Trials@uspto.gov
`571-272-7822
`
`
`
`
`
`Paper No. 8
`Filed: November 9, 2018
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`APPLE, INC.,
`Petitioner,
`
`v.
`
`UNILOC LUXEMBOURG S.A.
`Patent Owner.
`____________
`
`Case IPR2018-01028
`Patent 7,881,902 B1
`____________
`
`
`
`Before SALLY C. MEDLEY, JOHN F. HORVATH, and
`SEAN P. O’HANLON, Administrative Patent Judges.
`
`HORVATH, Administrative Patent Judge.
`
`
`
`DECISION
`Institution of Inter Partes Review
`35 U.S.C. § 314(a)
`
`
`
`
`
`
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`A. Background
`
`I. INTRODUCTION
`
`Apple Inc. (“Petitioner”) filed a Petition requesting inter partes review
`
`of claim 8 (“the challenged claim”) of U.S. Patent No. 7,881,902 B1
`
`(Ex. 1001, “the ’902 patent”). Paper 2 (“Pet.”). Uniloc 2017 LLC. (“Patent
`
`Owner”)1, filed a Preliminary Response. Paper 7 (“Prelim. Resp.”). We
`
`have jurisdiction under 35 U.S.C. § 314. Upon consideration of the Petition
`
`and Preliminary Response we are persuaded that Petitioner has demonstrated
`
`a reasonable likelihood that it would prevail in showing the unpatentability
`
`of challenged claim 8 of the ’902 patent. Accordingly, we institute inter
`
`partes review of claim 8.
`
`B. Related Matters
`
`Petitioner and Patent Owner identify the following as matters that
`
`could affect, or be affected by, a decision in this proceeding: Uniloc USA,
`
`Inc. v. Huawei Devices USA, Inc., 2-17-cv-00737 (E.D. Tx.); Uniloc USA,
`
`Inc. v. HTC America, Inc., 2-17-cv-01629 (W.D. Wa); Uniloc USA, Inc. v.
`
`Samsung Electronics America, Inc., 2-17-cv-00650 (E.D. Tx); Uniloc USA,
`
`Inc. v. Apple Inc., 4-18-cv-00364 (N.D. Ca);2 Uniloc USA, Inc. v. LG
`
`Electronics USA, Inc., 4-18-cv-02918 (N.D. Ca).3 Pet. 1–2; Paper 6 (2).
`
`Patent Owner also identifies the following matter before the Board, which
`
`
`1 In its updated mandatory notice, Uniloc Luxembourg S.A. identifies Uniloc
`2017 LLC as the patent owner, and Uniloc USA, Inc. and Uniloc Licensing
`USA LLC as real parties-in-interest. Paper 6 (1).
`
`2 Transferred from Uniloc USA, Inc. v. Apple Inc., 2-17-cv-00522 (E.D. Tx)
`
`3 Transferred from Uniloc USA, Inc. v. LG Electronics USA, Inc., 4-17-cv-
`00832 (E.D. Tx)
`
`2
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`involves the ’902 patent: Apple Inc. v. Uniloc Luxembourg S.A., Case
`
`IPR2018-00424 (PTAB) (challenging claims 1–6, 9, and 10 of the ’902
`
`patent). Paper 3, 2. Neither party identifies the following matters before the
`
`Board, which also involve the ’902 patent: HTC Corp. v. Uniloc
`
`Luxembourg S.A., Case IPR2018-01631 (PTAB); and Samsung Electronics
`
`America, Inc. v. Uniloc Luxembourg S.A., Case IPR2018-01653 (PTAB).
`
`Additionally, neither party identifies the following matters before the
`
`Board, which involve U.S. Patent No. 7,653,508 B1, from which the ’902
`
`patent depends: Apple Inc. v. Uniloc Luxembourg S.A., Case IPR2018-
`
`00387 (PTAB); Apple Inc. v. Uniloc Luxembourg S.A., Case IPR2018-01026
`
`(PTAB); LG Electronics, Inc. v. Uniloc Luxembourg S.A., Case IPR2018-
`
`01577 (PTAB); HTC Corp. v. Uniloc Luxembourg S.A., Case IPR2018-
`
`01589 (PTAB); and Samsung Electronics America, Inc. v. Uniloc
`
`Luxembourg S.A., Case IPR2018-01756 (PTAB).
