US 7003402

I'll search for authoritative information on US patent 7003402, including any litigation or CAFC activity.

Let me run additional targeted searches for litigation, PTAB, and assignment records for this specific patent.

US Patent 7,003,402 — Analyst Summary

Source of record: Full patent text (Google Patents, fetched 2026‑09‑24). Metadata cross‑checked against Justia Patents.

Bibliographic Data

Family members: WO 2002103397 A1; EP 1399764 A1; GB 2376527 B; CN 100487490 C; AU 2002302834 B2; RU 2309436 C2; NO 20035615 D0.

Abstract (verbatim)

"A method of processing data sequences obtained at substantially the same time at locations spatially separated from one another is disclosed. A sequence indicative of a measure of spread of the data sequences or of selected data sequences is determined. A common signal data sequence may also be determined from the (selected) data sequences. The measure of spread of the data sequences is preferably normalized, for example relative to the absolute value of the common signal data sequence. The measure of spread is a measure of the noise to signal ratio in the initially‑obtained data sequences. It may be used to control the parameters of other processing steps performed on the data sequences. Alternatively or additionally, the measure of spread may be output, for example for monitoring by an observer. The method can be applied to the processing of seismic data obtained using a single sensor seismic data acquisition system."

Plain‑Language Overview of the Independent Claims

The patent has three claim sets that read as independent in form (claim 1, claim 28, and claim 30). Note that claims 7 and 31 are formally dependent on claim 1/7 but add substantial separate subject matter (multi‑sensor selection), so I flag them below.

• Claim 1 (method — core concept): Compute, from two data sequences recorded by two spatially separated sensors that sampled substantially the same signal, a third sequence representing a "measure of spread" between them. That spread sequence is used as an estimate of the initial uncertainty in the signal. This is the fundamental claim — no common‑signal determination or normalization is required.

• Claim 7 (multi‑sensor variant; depends from claim 1): Rather than just two sensors, (a) select two or more sequences from a plurality of sequences taken at respective spaced sensors, and (b) determine a measure‑of‑spread sequence for the selected set.

• Claim 28 (apparatus): A device comprising means for performing the claim‑1 determination — i.e., the apparatus counterpart to the core method.

• Claim 29 (apparatus; depends from 28): The apparatus further comprises a programmable data processor.

• Claim 30 (storage medium): A storage medium containing a program for the data processor of the apparatus as defined in claim 29 — a Beauregard‑style program‑product claim.

Key dependent‑claim highlights: claim 2 (determine a common‑signal sequence), claim 5 (normalize the spread relative to the absolute value of the common signal), claim 6 (spread ÷ |common signal|), claim 8 (common signal for the selected multi‑sensor set), claim 13 (group sequences into ensembles), claim 18/19 (spread = standard deviation / variance), claim 21 & 23 (control processing parameters based on the spread), claim 25/27 (seismic data / single‑sensor seismic acquisition).

Technical Context (plain language)

The invention exploits Single Sensor Seismic (SSS) acquisition (e.g., Schlumberger's Q‑Marine system), where each sensor's output is digitized and individually retained rather than hard‑wired into analog groups. Because sensors within a group are spaced closely enough to over‑sample the wavefield and see "substantially the same" signal, differences between their recorded traces are attributable primarily to local noise. The patent computes a spread metric across those traces (preferably standard deviation, normalized by the absolute value of the ensemble mean) to yield a per‑sample noise‑to‑signal estimate that can be used for quality control, to steer/tune noise‑attenuation processing, and to propagate uncertainty into later inversion/imaging steps.

Litigation / CAFC 2026 Docket Search — Result

I ran targeted searches against USPTO/PTAB materials and Federal Circuit 2026 docket sources for patent 7,003,402 specifically (not similar numbers). I found no CAFC 2026 appeal, PTAB proceeding (IPR/PGR), or district‑court docket naming US 7,003,402. Searches for "7,003,402" returned only unrelated patents (e.g., 7,402,903 in Katana Silicon v. GlobalFoundries; unrelated "’402" patents in semiconductors and wireless printing). The patent is also listed as expired (fee‑related), which is consistent with no active enforcement.

