`571-272-7822
`
`
`
`
`
`
`Paper No. 8
`Entered: October 18, 2018
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`APPLE INC.,
`Petitioner,
`
`v.
`
`UNILOC 2017 LLC,
`Patent Owner.
`
`____________
`
`Case IPR2018-01027
`Patent 8,712,723 B1
`____________
`
`
`
`
`Before SALLY C. MEDLEY, MIRIAM L. QUINN, and
`SEAN P. O’HANLON, Administrative Patent Judges.
`
`O’HANLON, Administrative Patent Judge.
`
`
`
`
`DECISION
`Denying Inter Partes Review
`35 U.S.C. § 314(a)
`
`
`
`
`
`
`
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`
`I. INTRODUCTION
`Apple Inc. (“Petitioner”) filed a Petition for inter partes review of
`
`claims 4 and 19 of U.S. Patent No. 8,712,723 (Ex. 1001, “the ’723 patent”).
`Paper 2 (“Pet.”), 1. Uniloc Luxembourg S.A., a predecessor in interest of
`Uniloc 2017 LLC (“Patent Owner”), filed a Preliminary Response. Paper 7
`(“Prelim. Resp.”).
`
`Institution of an inter partes review is authorized by statute only when
`“the information presented in the petition . . . and any response . . . shows
`that there is a reasonable likelihood that the petitioner would prevail with
`respect to at least 1 of the claims challenged in the petition.” 35 U.S.C.
`§ 314(a). For the reasons set forth below, upon considering the Petition,
`Preliminary Response, and evidence of record, we conclude that the
`information presented in the Petition fails to establish a reasonable
`likelihood that Petitioner will prevail in showing the unpatentability of
`claims 4 and 19. Accordingly, we decline to institute an inter partes review.
`
`A. Related Matters
`The parties indicated that the ’723 patent is the subject of the
`
`following litigation:
`Uniloc USA, Inc. v. Apple Inc., No. 2-17-cv-00522 (E.D. Tex.,
`filed June 30, 2017).
`Uniloc USA, Inc. v. Samsung Elects. Am., Inc., No. 2-17-cv-
`00650 (E.D. Tex., filed Sept. 15, 2017),
`Uniloc USA, Inc. v. LG Elecs. USA, Inc., No. 4-12-cv-00832
`(N.D. Tex., filed Oct. 13, 2017),
`Uniloc USA, Inc. v. HTC America, Inc., No. 2-17-cv-01629
`(W.D. Wash., filed Nov. 1, 2017),
`
`2
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`
`Uniloc USA, Inc. v. Huawei Devices USA, Inc., No. 2-17-cv-
`00737 (E.D. Tex., filed Nov. 9, 2017),
`Uniloc USA, Inc. v. Apple, Inc., No. 4-18-cv-00364 (N.D. Cal.,
`filed Jan. 17, 2018).
`Apple Inc. v. Uniloc USA, Inc., IPR2018-00389 (PTAB, filed
`Dec. 22, 2017) (“the ’389 IPR”).
`Pet. 2; Prelim. Resp. 2; Paper 6, 3.
`
`B. The Challenged Patent
`The ’723 patent relates to monitoring and counting periodic human
`
`motions, such as steps. Ex. 1001, 1:12–14. The ’723 patent states that
`inertial sensors (e.g., accelerometers) are used in step counting devices
`allowing an individual to track the number of daily steps. Id. at 1:18−29.
`One problem recognized in the ’723 patent is the limitations of these step
`counting devices concerning the orientation of the device during use. Id. at
`1:29−34. Further, motion noise often confuses these devices resulting in
`missed steps or counting false steps, with a particular problem identified of
`inaccurate step measurements for slow walkers. Id. at 1:35−43.