`
`Further, neither party identifies the following matters before the
`
`Board, which involve U.S. Patent No. 8,712,723 B1, which is a continuation
`
`of the ’902 patent: Apple Inc. v. Uniloc Luxembourg S.A., Case IPR2018-
`
`00389 (PTAB); Apple Inc. v. Uniloc Luxembourg S.A., Case IPR2018-01027
`
`(PTAB); LG Electronics, Inc. v. Uniloc Luxembourg S.A., Case IPR2018-
`
`01458 (PTAB); and Samsung Electronics America, Inc. v. Uniloc
`
`Luxembourg S.A., Case IPR2018-01757 (PTAB).
`
`3
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`C. Evidence Relied Upon4
`
`References
`
`Effective Date5
`
`Exhibit
`
`Pasolini
`
`US 7,463,997 B2
`
`Oct. 2, 2006
`
`Fabio
`
`Tsuji
`
`
`
`US 7,698,097 B2
`
`Oct. 2, 2006
`
`US 7,297,088 B2
`
`Apr. 19, 2005
`
`1010
`
`1005
`
`1006
`
`D. Asserted Ground of Unpatentability6
`
`Petitioner asserts the following ground of unpatentability:
`
`References
`
`Basis
`
`Claim Challenged
`
`Fabio, Pasolini, and Tsuji
`
`§ 103(a)
`
`8
`
`II. ANALYSIS
`
`A. Discretion Under 35 U.S.C. § 325(d)
`
`Petitioner previously challenged claims 1–6, 9, and 10 of the ’902
`
`patent, including claims 5, 6, 9, and 10 as unpatentable over Fabio and
`
`Pasolini. See Apple Inc. v. Uniloc Luxembourg S.A., Case IPR2018-00424,
`
`
`4 Petitioner also relies upon the Declaration of Joseph A. Paradiso, Ph.D.
`(Ex. 1003).
`5 Petitioner relies on the filing dates of Pasolini, Fabio, and Tsuji as the
`effective date for determining their availability as prior art under 35 U.S.C.
`§ 102(e). Pet. 8–9.
`
`6 Petitioner also challenges claim 5 as obvious over Fabio and Pasolini.
`Pet. 8. However, Petitioner states it “seeks review with respect to only
`previously unchallenged claim 8,” and that “this petition is directed toward
`only claim 8.” Id. at 3. Moreover, Petitioner states that its challenge of
`claim 5, from which claim 8 depends, repeats the same analysis it used to
`challenge claim 5 in IPR2018-00424. Therefore, we interpret this Petition as
`challenging only dependent claim 8.
`
`4
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`slip op. at 8 (PTAB) (Paper 2) (“the ’424 petition”). Here, Petitioner
`
`challenges claim 8 of the ’902 patent as unpatentable over Fabio, Pasolini,
`
`and Tsuji. See Pet. 8.
`
`The Board has discretion to institute inter partes review, and can deny
`
`institution of follow-on petitions, like the present petition, as an exercise of
`
`that discretion. See 35 U.S.C. § 325(d) (“In determining whether to institute
`
`or order a proceeding under this chapter, chapter 30, or chapter 31, the
`
`Director may take into account whether, and reject the petition or request
`
`because, the same or substantially same prior art or arguments previously
`
`were presented to the Office.”).
`
`Petitioner argues the instant petition is not redundant to the ’424
`
`petition and should be allowed to proceed because “it seeks review with
`
`respect to only previously unchallenged claim 8,” and because “the prior art
`
`teaching the limitations of claim 8 was not located by Petitioner until after
`
`the ’424 petition was filed.” Pet. 3. Petitioner argues that in preparing this
`
`petition it has not benefited from Patent Owner’s preliminary response to the
`
`IPR2018-00389 petition (“the ’389 petition”), which challenged claims of
`
`the related ’723 patent, because “the analysis of claim 5 in this petition is
`
`verbatim identical to the analysis of claim 5 presented in the ’424 petition.”
`
`Id.
`
`Patent Owner argues the instant petition should be denied under 35
`
`U.S.C. § 325(d) because “Petitioner knew, or should have known, of the
`
`additional reference (Tsuji) at the time of filing” the ’424 petition. Prelim.