Uncertainty caveat: General web search does not comprehensively index PACER, CourtListener, or the cafc.uscourts.gov docket database. Therefore I can say I found no evidence of 2026 CAFC activity, but I cannot affirmatively rule out an unindexed filing. If you need a definitive answer, the authoritative step is a direct PACER/CourtListener docket query by patent number and by party name (WesternGeco / Schlumberger).

Notes on Literal Interpretation / Anomalies

• Claim 24 recites "The method of claim 31," and claim 31 in turn depends from claim 7 — an unusual dependency chain (a higher‑numbered claim depending on a lower‑numbered one). I have reported this literally as issued and have not "corrected" it; it may reflect a renumbering or clerical artifact in the granted text.
• The patent number is US 7,003,402 B2. I did not substitute or auto‑correct any similarly numbered patent.
• Assignee is recorded as WesternGeco L.L.C.; the reassignment record shows assignment from inventors Curtis and Christie to WESTERNGECO L.L.C. (recorded 2003‑11‑24).

If you'd like, I can also map each independent claim to the specification's supporting passages (steps 11–23 of FIG. 4) for a validity/claim‑construction‑style analysis.

I'll search for litigation involving US patent 7003402 specifically.

Initial searches returned irrelevant matches. Let me search more targeted litigation databases.

Let me search additional litigation-specific sources directly.

Litigation Search: US Patent No. 7,003,402

Bottom line

I found no litigation specifically involving US Patent No. 7,003,402 ("Method of processing data," WesternGeco L.L.C., Christie & Curtis). Within the searches I was able to run, no case in any court, PTAB proceeding, or ITC investigation was identified that names this patent.

What I searched

• General web searches for the patent number combined with "litigation," "lawsuit," "infringement," and "case."
• Searches targeting litigation-oriented terms and the assignee/inventors (WesternGeco; Christie; Curtis; seismic "method of processing data").
• Attempts to reach the Unified Patents litigation portal and other docket-type sources for this specific number.

The searches returned only false positives — i.e., unrelated uses of the digit string "7003402":

• A Sub-Zero/Wolf appliance retrofit-kit part number (7003402);
• A Brazilian design registration DI 7003402-8;
• A Westlaw citation "2023 WL 7003402" in an employment-discrimination case.

None of these involve the patent.

Important caveat about a related (but different) case

There is a well-known seismic patent campaign — WesternGeco LLC v. ION Geophysical Corp. (and a consolidated case against the Fugro entities), litigated in the U.S. District Court for the Southern District of Texas (lead case No. 4:09-cv-01827) and ultimately reaching the U.S. Supreme Court (585 U.S. 407 (2018)). However, the patents asserted in that case were U.S. Patent Nos. 6,932,017; 7,080,607; 7,162,967; 7,293,520; and 6,691,038 — not 7,003,402. I must be explicit: 7,003,402 is not among the patents asserted in that litigation, so it would be incorrect to list WesternGeco v. ION Geophysical (or the Fugro case) as litigation "involving" US 7,003,402.

Limitations and how to verify

• My searches were capped before I could fully exhaust docket-specific databases. As a result, "no litigation found" here should be treated as inconclusive rather than as a definitive negative.
• To confirm with authoritative sources, the following should be checked directly for the literal number 7003402 / 7,003,402:
  • Unified Patents litigation portal (portal.unifiedpatents.com/litigation/caselist),
  • PACER / CourtListener RECAP docket search,
  • USPTO Patent Center "Legal Events" / "Patent Trial" tabs and the USPTO Assignment database,
  • Google Patents "Litigation" tab and Lex Machina / Docket Alarm.
• Note that PTAB challenges (IPR/PGR/CBM) are sometimes tracked separately from district-court litigation; I did not identify any PTAB proceeding for this number either.

If you can share which database you want prioritized (Unified Patents, PACER, CourtListener, or PTAB), I can attempt a narrower, number-exact verification pass.

Summary table

I'll verify the structured ODP data against live web sources before writing anything up.

The first pass surfaced only noise. Let me run more targeted queries before concluding.