`
`The ’723 patent provides for accurate counting of steps without regard
`for the orientation of the step counting device, even if that orientation
`changes during operation. Id. at 2:33−38. In particular, the ’723 patent
`describes assigning a dominant axis after determining an orientation of the
`inertial sensor, where the orientation of the inertial sensor is continuously
`determined. Id. at 2:15−19. In one embodiment, the ’723 patent method
`determines rolling averages of the accelerations of each axis monitored by
`the inertial sensor in the device. Id. at 6:15−21. The largest absolute rolling
`average indicates the axis most influenced by gravity, which may change
`
`3
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`over time, as the device’s orientation changes because of rotation. Id. at
`6:20−25.
`
`With regard to the embodiment shown in Figure 8, reproduced below,
`the ’723 patent describes the method for measuring the acceleration along
`the assigned dominant axis to detect, and count, steps. See id. at 12:30−35.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Figure 8 illustrates a diagram for a method of recognizing a step.
`
`After measurements of acceleration data (step 805) and filtering those
`
`4
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`measurements (step 810), the method evaluates the orientation of the device
`and assigns a dominant axis (step 812). A processing logic determines
`whether a measurement is within a cadence window (step 815). The
`cadence window is the allowable time window for steps to occur. Id. at
`3:65−66. In one embodiment, the cadence window is determined based on
`the actual stepping period or actual motion cycle, but default limits or other
`determiners may be used to set the cadence window. Id. at 4:7−27. After
`each step is counted, the minimum and/or maximum of the cadence window,
`or window length, may be adjusted based on actual cadence changes. Id.
`Therefore, the cadence window is dynamic so that it continuously updates.
`Id. at 4:31−33.
`
`If the measurement of acceleration along the dominant axis is within
`the cadence window, and is within the range of acceleration thresholds
`(steps 820, 830), the motion is determined to be a step and is counted (step
`835). Otherwise, the step is not counted (step 840) and the method
`continues to evaluate subsequent measurements.
`
`C. The Challenged Claims
`Petitioner challenges claims 4 and 19 of the ’723 patent. Claim 4
`
`depends from independent claim 1 through intermediate dependent claim 3,
`and claim 19 depends directly from independent claim 14. Claims 1, 3, 4,
`14, and 19 are reproduced below:
`1.
`A method of monitoring human activity using an inertial
`sensor, comprising:
`
`assigning a dominant axis with respect to gravity based
`on an orientation of the inertial sensor;
`
`detecting a change in the orientation of the inertial sensor
`and updating the dominant axis based on the change; and
`
`5
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`
`counting periodic human motions by monitoring
`
`accelerations relative to the dominant axis by counting the
`periodic human motions when accelerations showing a motion
`cycle that meets motion criteria is detected within a cadence
`window; and
`
`updating the cadence window as actual cadence changes.
`
`The method of claim 1, wherein at least one of the
`3.
`motion criteria is a dynamic motion criterion, the dynamic
`motion criterion updated to reflect current conditions.
`
`The method of claim 3, wherein the dynamic motion
`4.
`criteria includes at least a lower threshold, wherein the lower
`threshold is adjusted based on at least one of a rolling average
`of accelerations and the orientation of the inertial sensor.
`
`14. A non-transitory machine readable medium containing
`executable computer program instructions which, when
`executed by a processing system, cause said system to perform
`a method for:
`
`assigning a dominant axis with respect to gravity based
`on an orientation of the inertial sensor;
`
`detecting a change in the orientation of the inertial sensor
`and update the dominant axis based on the change; and
`
`counting periodic human motions by monitoring
`accelerations relative to the dominant axis by counting the
`periodic human motions when accelerations showing a motion
`cycle that meets motion criteria is detected within a cadence
`window; and
`
`updating the cadence window as actual cadence changes.
`
`19. The non-transitory machine readable medium containing
`executable computer program instructions of claim 14, wherein
`the dynamic motion criteria includes at least a lower threshold,
`wherein the lower threshold is adjusted based on at least one of
`a rolling average of accelerations and the orientation of the
`inertial sensor.