`
`Resp. 3. Patent Owner argues this is so because Tsuji is one of a small
`
`number of references cited on the face of Fabio, and Petitioner knew about
`
`5
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`and relied on Fabio when it filed the ’424 petition.7 Id. at 4–6. Patent
`
`Owner further argues the instant petition should be denied because
`
`“Petitioner had the benefit of Patent Owner’s preliminary response” to the
`
`’389 petition, and has not stated that (a) it “did not choose to file the instant
`
`follow-on petition based in part on having the benefit of Patent Owner’s
`
`preliminary response,” and (b) “its arguments and statements against the
`
`newly challenged dependent claims were not formulated based in part on
`
`having the benefit of Patent Owner’s preliminary response.” Id. at 3, 6–7.
`
`Based on the facts of this case, we decline to deny the Petition as an
`
`exercise of our discretion. First, Petitioner is challenging claim 8, a claim
`
`Petitioner did not challenge in the ’424 petition. Second, although Tsuji is
`
`cited on the face of Fabio and Petitioner relied on Fabio in some of its
`
`challenges in the ’424 petition, those facts, by themselves, are insufficient
`
`here for us to deny the Petition for a newly challenged claim. Finally,
`
`although Petitioner did have the benefit of reading Patent Owner’s
`
`preliminary response in the ’389 petition, the subject matter of claim 8
`
`differs from the subject matter claimed in any of the claims of the ’723
`
`patent that was challenged in the ’389 petition. Patent Owner does not
`
`explain how the knowledge Petitioner gained from reading Patent Owner’s
`
`preliminary response to the ’389 petition could have helped Petitioner to
`
`formulate its positions regarding the patentability of claim 8 of the ’902
`
`
`7 Patent Owner alleges Tsuji is one of five references cited on the face of
`Fabio. Prelim. Resp. 5. More precisely, Tsuji is one of five U.S. Patent
`Documents cited on the face of Fabio. See Ex. 1006 [56]. Also cited are
`three Foreign Patent Documents, and the translations of a Japanese patent
`and patent application. Id. Thus, Tsuji is one of ten references cited on the
`face of Fabio.
`
`6
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`patent. Therefore, we decline to deny the petition as an exercise of our
`
`discretion, and proceed to the merits of Petitioner’s challenge.
`
`B. The ’902 Patent
`
`The ’902 patent relates to “a method of . . . counting periodic human
`
`motions such as steps.” Ex. 1001, 1:9–11. The method involves the use of a
`
`“portable electronic device that includes one or more inertial sensors. . . .
`
`[that] measure accelerations along a single axis or multiple axes.” Id. at
`
`2:24–28. The measured accelerations may be linear or rotational. Id. at
`
`2:28–29.
`
`Figure 1 of the ’902 patent is reproduced below.
`
`Figure 1 of the ’902 patent is a block diagram illustrating electronic device
`
`100. Id. at 1:47–48. Device 100 includes acceleration measuring logic 105
`
`(e.g., inertial sensors), dominant axis logic 127, and step counting logic 130.
`
`
`
`7
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`Id. at 2:19–24, 2:38–43, Fig. 1. Device 100 “may be used to count steps or
`
`other periodic human motions.” Id. at 2:29–30. In the context of the ’902
`
`patent, a “step” is “any user activity having a periodic set of repeated
`
`movements.” Id. at 3:34–38. According to the ’902 patent, device 100
`
`accurately counts steps “regardless of the placement and/or orientation of the
`
`device on a user,” and regardless of whether the device “maintains a fixed
`
`orientation or changes orientation during operation.” Id. at 2:31–35.
`
`
`
`Dominant axis logic 127 includes cadence logic 132 and rolling
`
`average logic 135. Id., Fig. 1. Inertial sensors 105 measure acceleration
`
`data, and cadence logic 132 analyzes this data to detect “a period and/or
`
`cadence of a motion cycle” or step, which may be based on user activity
`
`such as running or walking. Id. at 2:38–40, 3:14–18, 3:46–51. Cadence
`
`logic 132 determines “a cadence window 150 to be used by the step counting
`
`logic 130.” Id. at 3:11–14. Cadence window 150 is “a window of time
`
`since a last step was counted and that is looked at to detect a new step.” Id.
