UNITED STATES PAl‘EN'I‘ ANNE) TRADEMARK OFFICE
`
`BEFORE TEE PATENT TRifAL AND AF?EAL EGARD
`
`K42“) ELECTRGNICS, LLC
`
`Patitianer
`
`v
`
`ESEORT, ENC.
`
`Pawn: Owner
`
`Cage 1P3. EQES-QGZAG
`
`US. Patent No) éflémfififlfi
`
`EXHIBIT
`
`b CI?”
`
`£7?an 10% '
`
`
`
`”RAES‘Exhibit 1,016, psi. 1
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 1
`IPR2013-00240
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`llllllllllllllllllllllIllllIllllllllllllllllllilllllllmllillllllIllllllll
`
`U5005305007A
`
`[19]
`United States Patent
`5,305,007
`[11] Patent Number:
`
`Orr et a1.
`[45] Date of Patent:
`Apr. 19, 1994
`
`[54} WIDEBAND RADAR DETECTOR
`
`[75}
`
`Inventor-i: Steven K. Orr, Lowland; John 13»
`Kuhn, West Chester, both of 01110
`
`5,033,019 7/ 1991 White .................................. 364/726
`
`...... 342/20
`5,068,663 11/ 1991 Valentine et al
`
`1/1992 Orr ........................... 342/20
`5,079,553
`................... 324/78
`5,099,194
`3/1992 Sanderson et a1.
`
`{73] Assignce: Cincinnati Microwave Corporation,
`Cincinnati, Ohio
`
`Primary Examinerw—John B. Sotomayor
`Attorney Agent, or Firm-wLimbach & Limbach
`
`[21] Appl. No.: 48,128
`
`[22] Filed:
`
`Apr. 13, 1993
`
`Egg;
`58
`[
`1
`[56]
`
`Int. C105 ........................................3.4,.22230415853£;g
`
`1" Id f S 11“"3412 20 98’ 99 104
`w 0
`earn:
`342/116455 2/26 ’1
`2’26 4: 228’
`’
`l
`'
`’
`‘
`’
`References Cited
`U.Sr PATENT DOCUMENTS
`
`4/1986 Grimsley et a1.
`4,581,769
`455/226
`4,622,553 11/1986 Baba et a1.
`342/91
`
`4,630,054 12/1986 Martinson
`342/20
`
`4,698,632 10/1987 Baba et a1.
`342/17
`4,723,125
`2/1988 Elleaume ..
`. 342/194
`
`4,750,215
`.455/226
`6/1988 Riggs .........
`
`. 342/194
`4,772,889 9/ 1988 Elleaume ..
`
`4,862,175
`8/1989 Biggs et a].
`342/20
`5/1990 Elleaume ............................. 342/ 194
`4,929,954
`
`[57]
`
`ABSTRACT
`
`A wideband radar detection apparatus includes a signal
`detection section, high rate signal processing section
`and a low rate signal processing section. The signal
`detection section sweeps througharange of preselected
`frequencies and generates an output signal having a pair
`of single cycle sinusoids for every detected signal. The
`output signal is provided to the high rate signal process»
`ing section and a Sliding Window Discrete Fourier
`Transform is performed thereon to generate a set of
`complex values that are related to the fundamental en-
`ergy content at consecutive points in the sweep. The
`low rate signal processing section controls sweep pa-
`rameters and also evaluates the complex values. 1f the
`magnitude of the complex vaiues exceed a predefined
`threshold, then an alert is indicated
`
`11 Claims, 6 Drawing Sheets
`
`10 Mhz IF
`
`10.525 Gm,
`+/~25 AM:
`24.15 Ghz
`-+/-50 Ma:
`15:
`34-7 0'12
`+/"500 HhZ MIXER
`10
`
`BYPASS AMP
`
`
`25
`K0 84W Vac
`
`
`1024 Mhz
`2nd
`10 Mhz IF & DETECTOR
`IF AMP
`MIXER
`
`
`10 Mhz UMITER, MP
`
`
`:2 QWRATURE
`
`DEIECVOR
`
`
`MN<¥QC>U§
`
`
`' 11.559 Ghz
`
`
`10.34 Mhz
`+/- 60 Mhz
`
`
`151 LOCAL
`2nd LOCAL
`
`
`OSCIUATOR
`OSCILLATOR
`
`
`
`Vac
`
`
`
`
`SWEEP 00/1/7301.
`
`5115EP
`HICROPROCESSOR. DSP CHIP
`
`
`
`CIRCUIT
`0R DEIECNON CIRCUIIRY
`
`
`
`STORAGE, PROCESSING AND
`
`DECISION LOGIC
`
`X K N
`
`a
`
`USER
`INPUT
`
`
`
`K40 Exhibit 1016, pg. 2
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 2
`IPR2013-00240
`
`

