greater
`than the sub-chip step size searched.
`According to the present invention, as embodied and
`broadly described herein, an enhanced time of arrival
`method for use with a radio location system is provided. A
`receiver uses direct sequence spread-spectrum modulation
`to time stamp a received radio broadcast. The receiver
`receives a spread-spectrum modulated signal from a trans-
`mitter. The steps include determining,
`in the received
`spread-spectrum signal, a time position of a correlation peak
`and estimating. from the correlation peak. a leading edge of
`a correlation function. The steps further include moving a
`receiver chip code position. corresponding to the estimated
`leading edge of the correlation function, backward by about
`one chip time. relative to the chips in the received spread-
`spectrum signal. The receiver-chip-code position, relative to
`the chips in the received spread spectrum signal. is then
`moved forward by sub-chip increments until a correlation
`envelope reading exceeds a preset margin. The method uses
`the leading edge of the correlation function. as determined
`by the receiver-chip-code position when the correlation
`envelope reading exceeds the preset margin, to time stamp
`an arriving signal. The term “receiver-chip-code position” as
`used herein is the position of chips in a chip sequence
`generated at the receiver. The receiver chip sequence is used
`to despread the received spread-spectrum signal. as is well
`known in the art.
`
`The method, as broadly embodied herein, further includes
`comparing the leading edge to a level equaling a sum of an
`auto—correlation code noise and estimated background noise,
`and using the leading edge to time stamp the arriving signal
`only when the leading edge exceeds the sum. Signal-to-
`noise ratio may be enhanced by reducing the bandwidth of
`the receiver following estimation of the leading edge.
`Alternatively. signal-to-noise ratio may be enhanced by
`increasing the integration time.
`The method may also include taking a multiplicity of
`correlation envelope samples of the leading edge of the
`correlation function. each of the samples being separated by
`a portion of a chip time. The method then includes deriving
`slope information fiom this multiplicity of samples for
`enhancing the resolution of the time-of-arrival time stamp.
`The method may also include inputting the multiplicity of
`correlation envelope samples to a look—up table of best-fit
`results for increasing the resolution of the time-of-arrival
`time stamp.
`Additional objects and advantages of the invention are set
`forth in part in the description which follows, and in part are
`obvious from the description. or may be learned by practice
`of the invention. The objects and advantages of the invention
`also may be realized and attained by means of the instru-
`mentalities and combinations particularly pointed out in the
`appended claims.
`BRJEF DESCRIPTION OF THE DRAWINGS
`
`The accompanying drawings. which are incorporated in
`and constitute a part of the specification. illustrate preferred
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`together with the
`embodiments of the invention and.
`description, serve to explain the principles of the invention.
`FIG. 1 illustrates delay spread effect on correlation func-
`tion;
`FIG. 2 shows received signal strength output of receiver;
`FIG. 3 is a look—up table of signal increase over ambient
`noise vs. sub-chip position signal strength;
`FIG. 4 illustrates a receiver algorithm; and
`FIG. 5 is a receiver block diagram.
`
`THE
`DETAILED DESCRIPTION
`PREFERRED EMBODIMENTS
`
`Reference now is made in detail to the present preferred
`embodiments of the invention, examples of which are illus-
`trated in the accompanying drawings, wherein like reference
`numerals indicate like elements throughout
`the several
`views.
`
`As illustratively shown in FIG. 1, when a radio wave is
`transmitted. regardless of modulation or bandwidth,
`the
`radio wave propagates through a multiplicity of physical
`paths prior to being collected at a receiving antenna. The
`multipaths reverberate through a building or echo through a
`city and produce a delay spread profile. Information from
`this delay spread profile 101 is used to time stamp a feature
`of the incoming signal in order to provide a time-of-arrival
`measurement, or relative tirne—of-arrival measurement. In a
`direct sequence radio modulated system, the received cor-
`relation signal takes on the function as shown by the dashed
`lines 102. This function would typically provide a sharp,
`well-defined correlation peak 104 in the absence of multi-
`path effects. In the presence of severe multipath, however,
`the well-defined correlation peak can often be lost. The
`correlation function could indeed have two or more peaks.