`Ex. 1001, 15:13–24, 15:28–34, 16:23–35, 16:65–17:3.
`
`6
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`
`
`
`D. The Prior Art
`Petitioner relies on the following prior art references:1
`Reference
`Date
`U.S. Patent No. 7,463,997 B2 to
`Filed Oct. 2, 2006, issued
`Pasolini et al. (“Pasolini”)
`Dec. 9, 2008
`U.S. Patent No. 7,698,097 B2 to
`Filed Oct. 2, 2006, issued
`Pasolini et al. (“Fabio”)
`Apr. 13, 2010
`U.S. Patent No. 5,976,083 to
`Issued Nov. 2, 1999
`Richardson et al. (“Richardson”)
`Robert L. Harris, Information Graphics: A Comprehensive
`Illustrated Reference (1996) (“Harris”)
`
`Exhibit
`Ex. 1005
`
`Ex. 1006
`
`Ex. 1007
`
`Ex. 1011
`
`E. Asserted Grounds of Unpatentability
`Petitioner “seeks review with respect to only . . . claims 4 and 19” as
`
`being obvious under 35 U.S.C. § 103(a) over Pasolini, Fabio, and
`Richardson. Pet. 3, 12. Petitioner submits a declaration of Joseph A.
`Paradiso, PhD (Ex. 1003, “Paradiso Declaration” or “Paradiso Decl.”) in
`support of its contentions.
`
`II. ANALYSIS
`
`A. Level of Ordinary Skill in the Art
`Petitioner avers that the level of ordinary skill set forth in the instant
`
`petition is the same as in the ’389 petition. Pet. 4 n.1. Citing its declarant,
`Joseph A. Paradiso, PhD, Petitioner contends that a person having ordinary
`skill in the art (“POSITA”) at the time of the invention would have had “a
`
`
`1 Petitioner is reminded that exhibits should be submitted in letter format
`(8 ½ inch × 11 inch). See 37 C.F.R. § 42.6(a)(1).
`
`7
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`Bachelor’s degree in Electrical Engineering, Computer Engineering, and/or
`Computer Science, or equivalent training,” and “approximately two years of
`experience working in hardware and/or software design and development
`related to MEMS (micro-electro-mechanical) devices and body motion
`sensing systems.” Id. at 9 (citing Ex. 1003, 9). “Patent Owner does not
`offer a competing definition for [a] POSITA . . . .” Prelim. Resp. 8.
`
`We find Petitioner’s definition of the level of ordinary skill
`reasonable, and for purposes of this Decision, adopt it as our own.
`
`B. Claim Construction
`In an inter partes review, a claim in an unexpired patent shall be given
`
`its broadest reasonable construction in light of the specification of the patent
`in which it appears. 37 C.F.R. § 42.100(b); Cuozzo Speed Techs., LLC v.
`Lee, 136 S. Ct. 2131, 2144–46 (2016) (upholding the use of the broadest
`reasonable interpretation standard). Consistent with the broadest reasonable
`construction, claim terms are presumed to have their ordinary and customary
`meaning as understood by a person of ordinary skill in the art in the context
`of the entire patent disclosure. In re Translogic Tech., Inc., 504 F.3d 1249,
`1257 (Fed. Cir. 2007). The presumption may be overcome by providing a
`definition of the term in the specification with reasonable clarity,
`deliberateness, and precision. See In re Paulsen, 30 F.3d 1475, 1480 (Fed.
`Cir. 1994). In the absence of such a definition, limitations are not to be read
`from the specification into the claims. See In re Van Geuns, 988 F.2d 1181,
`1184 (Fed. Cir. 1993). Only those terms which are in controversy need be
`construed, and only to the extent necessary to resolve the controversy. Vivid
`Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d 795, 803 (Fed. Cir. 1999).