`

`US. Patent
`
`Apr. 19, 1994
`
`Sheetl of6
`
`5,305,007
`
`REVEH33Na)..2
`
`
`
`MQEESSQ.fi
`
`QEQEQ
`
`NM.
`
`mbbk>mm
`
`NM.
`
`N§~Q“
`
`333‘
`
`$3.5aninM«SEE
`
`QEQEQ.fi.3.23E
`
`mmmmS9%Q_“A0AI.AzA.
`
`
`55:useAE5:Nn+EN3:KSEafihmmmfinax
`
`8;3%ex
`
`bN
`
`n3?mmvmxm
`
`he.Na:2
`
`
`
`Ne:th\+
`
`New3.x.kNa}mNI\+
`
`NE@399X
`
`09>IE«£538xxmESEmo.
`.33Em4.63E3:whenE:on:\+
`
`ExtNewmmmi
`
`fig
`
`hi3
`
`sGE
`
`
`Q2<623.39qu“wuw‘kosm.
`98¢éoxmfimqm“
`
`m“
`
`
`
`M‘OmfiMUlethixESUSQ235meto3:0mud.
`
`SE28“3%.Swim
`
`hams
`
`K40 Exhibit 1016, pg. 3 '
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 3
`IPR2013-00240
`
`
`
`
`
`
`
`

`

`US. Patent
`
`Apr. 19, 1994
`
`Sheet 2 of 6
`
`5,305,007
`
`AUDIO OUTPUT OF QUADRAWRE
`DETECTOR
`
`L—t—al‘
`
`FIG. 2
`
`.5
`
`7.5
`
`P 2 m5
`
`.3 m3
`
`6 m3
`
`
`
`
`SWEEP
`CONTROL
`
`FIG. 4
`
`K40 Exhibit 1016, pg. 4
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 4
`IPR2013-00240
`
`

`

`US. Patent
`
`Apr. 19, 1994
`
`Sheet 3 of 6
`
`5,305,007
`
`8338:3333-3&8
`2%I
`I333.333
`
`:83
`
`3:33:33'
`
`3&8
`
`
`
`1W0IEUk
`
`9N
`
`
`33,630I83
`
`«33.535333.33.
`
`mmGE
`
`3.an-.333
`
`WK.ScbImnn
`
`330.3338
`
`3:8
`
`333330
`
`323133'
`
`3I333.
`LS0303.33...
`I,-
`II
`
`33833633I3&8
`muggw|..I333333
`3.3.333'3&8
`nmwmifi“3&8
`333033933I3&8
`33.3w.n:3&88383I3&8
`.z.3&8
`m3.96.5.3=r-we.33.3.$3338.83
`
`33.3333Joe? ,lmm83.33IWW3b$\°kIam..8333EmmaImm.63333336333
`
`
`
`u“.83333333333..3.3332323I333on333.6I8:3...3.3333823.3333I.Nw833.3.333.333.3333
`
`
`Quin»?QNcNW33N338
`
`hm.humkmukNM...
`
`
`33..3333333333383
`
`R.233nm.Na.KEN...
`
`El...3&3
`III-Flu!-“E"“$33.33.—'33:I..3.33333I3...3.
`
`El...33333nl.3333
`
`EINI...IE33§IN323
`g'...33.333333“$3333.33
`
`wwim-=30.2333.3.8
`
`EDIT—.1:S333
`
`'Emma
`
`K40 Exhibit 1016, pg. 5
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 5
`IPR2013-00240
`
`
`
`
`
`
`

`

`US. Patent
`
`‘E.
`
`
`
` x33$.on,-“18.252.3.m-k'pwtiCNIQ..RD.1.95.3.3.3.=E:Nm‘wwmfiw.ElmnSm:'anEumhingcml5&8IE:I333.9%,W.SnRDQ'-l~:o|\NU..qu,lAIE;“fl_I9.“ban?
`
`tubW1III...—a—Istu35vJustI
`
`
`
`—-
`
`4,a,,m-mmat
`
`mcm.GE
`
`70O,5
`
`w.5,anGsl
`
`0.GP.‘
`
`K40 Exhibit 1016, pg. 6
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 6
`IPR2013-00240
`
`
`

`

`US. Patent
`
`Apr. 19, 1994
`
`Sheet 5 of 6
`
`5,305,007
`
`STAR?
`
`R540 COMPLEX
`VALUE 0j FROM
`ASIC
`
`
`
`CALCULATE
`
`MAGN/TUDE (M)
`OF COMPLEX
`
`WILLIE AND
`
`
`
`
`INDEX DIS~
`' PLACEMENT
`
`
`
`AVERAGE c,-
`. mo MEMORY
`
`
`STORE MAX
`8:
`INDEX
`
`DISPLACEMENT
`
`INTO PEAK
`
`TABLE
`
`
`SEr MAX-=0
`
`N0
`
`SWEEP
`COMPLHE?
`
`COMMENCE
`PROCESS
`
`,
`
`Y
`
`FIG. 6
`
`0
`
`K40 Exhibit 1016, pg. 7
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 7
`IPR2013-00240
`
`