`Further, in bandwidth compressed systems, the correlation
`function does not yield a well defined correlation peak but
`rather a rounded-off top 103 with no distinct peak to time
`stamp. Such bandwidth compressed correlation functions .
`previously reduced resolution in time—of-arrival readings.
`The instant invention, however, uses the correlation function
`leading edge which can still provide accurate readings.
`The receiver first steps a chip code generator by a portion
`of a chip time or a whole chip time in order to search for a
`correlation peak. The receiver then uses the correlation peak
`information to estimate the leading edge of the correlation
`function. The leading edge should be approximately one
`chip time prior to the correlation peak. The receiver then
`either reduces bandwidth and/or increases integration time
`and then moves the chip code position to the corresponding
`estimated leading edge of the correlation function. The
`receiver then samples and stores the correlation result,
`moves the code position by a portion of a chip time. and
`measures additional correlation results. The correlation
`results can be in the form of signal strength indication,
`quieting detection, or lock detection from control
`loop
`means. The receiver then uses the sample to form a slope
`which more accurately predicts the leading edge of the
`correlation function. Alternatively. the receiver can use these
`samples to provide pointers to a look-up table which stores
`results yielding enhanced-accuracy time of arrival measure-
`ments.
`
`Any one of a number of existing techniques available in
`the art can be used to determine the approximate location of
`a correlation peak. These include sliding correlators. serial
`search methods. and frequency or chip clock slewing. with
`or without conventional Costas loop or tau dither loop or
`
`PETITIONERS 1003-0008
`
`

`
`5,742,635
`
`5
`
`delay lock loop hardware. In the instant invention, a corre-
`lation peak can be determined in either a minimum detect-
`able signal (MDS) or a strong signal case. In the MDS case,
`a small signal increase above the noise floor may be detected
`indicating the presence of a correlation peak. The correlation
`peak can be used to estimate the leading edge of the
`correlation function. Alternatively, an intermediate step of
`increasing signal averaging or reducing bandwidth can
`enhance the correlation peak prior to making an estimation
`of the beginning point of the correlation function.
`In the strong signal case. the correlation peak can be used
`to estimate the leading edge of the correlation function.
`Alternatively, slope information from the correlation peak
`can be taken from one or more readings taken a portion of
`a chip prior to the peak to estimate the correlation function
`leading edge. If two correlation peaks appeared due to the
`results of multipath, then the leading correlation peak can be
`used to estimate the leading edge of the correlation function.
`Alternatively, an average can be calculated between the two
`correlation peaks in order to estimate the leading edge of the
`correlation function.
`
`The correlation peak is first sought as an indication
`because it represents the highest signal-to-noise ratio (SNR)
`feature of the received signal. As an alternative to using the
`peak, if stronger signals were available, then the leading
`edge correlation slope can be used directly or the entire
`correlation function envelope can be used to predict the
`leading edge of the correlation function.
`Any of the above techniques can be used to estimate the
`leading edge of the correlation function. The more accurate
`this estimation, the shorter the dwell time required when
`taking additional samples of the leading edge of the corre-
`lation function. Less accurate
`measurements force a
`
`broader range of search to guarantee that the leading edge is
`included within the search. More accurate initial measure-
`ments allow the range of the search to be reduced. Reducing
`the range of the search is desirable so that the on-air time of
`the transmitted signal can be minimized. Minimizing the
`on-air time of the transmitted signal results in a high volume
`message traflic system by reducing message collisions.
`In a high signal-to-noise ratio situation, the first leading
`edge data sample which exceeds the auto-correlation code
`noise plus anticipated background noise can be used as a first
`signal. To ensure that the signal-to-noise ratio of leading
`edge samples is equal to that of correlation peak samples. the
`receiver bandwidth is reduced or the baseband integration
`time is increased to compensate for the difference. This
`technique has the advantage of yielding similar accuracy
`readings in both the strong signal and the MDS cases.