`
`8
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`Petitioner proposes constructions for “dominant axis” and “cadence
`
`window,” asserting that its “claim constructions . . . are based on the
`broadest reasonable construction.” Pet. 10–11. Petitioner avers that the
`claim construction set forth in the instant petition is the same as in the’389
`petition. Id. at 4 n.1. Patent Owner asserts that no claim construction is
`needed and disagrees with Petitioner’s proposed constructions. Prelim.
`Resp. 8–11. Patent Owner’s assertions are substantially similar to those it
`made in the ’389 IPR. See IPR2018-00389, Paper 6.
`
`For purposes of this Decision, we adopt the claim construction of
`“dominant axis” as set forth in the ’389 IPR to mean “the axis most
`influenced by gravity,” and determine that we do not need to construe
`“cadence window.”
`
`C. The Challenge
`Petitioner argues that claims 4 and 19 would have been obvious over
`
`Pasolini, Fabio, and Richardson. Pet. 39–54. In support of its showing,
`Petitioner relies upon the Paradiso declaration. Id. (citing Ex. 1003).
`Because challenged claim 4 depends from independent claim 1 through
`intermediate dependent claim 3 and challenged claim 19 depends directly
`from independent claim 14, Petitioner includes an analysis of claims 1, 3,
`and 14. Id. at 23–39. Petitioner avers that the analysis of claims 1, 3, and 14
`set forth in the instant petition is identical to the analysis in the’389 petition.
`Id. at 4.
`
`We have reviewed Petitioner’s assertions and supporting evidence.
`For the reasons discussed below, and based on the record before us,
`Petitioner has not demonstrated a reasonable likelihood of prevailing in
`
`9
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`showing that claims 4 and 19 would have been obvious over Pasolini, Fabio,
`and Richardson.
`
`1. Overview of the Prior Art
`a. Pasolini
`Pasolini discloses a pedometer and step detection method. Ex. 1005,
`
`1:10–12. The pedometer includes an accelerometer and a processing unit.
`Id. at 2:60–63. The pedometer is carried by the user, and the accelerometer
`senses vertical accelerations that occur with each step due to impact of the
`feet on the ground. Id. at 2:66–3:29. The accelerometer produces an
`acceleration signal corresponding to the detected accelerations, and the
`processing unit acquires, at pre-set intervals, samples of the signal. Id. at
`3:16–19, 3:30–33. The processing unit processes the acceleration signal to
`count the number of steps taken by the user. Id. at 3:32–34. Figure 2 is
`reproduced below:
`
`
`
`
`
`
`
`
`
`
`
`“Figure 2 shows a graph corresponding to the pattern of an acceleration
`signal during a step.” Id. at 2:39–40.
`
`Step counting may be accomplished by an algorithm that analyzes the
`acceleration signal to look for a positive phase of the step followed by a
`
`10
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`negative phase within a pre-set time interval from the occurrence of the
`positive phase. Id. at 3:63–66. During the positive phase, a positive-
`acceleration peak occurs (i.e., directed upwards) due to contact and
`consequent impact of the foot with the ground; during the negative phase, a
`negative-acceleration peak occurs (i.e., directed downwards) due to rebound,
`having an absolute value smaller than that of the positive-acceleration peak.
`Id. at 3:23–29. The processing unit compares the value of the acceleration
`signal with positive and negative reference thresholds to identify,
`respectively, the positive phase (positive acceleration peak) and the negative
`phase (negative acceleration peak) of the step. Id. at 3:36–41. Acceleration
`datum values are calculated for each acceleration sample based on the
`acceleration sample value and a mean value of the acceleration samples. Id.
`at 6:5, Fig. 3. The positive phase is detected when the acceleration datum
`exceeds the positive reference threshold and then drops below the positive
`reference threshold. Id. at 4:35–40. The negative phase is detected when
`the acceleration datum drops below the negative reference threshold within a
`certain time interval, in which case the processing unit increments the count
`of detected steps and the algorithm looks for a new potential positive phase
`of a step. Id. at 4:66–5:3, 5:40–41. If no negative phase is detected within
`the time interval, the algorithm looks for a new potential positive phase of a
`step without incrementing the step count. Id. at 4:62–65, Fig. 3.