`

`US. Patent
`
`Apr. 19, 1994
`
`Sheet 6 of 6
`
`5,305,007
`
`INDEX 4!: MAG
`
`INF0.
`
`
`
`
`SINGLE
`STABLE PEAK
`
`
`w 2
`CONSEcunVE
`
`SWEEPS? ,
`
`Y
`
`\
`
`SIGNAL ALERT
`
`‘
`
`_
`
`@
`
`M or?
`KA
`
`TX
`
`
`
`
`
`
`N K—BAND
`SPACING‘?
`
`
`v
`
`PASS
`"NEAR
`r 37'?
`
`Y
`
`W/IN 20: 0F
`SWEEP?
`
`
`
`
`
`
`
`
`
`RANGE
`cmssznmnm/s
`SAUSFIED?
`
`
`
`
`
`
`
`
`
`Y
`
`_-FIG. 7
`
`K40 Exhibit 1016, pg. 8
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 8
`IPR2013-00240
`
`

`

`1
`
`WIDEBAND RADAR numerou
`
`FIELD OF THE INVENTION
`
`5,305,007
`
`The present invention relates to a police radar detec~
`tor, and more particularly, to a wideband radar detector
`which efficiently and economically detects valid police
`radar signals which are present in the X, K, and Ka
`signal bands and which will ignore interfering signals
`generated by other radar detectors.
`BACKGROUND OF THE INVENTION
`
`An electronic assembly for detecting the presence of
`police radar signals is generally known, and will be
`referred to herein as a radar detector. In use, the radar
`detector is mounted in a vehicle and provides an audible
`and/or visual indication of the presence of a police
`radar signal.
`.
`Many known radar detectors cover two signal bands,
`namely the X band (10.525 GHz+/——25 MHz) and the
`K band (24.15 GHz+/—50 MHz). Other known radar
`detectors cover three signal bands, namely the X band,
`the K band, and a narrow Ka band (34.3 GHz+/—~ 100
`MHz).
`More recently, the Ka band has been widened and is
`now specified to be 34.7 GH2+/—500 MHz. However,
`a problem has arisen is trying to cover this widened Ka
`band. A radar detector generally has either a fixed or
`sweeping first local oscillator that emits a signal cen-
`tered around 11.559 GHz. The third harmonic of this
`signal (3X 11.559 GHz=34.677 GHz) falls within the
`wide Ka band. This signal is radiated out from the an—
`tenna of the radar detector and may be received by
`other radar detectors. If this signal is fixed, it appears to
`other radar detectors to be a police Ka radar signal and
`therefore causes these other radar detectors to generate
`an alert.
`As can be appreciated, the only difference between a
`valid police Ka band radar signal and an interfering
`signal caused by another radar detector is that the inter~
`fering signal has energy radiated at the fundamental
`frequency and the second harmonic frequency in addi»
`tion to the third harmonic frequency. Thus, it would be
`desirable to have a radar detector that could simulta—
`neously determine if there was energy present at the
`fundamental and second harmonic frequencies when a
`signal was detected in the 'wide Ka band, and not alert
`under these conditions.
`Furthermore, in recent years, a number of automatic
`door openers have been designed to use microwave
`signals to detect the proximity of people. Although
`these signals usually appear as X band sources to radar
`detectors, 3 group of x band door openers may have the
`signal properties associated with a k band source. Ac-
`cordingly, a new false signal rejection scheme is neces—
`sary.
`A disclosure of the general operation of police radar
`and police radar warning receivers is provided in U.S.
`Pat. No. 5,079,553, which is commonly assigned to
`“Cincinnati Microwave, Inc.” (hereinafter referred to
`as "CMI”) and is hereby incorporated by reference.
`U.S. Pat. No. 5,079,553 discloses a police radar warning
`receiver including a DSP circuit having a correlator
`and peak detector. The output of an FM discriminator is
`digitally sampled so that the magnitude of each digital
`sample word corresponds to the magnitude of the sig»
`nals and noise received at the X and/or K band frequen-