`Typically, MDS cases are accepted as yielding inherently
`poor accuracy. The anticipated strength increase of the
`leading edge correlation function signal strength above the
`uncorrelated tails can be calculated as
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`for the first '/é chip step search. where N=number of chips in
`the code. The improvement needed to equalize the strong
`signal in the MDS case is therefore
`
`20log(N)—20log( ":1
`
`_ 2 ) .
`
`This ratio is provided by an associated reduction in band-
`width and/or increase in integration time. In the instant
`invention. this results in an increased integration time of 17
`to 1.
`
`65
`
`6
`Once the estimation of the correlation peak leading edge
`is calculated, and the signal to noise ratio increased, then
`additional samples can be taken to more accurately establish
`the leading edge of the correlation function.
`One of several expanded SNR search schemes can be
`employed. First, a search scheme can be used which antici-
`pates that the initial prediction of the leading edge of the
`correlation function purposely comes before the initial sig-
`nal leading edge by one or more sub-chip step samples. In
`this case. the first sample taken should be at the auto-
`correlation function code noise floor. The chip code clock of
`the receiver can then be advanced in sub-chip multiples in
`order to produce a contour of signal strength improvement
`or quieting detector output improvement. These measure-
`ments are then stored for later processing.
`Alternatively, the leading edge of the correlation function
`can be initially estimated in a manner to purposely force an
`initial measurement after the leading edge of the correlation
`function. In this case, the receiver’s chip code clock should
`be decremented in sub-chip steps and measurements of
`signal strength or quieting detector output can be made
`which have decreasing strength until they stabilize at the
`auto—correlation function’s code noise level.
`A third approach is to produce an initial estimation in
`between these previous two schemes. This approach makes
`a best estimate of the exact beginning of the leading edge of
`the correlation function. Then, based on that estimated
`beginning signal strength. the receiver’s code clock can be
`incremented or decremented in sub-chip multiples, depend-
`ing upon whether the initial reading was at
`the auto-
`correlation function’s code noise level or above that level.
`Alternatively, a form of a binary search based on signal
`strength could be used after an initial sample was taken.
`If two or more such samples were taken,
`then this
`information can be used to create a slope vector. If three or
`more samples were used,
`then these samples can be
`weighted to produce a slope vector. This weighting can take
`into consideration that
`the leading samples have lower
`signal-to-noise ratio but lower multipath error, and that later
`samples have higher signal-to-noise ratio but higher multi-
`path error. In the third alternative. three or more samples can
`be stored and then used as input to a look-up table which has
`been pre-assigned with best estimate results based on theo-
`retically derived or empirical data.
`When searching in a quantized step of a sub-chip
`multiple, time-of-arrival resolution is limited to :t:1/2 of the
`quantized sub-chip step size. Combining signal strength and
`envelope information produces a vector with far greater
`resolution. However, the achievable accuracy is limited by
`signal-to-noise ratio and multipath environment. Lastly, the
`multiplicity of envelope samples can be transferred to a
`central processing hub which collects time-of-arrival infor-
`mation from multiple receivers. This information can be
`combined by the hub to further reduce errors and increase
`accuracy.
`This technique is equally applicable to digital parallel
`correlators. Bandwidth can be reduced by weighting mul-
`tiple correlation results or arranging multiple correlation
`readings. or by other techniques as are known in the art. in
`order to improve signal to noise ratio. Only a portion of the
`received signal need be analyzed at this slower information
`rate so that message trafiic would not be greatly impacted.
`The sample clock of the parallel correlator can be shifted to
`one chip time leading the correlation peak. Then the sample
`clock can be shifted in ‘/is chip increments in a similar
`manner to the serial correlator’s chip code position. This
`makes serial and parallel correlation techniques equivalent.