Proceedings overview

Zero AIA trial proceedings are on file for US 7,003,402 — no IPR, no PGR, no CBM — so there is no claim-level PTAB outcome, no FWD, and no § 315(e)(2) estoppel attached to this patent. That is the canonical answer from the structured "PTAB proceedings on file" block (USPTO Open Data Portal), and my live searching found nothing that contradicts it. The defensively significant corollary is that the patent is now expired: Google Patents records a legal status of "Expired – Fee Related" with an adjusted expiration of 2022-12-22 (US7003402B2). A defendant facing a demand letter on this patent is facing a dead patent with no litigated validity history at all — the opposite of a hardened, IPR-surviving asset.

Because there are no proceedings, the per-proceeding template has nothing to populate. Rather than manufacture entries, here is what I verified and what it means.

No proceedings to report

I could not identify any petitioner, any institution decision, any FWD, any settlement, or any Federal Circuit appeal arising from US 7,003,402. There is therefore no proceeding number to give, and I will not invent one. Caveat on confidence: the ODP feed is authoritative for AIA trials, but it is a feed, not a full-text docket scrape. I attempted to corroborate against PTAB E2E-style material via web search and the searches were dominated by false positives (see below); a belt-and-braces check would be to run "7,003,402" in PTAB E2E and in Docket Alarm under the patent-number field before relying on this for a filing decision. My conclusion is "no activity found," not "provably impossible."

False positives I discarded, so you don't chase them:

• US 6,725,402 — an Ocean Semiconductor-related patent; the Fintiv briefing surfacing in search (referring to parallel Ocean litigations and a tentative 2022-12-07 trial date) is about that patent, not 7,003,402.
• US 11,597,402 — Tesla v. Granite Vehicle Ventures, IPR2025-00943. Unrelated.
• "7003402" also appears as a Sub-Zero refrigerator door-panel part number. Purely coincidental.

Why the WesternGeco IPR cluster does not touch this patent

This is the most likely place a defendant would look for a shortcut, so be precise about it. WesternGeco's contested patents were the Bittleston family — US 7,080,607, US 7,162,967, US 7,293,520, US 6,932,017 — and the Zajac patent US 6,691,038. The PGS-filed IPRs against WesternGeco (six petitions, e.g. the IPR2015-003xx series, with ION joining) and the WesternGeco v. ION litigation all concern streamer positioning devices ("birds"), not data processing. The Federal Circuit affirmed the Board's invalidation of the instituted claims in those six proceedings — see WesternGeco LLC v. ION Geophysical Corp., 889 F.3d 1338 (Fed. Cir. 2018) (opinion) — but none of that reasoning, and none of that estoppel, reaches US 7,003,402. Separately, WesternGeco petitioned against PGS on US 6,906,981 and US 6,026,059 (e.g. IPR2015-00309/00310/00311, final decision 2016-06-08 holding claims 31, 32, 35–37 unpatentable while 33, 34, 38 survived). Again: different patents, different technical field, and WesternGeco was the petitioner, not the owner.

The subject matter of US 7,003,402 is instead the Christie/Curtis "measure of spread" / normalised noise-to-signal estimate for single-sensor seismic (SSS) data — see claim 1 and the NES_j formulation at col. 9. It claims priority to GB 0114744.6 (2001-06-16), filed as PCT/GB2002/002774 on 2002-06-17, granted 2006-02-21. Its foreign siblings are GB 2376527, EP 1399764, CN 100487490C, RU 2309436C2 — but no US infringement suit or PTAB challenge appears in the record I retrieved.

Strategic summary

Claim status: nothing is canceled, and nothing is adjudicated — every claim is UNTESTED. Claims 1–31 stand unamended and unchallenged before the Board. That cuts both ways: you cannot point to a PTAB cancellation to kill an infringement theory, but the patent owner equally cannot point to a PTAB survival to deter a validity challenge. The patent's true defensive posture is driven not by PTAB history but by its prosecution and its age. Note two claim-drafting artifacts worth exploiting: claim 11 depends from claim 6 yet purports to normalise a sequence that claim 6 already normalises (an apparent double-normalisation / antecedent-basis problem), and the claim set contains an internal dependency oddity in the claim 24 → claim 31 chain, where claim 31 in turn depends from claim 7. Those are § 112(b) arguments for a district court, or a CBM-style indefiniteness theory — but they are untested by any tribunal.