`cies. Each sample word is then manipulated in a digital
`
`10
`
`l5
`
`20
`
`25
`
`30
`
`35
`
`45
`
`50
`
`55
`
`65
`
`2
`correlator and coupled to an averagcr which performs
`accumulating and averaging operations for each sample
`interval or group of intervals. A peak detector com-
`pares averager words with a current dynamic thresh-
`old. To avoid false alarms, the DSP circuit includes an
`index memory operating in conjunction with the peak
`detector to provide sweep-to-sweep comparison. If
`none of the averager words exceed the dynamic thresh-
`old and one or two of the same averager words present
`the largest magnitude for an extended period of time, an
`alarm enable is provided. Also, the peak detector evalu-
`ates the spacing between those segments which have
`magnitudes exceeding the dynamic threshold to deter-
`mine whether the alarm enable should indicate an X or
`K band.
`U.S. Pat. No. 5,068,663 discloses a radar detector
`which utilizes an amplitude detection scheme to detect
`radar signals. As shown in FIG. 1 of that patent, the
`radar detector 100 monitors the X, Ku, K and Ka bands.
`Amplitude signals are down‘converted by a series of
`mixers and compared to a threshold. Detected ampli-
`tude signals must persist for a minimum period of time
`before the microprocessor 128 performs signal verifica-
`tion.
`U.S. Pat. Nos. 4,929,954, 4,772,889, and 4,723,125
`disclose devices for calculating a discrete moving win-
`dow Fourier transform for use in the processing of a
`pulse compression radar signal. As shown in FIG. 1 of
`U.S. Pat. No. 4,772,889, a plurality of stages (E) receive
`samples of the signal x0) for which a Fourier transfor-
`mation is sought. To reduce the number of operations
`performed when the number of stages (E) becomes
`large, the complex rotation performed by the operator 1
`is broken down into a first rotation in the first quadrant
`that is performed in a way common to all of the stages.
`Then an additional rotation for each stage equal to 0, l,
`2, or 3 times pi/Z is performed.
`U.S. Pat. No. 5,099,194 discloses a digital frequency
`measurement receiver having an improved bandwidth.
`As shown in FIG. 1 of the patent, RF signal 10 is mixed
`with a signal from a local oscillator 12 and then pro-
`vided to power dividers 32. The mixed signal is divided
`and supplied to analog to digital converters 42 and 44.
`Each converter Operates at a different sampling fre-
`quency. The signal is then supplied to a processor 50
`where a Fourier transform is performed to determine a
`frequency f.
`The ESCORT and PASSPORT radar detector prod-
`ucts, manufactured by Cincinnati Microwave, Inc., use
`a correlation scheme to detect the presence of a single
`period sinusoid, or s-curve. The signal is converted to a
`digital equivalent with a single bit of precision. Identifi-
`able sets of 0’s and 1’s will result from the sinusoid or its
`180 degree out of phase equivalent. These are come
`niently recognized by a low gate count digital circuit.
`The digitized result is correlated by counting the num-
`ber of occurrences of 0’s followed by 1’s. A detection
`occurs when at least 16 0’5 are followed by at least'lé
`l’s. The opposite case will also generate a detect and is
`represented by 16 1’5 followed by 16 0’5.
`The NEW ESCORT radar detector, also manufac-
`tured by Cincinnati Microwave, Inc., was designed to
`take advantage of techniques available in spectral pro-
`cessing. It focused on measuring spectral content of
`portions of the FM demodulator output data collected
`during the sweep. The detection criterion was chosen to
`
`K40 Exhibit 1016, pg. 9
`|PR2013-00240
`
`
`
`
`
`
`
`
`
`
`K40 Exhibit 1016, pg. 9
`IPR2013-00240
`
`