`
`at 3:65–4:1. Initially, cadence window 150 is set to a default value, but that
`
`default value can be dynamically updated after each step once a minimum
`
`number of steps have been detected to reflect the cadence or period of the
`
`detected steps. Id. at 3:57–61, 4:22–28, 4:61–5:6. The cadence or stepping
`
`period can be determined as a “rolling average of the stepping periods over
`
`previous steps.” Id. at 3:61–62.
`
`Cadence logic 132 also determines “one or more sample periods to be
`
`used by the rolling average logic 135.” Id. at 3:11–14, 5:31–34. The sample
`
`period is set to be “the length of, or longer than, the stepping period,”
`
`including a “multiple of the stepping period.” Id. at 5:34–37. Rolling
`
`average logic 135 “creates one or more rolling averages of accelerations . . .
`
`8
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`measured by the inertial sensor(s) over the sample period(s) set by the
`
`cadence logic 132.” Id. at 5:39–41. These rolling averages are used to
`
`determine an orientation of the electronic device and a threshold against
`
`which acceleration measurements are compared. Id. at 5:41–45.
`
`Dominant axis logic 127 includes dominant axis setting logic 140,
`
`which determines an orientation of device 100 or the inertial sensor(s) within
`
`device 100. Id. at 6:8–10. This may be done “based upon the rolling
`
`averages of accelerations created by the rolling average logic 135.” Id. at
`
`6:10–12. In particular, “[t]he axis with the largest absolute rolling average”
`
`over a given sampling period is identified as the “axis most influenced by
`
`gravity,” and is designated the dominant axis. Id. at 6:14–18, 6:23–25. The
`
`’902 patent explains that because the device orientation may change over
`
`time, the rolling average acceleration may change and “a new dominant axis
`
`may be assigned when the orientation of the electronic device 100 and/or the
`
`inertial sensor(s) attached to or embedded in the electronic device 100
`
`changes.” Id. at 6:16–22. In addition, dominant axis setting logic 140 can
`
`set the dominant axis to be a virtual “axis that is defined as approximately
`
`aligned to gravity” that is found “by doing trigonometric calculations on the
`
`actual axes based on the gravitation influence” on those axes. Id. at 6:25–
`
`34.
`
`Step counting logic 130 includes measurement selection logic 145,
`
`measurement comparator 155, and threshold comparator 160. Id. at 6:38–
`
`412. Measurement selection logic 140 “monitor[s] accelerations relative to
`
`the dominant axis, and select[s] only those measurements with specific
`
`relations to the dominant axis.” Id. at 6:44–47. “Selected measurements
`
`9
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`[are] forwarded to the measurement comparator 155 and the threshold
`
`comparator 160 to determine whether a step has occurred.” Id. at 6:57–59.
`
`A method for determining whether a step has occurred is disclosed in
`
`Figure 8 of the ’902 patent, which is reproduced below.
`
`
`
`
`
`10
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`Figure 8 is a flow diagram of a method for recognizing that a step has
`
`occurred. Id. at 2:1–4, 12:25–27. Acceleration measurement data is
`
`received and filtered to remove low and high frequency components. Id. at
`
`12:31–38, Fig. 8 (steps 805 and 810). A dominant axis is assigned as
`
`described above. Id. at 40–44, Fig. 8 (step 812). Because only steps within
`
`the cadence window are counted, a determination is made whether the
`
`measured acceleration is within the cadence window. Id. at 12:45–48, Fig. 8
`
`(step 815). If it is, three additional tests are performed to determine whether
`
`the measured acceleration can be counted as a step. First, the absolute value
`
`of the measured acceleration along the dominant axis must be greater than a
`
`lower threshold, such as the rolling average acceleration along the dominant
`
`axis. Id. at 7:9–12, 12:51–55, 12:64–65, and Fig. 8 (step 820). Second, the
`
`absolute value of the measured acceleration along the dominant axis must be
`
`greater than the absolute value of previous measured accelerations along the
`
`dominant axis. Id. at 7:9–12, 13:34–38, 13:53–56, and Fig. 8 (step 825).