`FIG. 2 illustrates a graph of signal amplitude versus time
`of a received signal strength indicator output typical of the
`
`PETITIONERS 1003-0009
`
`

`
`5,742,635
`
`7
`above techniques. The initial portion of the signal 201
`depicts the noise floor or the auto-correlation function code
`noise being measured at a resolution bandwidth approxi-
`mately equal to that of the data bandwidth. Such measure-
`ments can be made practically at up to three times the data
`rate bandwidth with a small sacrifice in SNR performance,
`but with the added advantage of decreased search time.
`The next
`time portion of the illustration depicts the
`leading edge, first arriving signal 202. This is the signal
`which is least affected by multipath, but is at a lower
`signal-to-noise ratio than the signal peak.
`The next portion of time indicates the correlation peak
`203 detected at the normal sampling bandwidth, approxi-
`mately that of the data.
`The correlation peak. or correlation envelope, or some
`combination thereof. is used to estimate the leading edge of
`the correlation function. To ensure that the MDS case can
`yield an accuracy similar to the strong signal case, the
`detector bandwidth is decreased and the integration time is
`increased by an appropriate amount so that the leading edge
`of the correlation function can have a greater signal-to-noise
`ratio to produce an enhanced accuracy time-of-arrival read-
`ing. If the radio location system was simultaneously trans-
`mitting data during the time when synchronization was
`being acquired. then the transmitter data modulation band-
`width likewise is reduced to allow all of the received signal
`energy to be detected.
`The next portion of time. depicted first reading 204,
`illustrates the noise floor being reduced, in the MDS case. by
`increasing the integration sample time and/or reducing the
`bandwidth prior to detection.
`The next period of time. illustrated by first arriving signal
`205. shows the expanded first arriving signal, which has
`been filtered and integrated over a longer time interval to
`increase the signal-to-noise ratio. Each of these samples is
`taken after the receiver’s chip code clock has been shifted a
`portion of a chip time. In the instant invention the chip code
`clock is shifted in ‘A; chip steps. The second reading 206 is
`taken once the chip code clock has been shifted a portion of
`a chip time. The third reading 207 is taken after the second
`reading 206 is complete and after the chip code clock has
`been shifted an additional portion of a chip time. Two or
`more of these sample measurements are then used to predict
`a slope vector 208 or provide look-up table information to
`increase the resolution of the time-of-arrival measurement.
`
`The information obtained at the leading edge of the
`correlation function is in two forms. The first form of
`
`information is determined by finding the position of the chip
`code position which experiences a signal strength increase
`above the noise floor. or above the auto-correlation noise.
`The other form of information is determined by noting the
`level of the increase in the signal strength from one sample
`to the next. If two samples were taken. then a simple vector
`can be calculated. Ifthree or more samples were taken. then
`a form of curve fitting algorithm may be used, since multi-
`point slope measurements are not necessarily a straight
`linear function. As an alternative to calculating the leading
`edge of the correlation based on slope information, a look—up
`table can be used. Such a table can be a two-dimensional
`array or can be a look-up of a multiplicity of readings which
`will form an N-dimensional array. where N equals the
`number of readings taken as the chip code position is moved
`a portion of a chip time. The table entries can be derived by
`numeric estimation techniques. Alternatively. empirical data
`can be collected from actual field experience using known
`position transrrritters and receivers.
`In this way. a first
`reading 301 and a second reading 302. as shown in FIG. 3.
`
`8
`can be bound to the most likely correct result. Further, the
`look-up table can be one of several selectable look—up tables.
`If the operating environment of a the transmitter were
`known. such as in-building, in-urban areas, in-rural area. etc.
`then look-up table entries can be designed, and later
`selected, for best-fit in that environment.
`FIG. 3 depicts a look-up table of signal increase over
`ambient noise versus sub-chip position signal strength. In
`the instant invention, a first reading 301 is taken in order to
`establish the ambient noise floor or the ambient auto-
`correlation noise floor. This step is not. however. essential to
`the invention. Once a reading is taken representative of the
`noise floor. this reading is used as a base line indication of
`signal improvement between t