`Alternatively, the step detection can be based solely on the positive phase of
`the step. Id. at 7:65–8:1.
`
`The accelerometer can be a linear accelerometer (id. at 2:61), and can
`also be a three-axis accelerometer (id. at 8:11–15). In the latter case, the
`processing unit identifies the axis having the highest mean acceleration
`
`11
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`value (due to gravity) as the main vertical axis to be used for step detection.
`Id. at 8:15–24.
`
`b. Fabio
`Fabio discloses a pedometer that includes an inertial sensor and a
`
`control unit. Ex. 1006, 2:34–40, Figs. 1–2. The pedometer is carried by the
`user, and the inertial sensor measures accelerations along its detection axis.
`Id. at 2:49–59. Steps are counted by analyzing the acceleration data for a
`positive peak exceeding a first threshold, followed by a negative peak
`exceeding a second threshold within a certain time window after the positive
`peak. Id. at 4:12–21. Figure 5 is reproduced below:
`
`
`
`
`
`
`
`Figure 5 shows a graph of the acceleration signal measured during a step of
`the user. Id. at 4:13–15, 6:23–26.
`
`In operation, the control unit initially implements a first counting
`procedure in which acceleration data is sampled at a pre-determined
`frequency. Id. at 3:13–21, Figs. 3–4. The user is considered to be at rest,
`and the control unit executes the first counting procedure to analyze
`acceleration data for an indication that the user is engaged in activity with a
`regular gait. Id. at 3:22–27. If a regular gait is detected, the number of
`detected steps during the first counting procedure is added to the number of
`total steps and the control unit executes a second counting procedure
`according to which each detected valid step is added to the number of total
`
`12
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`steps; if no regular gait is detected within a certain amount of time, the
`pedometer is set in a low power consumption state and the control unit
`executes a surveying procedure until the pedometer is moved. Id. at 3:27–
`56, Figs. 4, 7, 8. If, when executing the second counting procedure, an
`interruption in locomotion is detected, the control unit reverts to the first
`counting procedure. Id. at 3:44–49. If pedometer movement is detected
`during execution of the surveying procedure, the control unit reverts to the
`first counting procedure. Id. at 3:53–57, Fig. 3.
`
`c. Richardson
`Richardson discloses a personal fitness monitoring device that
`
`monitors the aerobic fitness of the user as the user exercises, and provides
`the user with information about the current exercise session. Ex. 1007, 1:5–
`13. The device includes a pedometer, a fitness assessment arrangement, a
`fitness prediction arrangement, a user interface, and an audio output switch.
`Id. at 4:15–19. The pedometer uses bodily movement and personal data
`input by the user to produce a locomotion parameter signal that represents
`the user’s movement. Id. at 4:20–25. “The locomotion parameters include
`the gait, duration, speed, and distance of each step, and optionally, grade and
`terrain characteristics.” Id. at 4:25–27. The fitness assessment arrangement
`receives locomotion, heart rate, and personal data to compute an estimate of
`the user’s fitness. Id. at 4:28–38.
`
`The pedometer includes an accelerometer subsystem that measures the
`on-going instantaneous profile of the user’s movement as magnitudes of
`acceleration in or near the vertical plane. Id. at 6:20–29. A step parameter
`assignment module of the pedometer uses the measured acceleration data to
`detect the user’s steps, and ascribes a gait and speed to each step. Id. at
`
`13
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`6:36–62. The step parameter assignment module uses a gait model, which is
`a statistical compendium of numerous users, and the user’s personal data to
`determine the user’s gait and speed. Id. at 6:62–7:23.