`
`Third, the absolute value of the measured acceleration along the dominant
`
`axis must be lower than an upper threshold. Id. at 7:9–12, 13:59–62, 13:66–
`
`14:1, and Fig. 8 (step 830). The upper threshold “prevent[s] sudden
`
`accelerations such as taps from being counted as steps.” Id. at 14:1–3.
`
`Device 100 is battery operated, and therefore has multiple operating
`
`modes to preserve battery life, including sleep mode 305, entry mode 315,
`
`stepping mode 325, and exit mode 335. Id. at 8:16–18. The power level of
`
`device 100 is linked to these modes. Id. at 8:18–19. The different modes
`
`and the relationships between them are shown in Figure 3 of the ’902 patent,
`
`which is reproduced below.
`
`11
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`
`
`Figure 3 of the ’902 patent is a state diagram showing the different modes of
`
`electronic device 100. Id. at 1:52–54. When no acceleration data is
`
`measured, device 100 is in sleep mode 305. Id. at 8:20–22. When
`
`acceleration data is detected, device 100 enters entry mode 315 to detect
`
`steps in the acceleration data. Id. at 8:22–25. If a predetermined number
`
`(N) of steps are detected in a sampling period, device 100 enters stepping
`
`mode 325; otherwise it reverts to sleep mode 305. Id. at 8:25–28. In
`
`stepping mode 325, steps are detected and counted as described above until
`
`no steps are detected within the cadence window, at which point device 100
`
`enters exit mode 335. Id. at 8:30–37. In exit mode 335, device 100
`
`determines whether a predetermined number (X) of steps are detected at a
`
`particular cadence. Id. at 8:38–40. If so, device 100 reverts to stepping
`
`mode 325; if not, device 100 reverts to entry mode 315.
`
`12
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`The method by which device 100 transitions from entry mode 315 to
`
`stepping mode 325 is shown in Figure 5, which is reproduced below.
`
`Figure 5 of the ’902 patent is a flow chart of device 100 operating in entry
`
`mode 315. Id. at 1:58–60. After setting a sampling rate (504), a first step is
`
`detected in the acceleration data (510), a default cadence window is set
`
`
`
`13
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`(514), and a temporary or buffered step count is set to one (520). Id. at
`
`9:55–10:8, 10:25. Next additional steps are searched for in the acceleration
`
`data (524) using the criteria discussed above, including whether the
`
`acceleration data falls within the cadence window. Id. at 10:25–30, 12:45–
`
`46, and Fig. 8.
`
`
`
`When additional steps are detected in the acceleration data (524), they
`
`are added to the buffered step count (560). Id. at 10:46–47. When the
`
`number of steps in the buffered step count is less than a predetermined
`
`number M (564), additional steps are looked for in the acceleration data
`
`(524). Id. at 10:47–52. When the number of steps in the buffered step count
`
`reaches predetermined number M (564), a new cadence window is set based
`
`on the cadence of the M steps (574), and additional steps are looked for in
`
`the acceleration data (524) until a predetermined number of N steps is
`
`counted in the buffered step count (580). Id. at 10:53–67. When the number
`
`of steps in the buffered step count reaches predetermined number N, the
`
`number of buffered steps are added to an actual step count (584), and device
`
`100 enters stepping mode 325. Id. at 10:67–11:3. In stepping mode 325, the
`
`cadence window is dynamically updated based on the rolling average of
`
`previously measured stepping periods. Id. at 11:13–17.
`
`As discussed above, measured acceleration data is only counted as a
`
`step when it falls within the cadence window. Id. at 10:25–30, 12:45–46,
`
`and Fig. 8. Measured acceleration data can fall outside the cadence window
`
`because it is too early or too late. If it is too early, time remains within the
`
`cadence window (530) and additional steps are looked for in the acceleration
`
`data (524). Id. at 10:32–36. If it is too late, no time remains within the
`
`cadence window (530), the buffered step count is reset (534), and the
`
`14
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`acceleration data is searched for another first step (540/510). Id. at 10:36–
`
`43.
`
`C. Illustrative Claims
`
`Of the challenged claims, claim 5 of the ’902 patent is independent,
`
`and claim 8 depends directly from it. Claims 5 and 8 are reproduced below.