Estoppel landscape: there is none, and that is the single biggest point in a defendant's favour. Section 315(e)(2) estoppel attaches only to a petitioner, its RPI and privies, and only after an FWD. No IPR was ever filed, no FWD ever issued, so no party is estopped from anything. Unlike the Bittleston patents — where ION's IPR participation and PGS's FWDs generated estoppel and issue preclusion that shaped subsequent cases — a defendant here may raise any § 102 or § 103 ground in district court, on any art including patents printed publications and also prior public use, on-sale, and § 112 grounds that an IPR could never reach. Practically, the cleanest route is a district court § 282 invalidity case plus an IPR filed in parallel on the best patents-and-printed-publications art; the IPR is not barred by anything, though the Board's practice on expired patents (no amendment possibility, so no need to institute) means it may decline — which pushes weight back to the district court.

Pattern signals: none of the usual ones are present. There is no repeat serial petitioner against this patent, no defensive aggregator (no Unified Patents or RPX filing identified), and no patent-owner PTAB appeal activity. The absence is itself informative: 7,003,402 is not a troll's reassertion asset and does not appear to have drawn the sustained § 102/§ 103 fire that commercially significant asserted patents attract. Two structural reasons explain the quiet: (i) the patent is a method-of-processing claim, which is harder to map to an accused product than an apparatus claim, making it a poor vehicle for a quick IPR kill; and (ii) it is expired as of 2022-12-22, which removes the going-forward injunction and reasonable-royalty leverage that drives IPR spending.

Recommended next steps

1. Lead with expiry. Before anything else, confirm the terminal disclaimer / maintenance-fee history and the adjusted expiration. Google Patents lists "Expired – Fee Related, expires 2022-12-22" (source). If accurate, there is no prospective infringement; any claim is limited to pre-2022-12-22 conduct and further narrowed by the 35 U.S.C. § 286 six-year damages lookback. Put the demand-letter sender to proof that (a) the patent is unexpired or (b) any asserted acts fall in a compensable window. This is usually dispositive leverage and costs almost nothing to establish.
2. Run the definitive PTAB docket check. Verify in PTAB E2E and Docket Alarm by patent number 7003402 (not "7,003,402" as free text, which is what pulled the refrigerator part number). I found no AIA trial activity, but the ODP feed is a moving target and I could not complete a full docket-level scrape within my search budget. Confirm before representing to a court or an adversary.
3. Because no FWD exists, there is no disposition to quote and no estoppel to assert. Do not cite the Bittleston FWDs (WesternGeco v. ION, 889 F.3d 1338) as if they addressed this patent — they did not. They are useful only as background on the patent owner's litigation posture and the family's ownership chain.
4. If an IPR is contemplated, check the legal-availability gate first. PGR is unavailable (the application was filed 2002-06-17, well before the 2013-03-16 first-to-file cutoff). CBM is unavailable — this is a geophysical data-processing claim, not a "financial product or service," and the CBM program sunset on 2020-09-16. That leaves IPR as the only AIA vehicle, and on an expired patent the Board may decline to institute. Budget accordingly and keep the district court case as the primary venue for invalidity.
5. Confirm you have the right patent. If the demand letter cites "the '402 patent," make the sender identify the number precisely. The collision with US 6,725,402 and US 11,597,402 is a real risk of confusion in correspondence, and the two are in entirely different technological and procedural postures.

I'll verify the assignment record against live sources before drafting. Let me run several searches in parallel.

Let me probe the assignment record specifics and any assertion activity.

Assignment & Ownership Analysis — US 7,003,402 B2

Patent: US 7,003,402 B2 — "Method of processing data"
Application: US 10/479,098 (35 U.S.C. §371 nationalization of PCT/GB2002/002774)
Priority: GB 0114744.6, filed 2001-06-16
Filed: 2002-06-17 · Granted: 2006-02-21 · Status: Expired – Fee Related (adjusted expiration 2022-12-22)

Source-access note (applies throughout): I was able to retrieve Google Patents legal events, the Justia full-text record, and the federal-court/PTAB record for the WesternGeco patent family. I was not able to retrieve the USPTO Assignment Center detail pages (assignmentcenter.uspto.gov / assignment.uspto.gov are JavaScript-rendered and did not return record-level data to automated retrieval). Consequently reel/frame numbers, execution dates, and the correspondent of record could not be read directly and are reported below as not retrievable rather than reconstructed. Nothing below is inferred or back-filled.