`

`5,305,007
`
`4
`value from the ASIC and average it into memory. Peak
`detection and signal verification are also handled in the
`DSP chip.
`BRIEF DESCRIPTION OF THE. DRAWINGS
`
`3
`see if the amplitude of the s-curve component exceeded
`a threshold.
`Detecting signals in a wideband creates problems that
`are not overcome by the prior art. A wideband radar
`detector picks up the 1000 MHz wide Ka band as well
`as the X and K Bands handled by more primitive prod-
`ucts. The Ka band is 5 to 10 times wider than X and K
`bands. If the sweep time is held constant, a Ka sweep
`would then produce an s-curve that is 10 times higher in
`frequency than that of the X/K sweep. Equivalent anal-
`ysis processing would require 10 times the throughput.
`Additional complications arise when the competing
`considerations of product cast and product sensitivity
`are taken into account.
`Thus, a low cost but high throughput process is
`needed. Also, for flexibility, the process should be opti-
`mized in a software setting.
`SUMMARY OF THE INVENTION
`
`The present invention is a radar detector that detects
`signals in the broad Ka band while ignoring interfering
`radiation from other radar detectors. The radar detec~
`tor utilizes a high rate processor and a low rate proces-
`sor to evaluate received signals. The high rate processor
`provides high throughput and the low rate processor
`provides programmable features.
`In a preferred embodiment, the high rate processing
`is performed by a custom Application Specific Inte-
`grated Circuit (“ASIC”) that uses a pipelined architec-
`tural approach to enable high speed processing and
`information throughput. According to the present in-
`vention, the throughput rate is equal to the clock rate.
`The low rate processing is performed by a low cost
`programmable digital signal processor (“DSP”) chip.
`In operation,
`the detector swoops a voltage con-
`trolled oscillatcr (“VCO”) through a range of frequen-
`cies to detect signals in the X, K, and Ka signal bands.
`The signals mix with the sweeping VCO to produce a
`new set of sweeping frequencies that are down con-
`verted and passed to an FM demodulator, where a
`single period sinusoid (“s-curve”) is produced for any
`signal that falls within the desired range.
`The ASIC employs a sliding window discrete Fou-
`rier Transform (“SWDFT”) to reduce the number of
`processing operations. A discrete Fourier Transform
`(“DI-T”) is obtained over some narrow range of fre-
`quencies. By using the SWDFT, once the DFT of a
`window in known, the DFT of an adjacent window can
`easily be calculated. Successive DFI‘ calculations pro-
`duce complex values representative of the energy con-
`tent at consecutive points of the sweep. The magnitude
`of these complex values may then be compared to a
`threshold to determine if a valid signal is present. Alter»
`natively, this step may be approximated by evaluating
`every Nth value.
`The complex values are averaged to improve the
`signal to noise ratio. Also, by averaging each new com-
`plex value into the previously calculated average, a
`“sliding average” is computed. Thus, a continuously
`updated result is obtained.
`*
`The DSP chip sends a control value to the ASIC in
`order to tune the ASIC to a particular s-curve fre-
`' quency. The time it takes the ASIC to collect and pro-
`duce a new complex output value is known as the prt»
`cessor data period. This time period determines the
`maximum execution time available to the DSP chip for
`processing each new value. During the time between
`subsequent sample values, the DSP chip must read the
`
`
`
`
`
`
`
`10
`
`15
`
`35
`
`45
`
`S5
`
`65
`
`FIG. 1 is a block diagram of a preferred embodiment
`of a radar detector according to the present invention.
`FIG. 2 is a graphical representation of a typical out-
`put s-curve from the quadrature detector portion of the
`present invention.
`FIG. 3 is a graphical representation of the output
`from the quadrature detector portion of the present
`invention when an interfering radar source is received.
`FIG. 4 is a block diagram of digital signal processing
`portion of the embodiment shown in FIG. 1.
`FIG. 5 shows the interconnections within a program-
`mable prototype of the ASIC portion of the embodi-
`ment shown in FIG. 2.
`FIG. 6 is a flow chart showing steps performed in the
`Sweep and Peak Detection Processing used in the em—
`bodiment of FIG. 1.
`FIG. 7 is a flow chart showing the PROCESS step of
`FIG. 4 in more detail.
`
`DETAILED DESCRIPTION OF A. PREFERRED
`EMBODIMENT
`
`l. Receiving and Detecting a Signal
`A block diagram of the present invention is illustrated
`in FIG. 1. An incoming signal is received at antenna 10
`and mixed by a first mixer 12 with a sweeping signal
`from a first local oscillator 14 to generate a first sweep—
`ing intermediate frequency signal IF t. A sweep circuit
`16 controls the first local oscillator 16 in response to
`master control by a microprocessor 18.
`Two separate sweep cycles are provided by the
`sweep circuit 16. During the first sweep cycle, the X
`and K bands are processed, and during the second
`sweep cycle, the Ka band is processed, as will be de-
`scribed in more detail below.
`When processing X and K hand signals, i.e., during
`the first sweep cycle, the sweeping signal from the first
`local oscillator 14 is 11.559 GHZ+/-—60 MHz. For X
`band signals, the first mix is a fundamental high side mix
`(upper heterodynelthat results in a sweeping intermedi-
`ate frequency signal
`IF; of
`1024 MHz (11.559
`OBI-10.525 (3111). For K hand signals, the first mix is
`a second harmonic low side mix (lower heterodyne)
`that also results in a sweeping intermediate frequency
`signal IF: of 1034 MB: (24.15 GHz—(2X11.559 6112)).
`The first sweeping intermediate frequency signal IFI
`is amplified by amplifier 20 and then mixed by a second
`mixer 12 with a fixed signal from a second local oscilla-
`tor 24 to generate a second sweeping intermediate fre-
`quency 11:2. The fixed signal from the second local
`oscillator 24 is 1034 MHz.
`When processing Ka hand signals, i.e., during the
`second sweep cycle, the amplifier 20 and second local
`oscillator 24 are turned off. The microprocessor 18
`causes the sweep circuit 16 to increase the sweeping
`signal
`from the first
`local oscillator 14 to 11.559
`GHZ+/-—-200 MHz. Thus, the third harmonic of the
`first local oscillator 14 (34.677 GHz+/—600 MHz) is
`mixed with the incoming signal and then amplified by
`bypass amplifier 26 to produce a third sweeping inter-
`mediate frequency signal IF3 at 10 MHz.
`The second sweeping intermediate frequency IF; and
`the third sWeeping intermediate frequency IF; are
`
`K40 Exhibit 1016. pg. 10
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 10
`IPR2013-00240
`
`