`
`5. A method for a mobile device comprising:
`
`receiving acceleration data that meets stepping
`criteria from an accelerometer included in the
`mobile device;
`
`incrementing a step count in a step count buffer;
`
`when at least one of a) the step count is below a
`step count threshold, or b) a current user cadence
`fails to match a step cadence of a user profile,
`using a default step cadence window to identify a
`time frame within which to monitor for a next step;
`and
`
`when the step count is at or above the step count
`threshold, determining a dynamic step cadence
`window and using the dynamic step cadence
`window to identify the time frame within which to
`monitor for the next step.
`
`8. The method of claim 5, wherein determining
`the dynamic step cadence window comprises:
`
`computing a rolling average of stepping periods of
`previously counted steps; and
`
`setting the dynamic step cadence window based on
`the rolling average of stepping periods
`
`
`Ex. 1001, 15:46–16:6, 16:22–27.
`
`15
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`D. Claim Construction
`
`In an inter partes review, claim terms of an unexpired patent are given
`
`their broadest reasonable interpretation in light of the specification of the
`
`patent in which they appear. 37 C.F.R. § 42.100(b). Under the broadest
`
`reasonable interpretation standard, claim terms are generally given their
`
`ordinary and customary meaning, as would be understood by one of ordinary
`
`skill in the art, in the context of the entire disclosure. In re Translogic Tech.,
`
`Inc., 504 F.3d 1249, 1257 (Fed. Cir. 2007). Only claim terms which are in
`
`controversy need to be construed and only to the extent necessary to resolve
`
`the controversy. See Vivid Techs., Inc. v. Am. Sci. & Eng’g, Inc., 200 F.3d
`
`795, 803 (Fed. Cir. 1999).
`
`Petitioner proposes a construction for the term “cadence window,”
`
`which Petitioner argues is defined in the Specification to mean “a window of
`
`time since a last step was counted that is looked at to detect a new step.”
`
`Pet. 7 (citing Ex. 1001, 3:66–4:1). Patent Owner contends this term need not
`
`be construed to resolve any controversy in this proceeding. Prelim. Resp. 8–
`
`9. At this stage of the proceeding, we agree with Patent Owner that the term
`
`“cadence window” need not be construed because its express construction is
`
`not needed to resolve any contention between the parties.
`
`E. Overview of the Prior Art
`
`1. Fabio
`
`Fabio discloses a method for “controlling a pedometer based on the
`
`use of inertial sensors.” Ex. 1006, 1:10–11. The pedometer can be
`
`“integrated within a portable electronic device, such as a cell phone.” Id. at
`
`2:34–36. The method involves:
`
`16
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`generating a signal correlated to movements of a
`user of the pedometer; detecting steps of the user
`based on the signal; checking whether sequences
`of the detected steps satisfy pre-determined
`conditions of regularity; updating a total number of
`valid steps if the conditions of regularity are
`satisfied; and preventing updating of the total
`number of valid steps if the conditions of
`regularity are not satisfied.
`
`Id. at 1:62–2:3. Fabio detects user steps from the sampled acceleration data
`
`AZ of its inertial sensor according to a method illustrated in Figures 5 and 6,
`
`which are reproduced below.
`
`
`
`Figure 5 of Fabio is a graph illustrating quantities used to detect user steps
`
`from acceleration data. Id. at 2:22–23. A step is detected when a positive
`
`peak of acceleration signal AZ is greater than threshold AZP, and a negative
`
`peak is less than threshold AZN and falls within fixed time window TW from
`
`the positive peak. Id. at 4:15–21.
`
`17
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`
`
` A detected step is validated when it falls within a variable time
`
`window TV as illustrated in Figure 6 of Fabio, which is reproduced below.