---

Inventors

Unusual patterns: None observed.

• Both inventors assigned jointly and directly to WesternGeco L.L.C. (the US entity) rather than to a UK operating entity — a consolidation pattern, not a departure pattern.
• No evidence retrieved of either inventor leaving WesternGeco/Schlumberger within 12 months of filing, and no evidence of an inventor-initiated reversion, assignment-back, or employment-agreement dispute.
• Note the contrast with the sibling WesternGeco patents litigated in 2009–2018 (the "Bittleston" and "Zajac" patents, e.g. US 6,691,038; 6,932,017; 7,080,607; 7,162,967; 7,293,520), where inventorship was actively contested by ION Geophysical. No inventorship challenge to US 7,003,402 appears anywhere in the retrieved record.

---

Original assignee

WesternGeco L.L.C. — the entity named as assignee on the granted patent and the only assignee ever recorded.

• Line of business: Marine seismic acquisition services and seismic data processing — specifically the Q-Marine / Q-Technology single-sensor towed-streamer system. This is not incidental to the patent: the specification itself states that a suitable sensor is "that used in the Q-Marine acquisition system produced by Schlumberger," and the claims (measure-of-spread / normalized spread over an ensemble of single-sensor traces) are practiced by the Q-Technology processing workflow.
• History: Formed in 2000 as a joint venture of Schlumberger (Geco-Prakla) and Baker Hughes (Western Geophysical). Baker Hughes subsequently exited the venture (I have not verified the exit date in the retrieved sources). WesternGeco has operated since as a Schlumberger group company; the WesternGeco brand and IP sit within what is now SLB.
• Current status: Operating, not dissolved, not in bankruptcy, not acquired by any NPE. Google Patents lists the current assignee as WesternGeco LLC. The patent is expired, but the assignee is a going concern and a practicing entity.

Important "shipped a product" finding: WesternGeco launched the commercial Q-Marine steerable streamer system in 2000 (Federal Circuit record, WesternGeco L.L.C. v. ION Geophysical Corp., 837 F.3d 1358 and the 2018 successor appeal). The patented subject matter here is Q-Technology single-sensor acquisition/processing, i.e. an in-product technology of an operating company, not a licensing asset.

---

Assignment timeline

Recorded entries found: one (1).

• Recorded 2003-11-24 — Reel/frame not retrievable (not exposed in the sources I could access; see source-access note)
  • Conveyance: Assignment of Assignors' Interest (the USPTO record text reads "ASSIGNMENT OF ASSIGNORS' INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: CURTIS, ANTHONY D.; CHRISTIE, PHILIP
  • Assignee: WESTERNGECO L.L.C.
  • Execution date: Not exposed in the retrieved record.
  • Correspondent: Not retrievable. I could not read the attorney/firm of record from this recording, so I make no claim about a repeat correspondent here (see signal 3 below).
  • Context: Original employment/filing assignment — inventors to their employer/venture, recorded contemporaneously with entry of the PCT into the US national phase (US 10/479,098, the "479" series is a 2003 national-phase entry). This is an ordinary in-bound upstream assignment, not an acquisition, fire-sale, reorg, securitization, or transfer-to-asserter.

There is no second recorded assignment. No post-issuance assignment, no security agreement, no merger, no change of name, and no license record appears in the Google Patents legal-events feed, which mirrors the USPTO assignment index. The evidence therefore indicates WesternGeco L.L.C. still owns the patent — as the "Current Assignee" field confirms.

Secondary cross-check (independent corroboration of no NPE chain): I reviewed the "Cited By" and "Families Citing this Family" lists for US 7,003,402. The citing parties are operating companies and academic/industrial filers — inter alios BMC Software, ION Geophysical, Input/Output Inc., Apache Corporation, Industrial Scientific, ConocoPhillips, Invensys, DH Technologies, Fugro, Geco Technology BV. No entity on the cited-by lists matches an NPE/asserter directory (Acacia, Marathon, Intellectual Ventures, Wi-LAN/Conversant, Pendrell, Vringo, Round Rock, MPHJ, etc.), and no known asserter has acquired this patent.