`

`5,305,007
`
`5
`
`25
`
`35
`
`5
`summed by summer 28, although the net effect is that
`signal IFz is passed through the summer during the first
`sweep and signal IF3 is passed through the summer
`during the second sweep. Each signal is then passed
`through amplifier 30, bandpass filter 32, and then lim-
`ited and demodulated by a quadrature detector 34 dur-
`ing its respective sweep cycle in a manner that is known
`and explained in U.S. Pat. No. 5,049,885, which is ex-
`pressly incorporated herein by reference.
`The output of the quadrature detector 34 is a pair of 10
`single cycle sine waves that are referred to herein a
`“rt-curves” as shown in FIG. 2. As explained in U.S. Pat.
`No. 5,049,885, the s~curves define positions in time rela-
`tive to the start of the sweep which correspond to the
`frequency at which the incoming signal is received. The 15
`first s-curve is related to the actual incoming signal,
`while the second s—curve is a result of an “image” of the
`imcoming signal which is created by the heterodyning
`receiver. When the sweeping signal is slightly above 10
`MHz, the output of the quadrature detector 34 is a 20
`negative voltage. When the sweeping signal is slightly
`below 10 MHZ, the output of the quadrature detector 34
`is a positive voltage. When the sweeping signal is out-
`side the bandwidth of the quadrature detector 34, no
`output signal is observed.
`The time period t between the s-curves is a function
`of how fast the first local oscillator is sweeping, and of
`the frequency of the amplifier 30, filter 32, and quadra—
`ture detector 34. For example, if the first local oscillator
`14 is sweeping a total of 120 MHz in 120 msec, then for 30
`X band signal
`the radar detector
`is sweeping 1
`MHz/m8. Since the K band coverage is a second har-
`monic mix, the radar detector sweeps 240 MHz in 120
`mS, he. 2 MHz/m8. The time spacing between s~curves
`can be determined by the following equation:
`
`(sweep time X (2 xIF frequency)+sweep width).
`
`Thus, assuming a 120 MHz sweep time, the spacing for
`a valid signal of each type would be as follows:
`l;=(120 ms " (2 ‘ 10 MH2)+12O MHz)=20 m5;
`
`tk=(120 ms ‘ (2 * 10 MHZ)+24O MHZ)=10 ms;
`
`txg=(120 ms ‘ (2 ’ 10 MH1)+ 1200 MHz)=2 ms.
`
`40
`
`45
`
`50
`
`Therefore, absent the teaching of the present invention,
`a radar detector sensing a 2 ms time spacing between
`s-curves would generate an alert for a K21 band radar
`signal.
`However, if the incoming signal is from an interfering
`source, i.e., another radar detector, then three pairs of
`s~curvcs will be generated as shOwn in FIG. 3, including
`one pair spaced at 2 ms, one pair spaced at 3 ms, and one
`pair spaced at 6 ms. The pair of s-curves spaced at 2 ms 55
`are due to the third harmonic of the signal from the first
`oscillator 14 mixing with the third harmonic of the
`signal received from the interfering radar detector. The
`pair of s-curves spaced at 3 ms are due to the second
`harmonic of the signal from the first oscillator 14 mixing 60
`with the second harmonic of the signal received from
`the interfering radar detector. The pair of s-curves
`spaced at 6 ms are due to the fundamental signal from
`the first oscillator 14 mixing with the fundamental sig-
`nal received from the interfering radar detector.
`For example, if the interfering radar detector has a