`
`
`
`Figure 6 of Fabio is a graph illustrating quantities used to validate user steps
`
`detected from acceleration data. Id. at 2:24–25. Figure 6 illustrates a
`
`sequence of user steps detected at times TR(1), TR(2) . . . TR(K-2), TR(K-1),
`
`and TR(K) according to the method disclosed in Figure 5. The time between
`
`steps detected at times TR(K-1) and TR(K-2) is ∆TK-1; the time between steps
`
`detected at times TR(K) and TR(K-1) is ∆TK. Id. at 4:28–35. For the step
`
`detected at time TR(K) to be validated as a step, it must fall within variable
`
`time window TV, i.e., TR(K) must be greater than TR(K-1) + ½ (∆TK-1) and
`
`less than TR(K-1) + 2 (∆TK-1). Id. at 4:35–52. Although Figure 6 depicts TV
`
`as being asymmetric about time TR(K-1) + ∆TK-1 and as having a variable
`
`width ∆TK-1 (e.g., due to its dependence on the variable time between
`
`previous steps), Fabio teaches time window TV can be “symmetrical and a
`
`have a different amplitude.” Id. at 4:52–53. Fabio further teaches time
`
`window TV ensures “the duration ∆TK of a current step K is substantially
`
`homogeneous with respect to the duration ∆TK-1 of an immediately
`
`preceding step.” Id. at 4:28–31.
`
`18
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`When counting steps, Fabio discloses that isolated “or very brief
`
`sequences of steps are far from significant and should preferably be ignored
`
`because they are, in effect, irrelevant.” Id. at 1:47–50. To ignore such steps,
`
`Fabio discloses a two-stage counting procedure, illustrated in Figure 3,
`
`which is reproduced below.
`
`
`
`Figure 3 of Fabio is a flow chart depicting a two-stage counting procedure.
`
`Id. at 2:17–19. Initially, valid step counter NVT, valid control step counter
`
`NVC, and invalid step counter NINV are set to zero (100). Id. at 3:13–18.
`
`Next, first counting procedure 110 counts steps by sampling acceleration
`
`signal AZ at a predetermined frequency. Id. at 3:19–21. First counting
`
`procedure 110 terminates and sets a state flag to C when “a regular gait of
`
`the user is recognized,” or terminates and sets the state flag to PD when “a
`
`time interval . . . that has elapsed from the last step recognized is longer than
`
`a first time threshold.” Id. at 3:27–36. If the state flag is set to C, second
`
`counting procedure 130 executes, otherwise survey procedure 140 executes
`
`and the pedometer is placed in a wait or power down state. Id. at 3:37–41,
`
`3:50–53.
`
`19
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`
`
`Fabio illustrates first counting procedure 110 in Figure 4, which is
`
`reproduced below.
`
`
`
`Figure 4 of Fabio is a flowchart illustrating first counting procedure 110. Id.
`
`at 2:20–21, 3:58–59. Acceleration data AZ is sampled (200), and if the time
`
`Tc between the sample time and the time of the last detected step is greater
`
`than a first threshold TS1 (205), state flag FST is set to PD (210), and survey
`
`procedure 140 is called. Id. at 3:60–4:2; see also Fig. 3. Otherwise, if Tc is
`
`less than a second, shorter, threshold TS2 (215), the sample is checked to see
`
`if it is both recognized as a step according to the procedure disclosed in
`
`Figure 5 (225) and validated as a step according to the procedure disclosed
`
`in Figure 6 (230). Id. at 4:2–55. As discussed above, for the sample to be
`
`validated as a step, it must fall within time window TV.
`
`If the sample is recognized (225) and validated (230) as a step, the
`
`number of valid control steps NVC is incremented (255) and compared to a
`
`20
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`threshold NT2 (260). Id. at 5:10–22. If NVC is less than NT2, another sample
`
`of acceleration data is obtained (200) and the process continues. Id. at 5:22–
`
`27. However, if NVC is equal to NT2 (260), the number of valid control steps
`
`NT2 is added to the number of valid steps NVT, state flag FST is set to the
`
`count value C (i.e., NVT), and second counting procedure 130 is called (265).
`
`Id. at 5:30–39; see also Fig. 3.
`
`It the acceleration sample is recognized as a step (225), but not
`
`validated (230) because it falls outside of time window TV, an invalid step
`
`count NINV is incremented (235), a new acceleration data sample is obtained
`
`(step 200), and the process is repeated.
`
`Fabio’s second counting procedure 130 is illustrated in Figure 7,
`
`which is reproduced below.
`
`
`
`21
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`Figure 7 of Fabio is a flowchart illustrating second counting procedure 130.