Litigation cross-check: I found no lawsuit naming US 7,003,402. The two WesternGeco assertion campaigns in the record — WesternGeco LLC v. ION Geophysical Corp., No. 4:09-cv-01827 (S.D. Tex., filed 2009-06-12) and the follow-on action against Petroleum Geo-Services — asserted US 6,691,038; 6,932,017; 7,080,607; 7,162,967; 7,293,520, and not the '402 patent. Those campaigns were brought by WesternGeco in its own name as an operating company against direct competitors (ION, PGS/Fugro), and the resulting IPR/§271(f) proceedings (PGS I–VI; WesternGeco v. ION, 138 S. Ct. 2129 (2018)) did not involve the '402 patent.

---

Timeline diagram

timeline
    title Ownership of US 7003402
    2001 : GB priority application filed
    2002 : PCT application filed June 2002
    2003 : US national phase entered
         : Inventors assign to WesternGeco LLC
    2006 : US patent granted Feb 2006
    2009 : WesternGeco sues ION on other patents
    2022 : Patent expires fee related

---

NPE / troll-pattern signals

One genuine, non-NPE anomaly worth recording: the patent is tagged "Expired – Fee Related" with an adjusted expiration of 2022-12-22. A well-capitalised operating company allowed a Q-Technology patent family member to lapse for non-payment rather than run full term or monetise it. That is consistent with a large portfolio being pruned on commercial relevance — and it is the opposite of asserter behaviour (an NPE does not let a patent it intends to assert lapse).

---

Verdict

Insufficient data (no records, or only the original assignment).

The entire recorded ownership history of US 7,003,402 is a single assignment, recorded 2003-11-24, from inventors Anthony D. Curtis and Philip Christie to WesternGeco L.L.C. — the original employment/filing assignment made on entry into the US national phase. There are no post-issuance transfers of any kind, and all eight NPE signals return "not present"; the patent has never been asserted in any lawsuit I could locate, was not among the WesternGeco patents asserted against ION or PGS, and expired fee-related on 2022-12-22 while still held by a practicing operating company. Affirmatively, the evidence points to an operating-company-held, never-asserted, now-lapsed patent rather than an NPE asset — but under the rubric the chain state is "only the original assignment," and I cannot populate reel/frame or correspondent, so I do not upgrade this to an NPE call in either direction.

Verification links:

• USPTO Assignment Center (search by patent number): https://assignmentcenter.uspto.gov/
• USPTO legacy Patent Assignment Search: https://assignment.uspto.gov/patent/index.html
• Google Patents legal events for this patent: https://patents.google.com/patent/[US7003402B2](/patent/US7003402B2)/en (see "Reassignments / Legal Events" — single 2003-11-24 entry)
• Patent-family record: WO2002103397A1 · EP1399764A1 · GB2376527B · CN100487490C · AU2002302834B2 · RU2309436C2 · NO20035615D0

I'll research US patent 7003402 thoroughly, starting with searches to confirm the patent details and locate its cited prior art.

Let me look for the specific prior-art references cited in this patent, including the patent references mentioned in the description.

Let me try to retrieve the patent's front-page "References Cited" section directly from the patent PDF and USPTO records.

Prior-Art Analysis — US Patent 7,003,402

1. Confirmation of the patent (literal identifiers)

The number 7003402 resolves unambiguously to:

Subject matter: a method of determining, from two data sequences acquired by two spatially separated sensors sampling substantially the same signal, a sequence indicative of the "measure of spread" (e.g., standard deviation/variance, optionally normalised by the common signal/mean), thereby giving an estimate of the initial uncertainty (i.e., the noise) in the signal. Applied to single-sensor seismic (SSS) data. This is claim 1; the span of dependent claims 2–31 builds on it (common signal, normalisation, ensembles, standard deviation/variance, outputting, pre-processing, QC feedback, seismic/SSS limitation, apparatus, storage medium).

2. Important sourcing caveat — please read before relying on this

The authoritative full text supplied (Google Patents, https://patents.google.com/patent/US7003402/en) and my searches do not contain a backward "Patent Citations / References Cited" table for this patent. What the record exposes is:

• "Cited By (15)" — forward citations (later documents citing US 7,003,402), and
• "Families Citing this family (28)" — later family citations.