`
`To determine the occurrence of a step, the device samples the
`acceleration data measured by the accelerometer subsystem and stores the
`data in one of two buffers. Id. at 27:60–28:34. The data in one buffer is
`analyzed while data is input into the other buffer. Id. at 28:34–36. The
`system computes a moving average at each sample time and uses the moving
`average as a baseline for step detection. Id. at 28:36–39. Figure 13a shows
`a typical acceleration waveform, and is reproduced below.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Figure 13a is diagrammatic illustration detailing how the personal fitness
`monitoring device detects steps as footfalls. Id. at 3:50–52. As indicated in
`the legend of Figure 13a, measured acceleration data is shown in solid lines
`and the calculated baseline is shown in dotted lines. The system detects
`peaks in the acceleration curve, and determines a peak to be indicative of a
`
`14
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`footfall (i.e., a step) if it occurs later than a minimum time after a previous
`footfall, if it is not on the falling side of the waveform, and if the peak is
`greater than a minimum height above the baseline. Id. at 28:48–60. As
`illustrated, running steps produce higher peaks and more frequent footfalls
`than walking steps. Id. at 28:61–62.
`
`2. Claims 1, 3, and 14
`Petitioner notes that challenged claim 4 depends from independent
`
`claim 1 through intermediate dependent claim 3 and challenged claim 19
`depends from independent claim 14, and avers that “the analysis of claims 1,
`3 and 14 in this petition is verbatim identical to the analysis of claims 1, 3
`and 14 presented in the ’389 petition.” Pet. 3–4. Patent Owner presents
`arguments for claims 1 and 14 that are substantially similar to those it made
`in the ’389 IPR, and argues that the challenge to claims 4 and 19 should be
`denied for the same reasons. Prelim Resp. 15–23; see also IPR2018-00389,
`Paper 6.
`
`We do not repeat here our findings and determinations made in the
`’389 IPR, in which we determined that Petitioner had established a
`reasonable likelihood of prevailing on its challenges to claims 1, 3, and 14
`(which are not challenged in this proceeding).
`
`3. Claims 4 and 19
`The recitations of claims 4 and 19 are substantially similar, the
`
`difference being that claim 4 recites a method and claim 19 recites a non-
`transitory machine readable medium. See Ex. 1001, 15:31–34, 16:65–17:3.
`In addressing claim 19, Petitioner merely cites back to its analysis of claim
`4. Pet. 54. Patent Owner similarly addresses claims 4 and 19 collectively.
`
`15
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`Prelim. Resp. 11–15. Our analysis likewise focuses on language that is
`common to both claims 4 and 19.
`
`a. Rationale to Combine Pasolini, Fabio, and Richardson
`Petitioner notes that, at each sample time, Richardson computes a
`
`moving average of acceleration, which serves as a baseline for describing
`the acceleration waveform of a step. Pet. 44 (citing Ex. 1007, 28:33–36). 2
`Petitioner argues that “[t]he acceleration data used to determine the baseline
`is limited to the data for the current sample time (i.e. stepping period) that is
`stored in one of two buffers—either A or B.” Id. (citing Ex. 1007, 28:32–
`36; Ex. 1003, 59). Petitioner argues that it would have been obvious to use
`Richardson’s teaching of adjusting the threshold values because “using a
`moving average, where a sample size can be varied, would serve to likewise
`smooth-out the variations in the data that may occur due to a change in the
`user’s pace, gait, or walking surface.” Id. at 45 (citing Ex. 1003, 60).
`Petitioner relies on Harris to teach “that a moving average is often used in
`data analysis ‘to smooth the curve of a data series and make general trends
`more visible.’” Id. (quoting Ex. 1011, 243). Petitioner then argues that
`“applying a moving average of accelerations with a smaller sample size (i.e.,
`one stepping period), as disclosed in Richardson, would be beneficial to
`Pasolini in that it would yield a smoother acceleration threshold should the
`user’s walking surface change within a single sample period.” Id. (citing Ex.
`1003, 60) (emphasis added).
`
`
`2 It appears that Petitioner’s citation to lines 33–36 is a typographical error,
`with the intended reference being to lines 36–39, which is where the quoted
`language appears.