`fixed first local oscillator generating a signal at 11.559
`GI-Iz, then a second harmonic signal radiates at 23.118
`
`65
`
`6
`GHz and a third harmonic signal radiates at 34.667
`GHz. When the first oscillator 14 of the present inven-
`tion sweeps through 11.549 GHz, an s«curve is gener-
`ated due to the 10 MHz difference between the receiver
`and the interfering source:
`
`(11.549 GHz— 11.559 Gllz= 10 MHz).
`
`When the first oscillator 14 of the present invention
`sweeps through ll.554 GHz, an sucurve is also gener-
`ated due to the 10 MHz difference between the second
`harmonic mix of the receiver and the interfering source:
`
`((11.54 GB: ‘ 2)—23.1 18 GHz=1O MHZ).
`
`The same thing occurs at the third harmonic when the
`first oscillator 14 sweeps through 11.5557 GHz;
`
`((115557 SH: ' 3)34.677 GHz=10 MHZ).
`
`Since conversion loss of a third harmonic mix will be
`greater than that of a fundamental or second harmonic
`mix, the responses of each type of mix will differ in
`amplitude. These responses can be easily seen by using
`a tuner in the antenna to reduce out fundamental radia-
`tion.
`
`The use of signal processing techniques can thus be
`used to determine the difference between various re-
`sponses in order to make a decision to alert for a valid
`signal, as will now be described in detail.
`
`2. Digital Signal Processing of a Detected Signal.
`Once an s-curve pair has been successfully detected
`by quadrature detector 34, it is presented to the micro-
`processor/D8? 18 for digital signal processing. As
`shown in FIG. 4, this portion of the invention is primar-
`ily centered on the idea of spitting the signal processing
`into two groups: high rate processing and low rate
`processing. In order to meet the throughput require—
`ments, the high rate data reduction hardware is consoli-
`dated into an Application Specific Integrated Circuit
`(“ASIC”) 50 using a classic pipelined architectural ap-
`proach found in dedicated signal processing applica—
`tions. In order to meet programmable needs, the ASIC
`50 passes low rate data to a low cost programmable
`digital signal processing (“DSP”) chip 54. The DSP
`chip 54 is programmed to collect low rate data, perform
`post processing, and report the results to the product
`user, Preferably, the DSP chip 54 is Texas Instruments’
`TMSSZOCIS chip (hereinafter referred to as the “c15”).
`
`A. High Rate Processing
`i. Pipelined Architecture using SWDFT
`As stated above, the high rate processing section of
`the preferred embodiment includes an ASIC 50 which
`uses a pipelined architectural approach. A pipelined
`architecture refers to the arrangement in which calcula-
`tions are performed in the process. The process is bro-
`ken up into a series of calculations that are all performed
`in parallel in the pipeline. On each clock edge, the re-
`sults of a given step are passed on to the next stage of
`the pipeline. This enables the hardware to achieve a
`throughput rate that is equal to the clock rate of the
`pipeline.
`Inc.’s NEW ESCORT
`In Cincinnati Microwave,
`radar detector, a Motorola 56000 Digital Signal Proces-
`sor chip was used for radar signal detection and pro-
`cessing. However, this product performed all process-
`
`K40 Exhibit 1016, pg. 11
`|PR2013-00240
`
`
`
`
`
`
`
`
`
`
`
`
`K40 Exhibit 1016, pg. 11
`IPR2013-00240
`
`