`
`Id. at 6:12–13. Acceleration data AZ is sampled (300), and if the time Tc
`
`between the sample time and the time of the last detected step is greater than
`
`threshold TS2 (305), second counting procedure 130 is terminated. Id. at
`
`6:14–20; see also Fig. 3. Otherwise, the sample is checked to see if it is both
`
`recognized as a step according to the procedure disclosed in Figure 5 (315)
`
`and validated as a step according to the procedure disclosed in Figure 6
`
`(320). Id. at 6:21–39. If the sample is recognized and validated as a step,
`
`the number of valid steps NVT is incremented (325), another sample of
`
`acceleration data is obtained (300), and the process continues searching for
`
`valid steps to count. Id. at 6:40–53. Second counting procedure 130
`
`continues until either (a) the time Tc between an acceleration data sample
`
`and the time of the last detected step is greater than threshold TS2 (305), or
`
`(b) a sample is not validated as a step (320) and the incremented number of
`
`invalid steps NINV (340) reaches threshold NT4 (345). Id. at 6:14–20, 6:54–
`
`7:6. When second counting procedure 130 terminates, first counting
`
`procedure 110 is executed. Id., Fig. 3.
`
`
`
`In addition to disclosing a two-stage counting procedure as described
`
`above, Fabio discloses that when the pedometer’s inertial sensor contains
`
`two or more detection axes, step recognition is “performed by selecting the
`
`acceleration signal corresponding to the detection axis nearest to the
`
`vertical.” Id. at 8:21–25. The nearest-to-vertical axis “is selected on the
`
`basis of the value of the DC component of the respective acceleration signal,
`
`which is correlated to the contribution of the acceleration of gravity.” Id. at
`
`8:27–30. This allows steps to be counted “independently of how [the
`
`pedometer] is oriented.” Id. at 8:32–33.
`
`22
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`Fabio uses survey procedure 140 to identify the nearest-to-vertical
`
`detection axis. See id. at 7:21–59, Fig. 3. Survey procedure 140 executes
`
`whenever first counting procedure 110 terminates for failure to detect a step
`
`within a time TS1 from a previously detected step. Id. at 3:29–32, 3:50–53,
`
`Fig. 3. It initially determines and stores mean value AZM of the DC
`
`component of acceleration signal AZ in non-volatile memory. Id. at 7:27–
`
`31. Mean value AZM represents the inertial sensor’s acceleration due to
`
`gravity along its detection axis, and indicates the pedometer’s orientation.
`
`Id. at 7:30–37. The pedometer then enters a sleep state, and periodically
`
`wakes up (e.g., every 10 seconds) to determine an updated mean value AZM'
`
`of the DC component of acceleration signal AZ. Id. at 7:46–48. If AZM' is
`
`substantially the same as AZM, the pedometer goes back to sleep. Id. at
`
`7:54–59. Otherwise, first counting procedure 110 is executed, i.e., first
`
`counting procedure is executed whenever the pedometer’s orientation
`
`changes. Id. at 7:51–54.
`
`2. Pasolini
`
`Pasolini discloses a pedometer having a “step detection method using
`
`an algorithm for self-adaptive computation of acceleration thresholds.” Ex.
`
`1005, 1:10–12. The pedometer can be housed inside a portable device, such
`
`as a mobile phone, and includes an accelerometer having multiple detection
`
`axes. Id. at 2:60–63, 8:11–13, 8:31–34. The pedometer’s step detection
`
`algorithm “identif[ies] the main vertical axis to be used for step detection as
`
`the axis of detection that has the highest mean acceleration value Accm (on
`
`account of gravity),” and does so “at each acquisition of a new acceleration
`
`sample . . . to take into account variations in the orientation of the pedometer
`
`23
`
`

`

`IPR2018-01028
`Patent 7,881,902 B1
`
`
`device 1, and consequently of the accelerometer 2 arranged inside it.” Id. at
`
`8:16–24.
`
`Pasolini’s step detection algorithm “compares the value of the
`
`[measured] acceleration signal A . . . with a positive reference threshold S+
`
`and with a negative reference threshold S-, for identifying, respectively, the
`
`positive phase (positive acceleration peak) and the negative phase (negative
`
`acceleration peak) of the step.” Id. at 3:35–41. Once a positive acceleration
`
`phase is identified, a negative acceleration phase is searched for “within a
`
`certain ti