`
`16
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`Thus, Petitioner proposes to rely on the teachings of Richardson
`
`because it is known to use a moving average of past values to smooth-out
`variations in the data. Id. Petitioner relies on Harris to teach the concept of
`smoothing data via a moving average (also known as a rolling average). Id.
`Harris explains that “[e]ach point on a moving average curve is generally
`calculated by averaging the value for the current period plus a fixed number
`of prior periods” and “the greater the number of intervals, the smoother the
`moving average curve.” Ex. 1011, 243 (emphasis added). Petitioner
`contradicts this stated rationale for using Richardson’s teachings by arguing
`that one should only use “data that is generated in the current sample period,
`rather than an average based on the data from both the current and previous
`stepping periods.” Pet. 44–45. Petitioner’s contention that using data from
`only the current stepping period would “yield a smoother acceleration
`threshold” does not comport with Petitioner’s stated rationale or the
`teachings of Harris upon which Petitioner and its declarant rely. Id. at 45.
`Petitioner’s declarant presents the same conclusory statements, and, thus,
`fails to explain adequately how Petitioner’s contention of using a single data
`period follows from its stated reliance on the benefits of using a moving
`average of past values. See Ex. 1003 ¶ 70.
`
`Moreover, we find insufficient the factual support for Petitioner’s
`contentions that “Richardson’s moving average is based on the data that is
`generated in the current sample period, rather than an average based on the
`data from both the current and previous stepping periods” and that
`Richardson discloses “applying a moving average of accelerations with a
`smaller sample size (i.e., one stepping period).” Pet. 44–45.
`
`17
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`Richardson’s sole description of its use of a moving average is “[t]he
`
`first step is to compute at each sample time a moving average of acceleration
`168, which serves as a baseline for describing the acceleration 168
`waveform of a locomotor step.” Ex. 1007, 28:36–39. This baseline is
`illustrated in Figure 13a, which is reproduced in section II.C.1.c above.
`Contrary to Petitioner’s contention, there is no explicit disclosure in
`Richardson that the moving average, or baseline, is generated based on data
`in only the current sample period. As seen in Figure 13a, the baseline
`smoothing is substantially constant throughout the entire baseline curve,
`including at the initiation of the sample period at time 0.0. If, as asserted by
`Petitioner, the moving average begins anew at each sample period, one
`would expect the baseline curve to lag the acceleration curve at initiation,
`given that there would be no prior values to include in the averaging
`calculation. See Ex. 1011, 243 (“the fewer the time intervals used in the
`averaging process, the more closely the moving average curve resembles the
`curve of the actual data”). As illustrated in Figure 13a, however, this is not
`the case. Moreover, we note that the smoothing of Richardson’s baseline
`curve appears to have a high level of smoothing, indicating the use of a
`greater number of intervals in the moving average. See id. (“the greater the
`number of intervals, the smoother the moving average curve”). In the
`present record, Petitioner and Petitioner’s declarant provide conclusory
`statements for supporting their contentions, which we find unpersuasive.
`See Pet. 39–40; Ex. 1003, 51–52.
`
`18
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`
`b. Claim Analysis
`i. wherein the lower threshold is adjusted based on at
`least one of a rolling average of accelerations
`
`Petitioner relies on Pasolini to disclose the use of an acceleration
`average, stating that, “[i]n Pasolini, the self-adaptive calculation of the
`negative threshold S− is based on an average of the current and previous
`acceleration data.” Id. at 48 (citing Ex. 1003, 64). Petitioner notes that
`Pasolini’s method eliminates the d.c. component so as to determine the
`acceleration datum CalAcc with zero mean value, and does so by subtracting
`the mean value of the acceleration sample (Accm) from the newly measured
`acceleration sample (Acc). Id. at 48–49 (citing Ex. 1005, 5:43–45, 5:55–
`6:5). Petitioner argues that “[a] POSITA would have recognized that the
`acceleration datum CalAcc is based on an ‘average of accelerations’ because
`CalAcc is calculated from the new acceleration sample and the values of
`previously acquired accelerations.” Id. at 49 (citing Ex.1003, 65).