`

`7
`ing setps sequentially. Thus, each time a new input was
`fed in, the Motorola 56000 devoted over 90 percent of
`its processing time to performing these menial calcula—
`tions. As can be apprecitaed, the throughputof such a
`device is inadequate for timely wide Ka band process-
`mg.
`The NEW ESCORT product did, however, take
`advantage of spectral processing techniques by measur-
`ing the spectral content of portions of output data from
`the FM demodulator collected during a sweep. A de—
`tection criterion was chosen to see if the amplitude of
`the s~curve component exceeded a threshold.
`A numerically Wasteful solution involves converting
`the FM demodulator data into a frequency domain
`equivalent using a Fourier Transform. The component
`of interest will appear as a single value in the Fourier
`Series. This value can then be taken from the Fourier
`Series and compared to a threshold. However, the s-
`curve is not present during the entire sweep. Also, the
`period of the s—curve grows with signal strength. Fur-
`ther, a Fourier Transform requires a substantial amount
`of processing. Therefore, it is not a suitable approach to
`the problem of widcband signal processing.
`A better approach is based on measuring the spectral
`content of successive segments (or windows) of data
`from the sweep.‘The window size is chosen to be close
`to the period of the s-curve. The detection criterion is
`chosen to see if the amplitude of the fundamental com
`ponent
`in successive windows exceeds a threshold.
`Each time a new sample is acquired from an A/D con-
`verter, the window can be updated. Updating is accom~
`plished by incorporating the current sample point while
`eliminating the oldest sample point. This results in an
`effective shift of the window position so that it includes
`the N most recent points acquired during the sweep.
`The set of values in the window is converted into the
`frequency domain by a method such as the Fourier
`Transform. The component of interest can then be ex-
`amined in each from successive windows.
`A digital approximation to the continuous Fourier
`Transform is the Discrete Fourier Transform (“DFT”).
`The Fast Fourier Transform (“FFT”) is a mathemati-
`cally efficient method for finding the spectrum of a time
`series, however, it is less efficient than the DFI‘ for
`calculating a single value within the spectrum. Thus,
`the DFT is suitable for the present application even
`though the period of the scurve increases with signal
`strength. As a result, the algorithm can be simplified to
`the task of solving for the DFT of successive windows
`within the sweep. Each DFT calculation thus produces
`an output value corresponding to a single component
`found in its input window.
`The DPT will hereinafter be represented by the ex-
`pression H(n/NT) and may be determined according to
`the following equation:
`
`N—l
`”(n/NT) . r kzo “km—flunk/N
`
`where
`T= l/sample rate
`N= # points/period
`n=the component of interest (e.g. l in the case of the
`fundamental frequency)
`See E. Oran Brigham, “The Fast Fourier Trans~
`form,"p.102, expression 6—24.
`Since the present application involves calculating the
`DFT of a sliding window of time samples through the
`
`5,305,007
`
`
`
`
`
`
`
`
`
`
`
`sweep, a little exploration of the DFT of a sliding time
`window reveals an interesting approach that drastically
`cuts the number of multiply and adds associated in the
`previous equation.
`If the DPT of a window is known at time 1.. then the
`DFT for the adjacent window L+l can be calculated
`from the DFT at window L with far fewer operations.
`This concept is expressed in the following expression
`and is known as the Sliding Window Discrete Fourier
`Transform ("SWDFT").
`
`10
`
`HL+I(1/~n,,nwL/N c
`”Ltl/NT)+h(L7)-h((L—MD
`
`15
`
`where
`
`20
`
`25
`
`35
`
`45
`
`55
`
`65
`
`HL+;(l/N'I‘)=the DFT of the L+l window
`h(LT)=the next sample
`h((L—N)T)=the oldest sample in the window
`Using this equation, only one real by complex multi-
`ply and two real adds are involved per input sample.
`The set of operations is performed once per sample at
`the chosen sample rate.
`Successive DFT calculations produce a set of com-
`plex values that are related to the fundamental energy
`content at consecutive points throughout the sweep.
`ii. Structure of ASIC
`a. Generally
`The following is a description of the structure of the
`ASIC that performs the SWDFI‘. Initially, it is impor-
`tant to note that to reduce product cost and parts count,
`all necessary digital circuit functions were absorbed
`into the ASIC. These additional
`functions include
`sound synthesis, LED control, and user interface.
`The ASlC design began with an approach based on
`the method used in the NEW ESCORT product de-
`scribed earlier, which consisted of a circuit for perform-
`ing the SWDFT, a low pass filter, and a decimator. The
`NEW ESCORT received its input data from a 6 bit
`A/D converter and used an 81: static RAM chip for
`storing 24 bit averaged values. Thus, the DSP had a 24
`bit data bus with 48 bit accumulators.
`According to a preferred embodiment of the present
`invention, a 2 bit window comparator circuit is de-
`signed to output binary code (0,0) if the signal value is
`within a window of+ / wk, binary code (0,1) if the
`signal value is greater than +k, and binary code (1,0) if
`the signal value is less than -k. (Since only 3 of the 4
`states available in 2 bits are used, the A/D is actually
`not a 2 bit A/D, but rather log2(3) or 1.58 bits). The
`value for k is set by analog means, and is rather critical.
`If it is too large, subtle features in the sweep signal are
`not passed to the digital process. If too small. signal
`noise is exaggerated. The value for k is a function of the
`signal amplifier gain. The gain is adjusted by maximiz-
`ing the separation between signal and noise when a
`weak signal is present.
`The benefit of reducing the number of bits of input
`data significantly reduces wordlengths in the remaining
`steps in the process. The 2 bit input value is passed into
`a circuit that calculates the SWDF'T. This involves
`accumulating the complex product of the difference
`between the next input and the input that was received
`N points previously.
`The N point delay is accomplished with a shift regis-
`ter. Each input is fed into a shift register that is 128
`points (by 2 bits) deep.
`b. Hardware Description Language
`
`K40 Exhibit 1016, pg. 12 ~
`|PR2013-00240
`
`K40 Exhibit 1016, pg. 12
`IPR2013-00240
`
`

`

`5
`
`10
`
`15
`
`20
`
`25
`
`35
`
`Here, the vector lshift is used to carry the least signif—
`icant bit of the input from the window comparator and
`hshift is used to carry the most significant bit.
`The value N is the number of points in the window
`size for the SWDFT. A‘larger value for N produces a
`more accurate DFT measurement and also increases the
`dynamic range of the output of the SWDFT stage. The
`larger the dynamic range, the greater the number of bits
`required in downstream processing. Practical values
`were shown to be 32, 64, and l28. Design generality has
`been achieved by providing a method for the host pro-
`cessor (t