`According to Petitioner, “a POSITA would have understood that Pasolini’s
`negative acceleration threshold S− is adjusted based on an ‘average of
`accelerations’ because it is adjusted based on an averaging of the
`acceleration samples as each new acceleration sample is acquired.” Id. at 50
`(citing Ex. 1003, 66).
`
`Petitioner relies on Richardson to teach the use of a rolling average of
`accelerations to establish a lower threshold for step detection. Id. at 50–52.
`Petitioner argues that “a POSITA would have recognized that Richardson’s
`‘moving average’ of acceleration data in the buffer for [a] single sample
`period is a ‘rolling average.’” Id. at 51 (citing Ex.1003, 68; Ex. 1001, 6:3–
`14; Ex. 1011, 243–44, 330). Petitioner argues that “when Richardson
`
`19
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`describes using a moving average of accelerations, that it is computing an
`average of the current sample period rather than an average of both the
`current and previous sample period.” Id. (citing Ex.1003, 69; Ex.1011, 243–
`44, 330).
`
`Patent Owner argues there is no averaging in Pasolini because, when
`removing the d.c. component from the acceleration sample, “what Pasolini
`is concerned with is the mean amplitude of the waveform, not an average of
`accelerations.” Prelim. Resp. 12 (citing Ex. 2001 ¶ 47). Patent Owner
`argues that “[t]he only operations in either of the formulas of Pasolini [(for
`calculating the mean value of the acceleration sample Accm and acceleration
`datum CalAcc)] are addition, subtraction, and multiplication by an
`undetermined constant value.” Id. at 13 (citing Ex. 1005, 5:66).
`
`Based on the record before us, Petitioner has not demonstrated that
`Pasolini bases its lower threshold on an average of accelerations. Pasolini
`removes the d.c. component (which is “due substantially to the acceleration
`of gravity”) from the acceleration curve to give the curve a zero mean value.
`Ex. 1005, 5:57–61. This centers the acceleration curve along the x axis, and
`allows the positive and negative peaks to be compared via their absolute
`values. See, e.g., id. at Figs. 2, 5, 6, 3:20–29; see also Ex. 1007, 28:57–58,
`Fig. 13a (discussing and illustrating its acceleration curve centered around
`1 G).
`Pasolini does not use an average of the acceleration data in its
`
`calculation of the threshold values. Rather, Pasolini updates the positive and
`negative envelope values based on the peaks of the acceleration signal. Ex.
`1005, 6:6–53. Specifically, regarding the negative envelope value, Pasolini
`compares each acceleration datum CalAcc (from which the d.c. component
`
`20
`
`
`
`IPR2018-01027
`Patent 8,712,723 B1
`
`has been eliminated) to the existing value Env−, and, if CalAcc is less than
`Env−, the new value for Env− is set equal to CalAcc. Id. at 6:21–25. If
`CalAcc is greater than Env−, the new value for Env− is set equal to a proper
`fraction of the previous Env− value. Id. at 6:25–28. Once the new value of
`the negative envelope value is determined, the value of the negative
`reference threshold S− is set equal to a certain proper fraction of the negative
`envelope. Id. at 6:44–47. Thus, it is the distance from the average value to
`the peak that it used to set the envelope values, from which the thresholds
`are set. This distance would be the same regardless if the mean value is
`centered along the x axis (with the d.c. component removed) or elsewhere.
`
`Even if we agreed that Pasolini’s “acceleration datum CalAcc is based
`on an ‘average of accelerations’” (Pet. 44 (emphasis added)), this does not
`mean that the lower threshold is adjusted based on the average acceleration
`as required by claim 5. To the contrary, becaus