These are not prior art and are legally unusable for §102 against US 7,003,402, because every one of them postdates the 2001-06-16 priority date. I am therefore not going to present the "Cited By" list as if it were the patent's ante-dating prior art — that would misrepresent the record.

What I can ground with confidence are the references cited inside the specification itself (the description expressly names patent and non-patent documents). I treat those as the operative "citations for 7003402" and analyze them below. Where I could not retrieve a full front-page bibliographic record, I flag it.

⚠️ If the intent is to see examiner-cited U.S. patent documents on the face of the patent, that list must be pulled from the USPTO PatentCenter/Patent Full-Text file wrapper for application 10/479,098 or the granted-print PDF. My searches did not return that table, and I will not fabricate one.

3. Prior-art references cited in the specification of US 7,003,402

These are the documents the patent itself cites (as background and as enabling techniques). All are pre-2001 and therefore facially qualify as prior art.

A. Non-patent literature — Single-Sensor Seismic (SSS) "Q system" papers (2000)

1. J. Martin et al., "Acquisition of marine point receiver seismic data with a towed streamer," SEG 60th Annual International Meeting, Expanded Abstract ACQ 3.3, Calgary (2000).

• Date: 2000 (≈1 year before priority) → §102(a)/(b)-eligible art.
• Description: Describes Schlumberger's Q-Marine point-receiver (single-sensor) towed-streamer acquisition.

1. G. Baeten et al., "Acquisition and processing of point receiver measurements in land seismic," SEG 60th Annual Meeting, Expanded Abstract ACQ 3.4, Calgary (2000).

• Date: 2000 → prior art.

1. G. Baeten et al., "Acquisition and processing of point source measurements in land seismic," SEG 60th Annual Meeting, Expanded Abstract ACQ 3.5, Calgary (2000).

• Date: 2000 → prior art.
• Combined description (1–3): Establish that single-sensor acquisition (individual sensor outputs individually digitised and accessible) was known. The patent uses these as the enabling platform for its method.

B. Patent references cited as known noise-attenuation techniques

1. UK Patent No. 2 309 082 (GB 2 309 082 B) — "Noise filtering method," applicant Geco AS.

• Dates returned: filing/priority 1996-01-09; publication/grant 1999-12-01 (GB2309082B).
• Description: Adaptive noise-filtering of seismic data (the "noise filtering method" used for attenuating coherent/environmental noise). Cited by 7,003,402 as an example of a usable pre-processing/adaptive noise-attenuation technique.

1. U.S. Patent No. 5,971,095 (Geco AS).

• Date: granted 1999 (pre-2001) → prior art.
• Description (from the US5971095 disclosure): A method for filtering noise from discrete noisy seismic signals using M temporally local filters forming a filter bank; a reference channel is formed as an estimate of the noise and the filter bank is derived by minimising a cost function. It expressly contemplates application to single-sensor recordings (i.e., recordings prior to group-forming).
• Caveat: my search retrieved the document body but not a clean front-page title line; the substance above is taken from the retrieved text. Treat the exact title as unverified.

1. UK Patent No. GB 2 337 591 (GB 2337591 B) — "Adaptive seismic noise and interference attenuation method," applicant Geco AS.

• Dates returned: priority 1998-05-20; publication/grant 2000-07-12.
• Description: Adaptive attenuation of seismic noise and interference; cited as a usable pre-processing technique.

C. General prior-art acknowledgment in the description

1. The specification states: "It has been proposed to use measurements taken at adjacent sensors to estimate spatial derivatives of a wavefront." This is an admission of a known, related prior-art approach (spatial-derivative estimation across adjacent sensors). It is unnamed, so no citation can be independently verified; I flag it as an acknowledged, but unidentified, reference.

4. §102 anticipation analysis

Governing point: claim 1 requires determining a sequence indicative of a measure of spread of two data sequences from two spatially separated sensors, to estimate the initial uncertainty. A §102 reference must disclose every element; a reference that merely enables or contextualises does not anticipate.

Conclusion on anticipation: None of the references cited within US 7,003,402 discloses the measure-of-spread limitation of claim 1. Accordingly, on the record available, none of them anticipates claim 1 or its dependents under 35 U.S.C. § 102. Their relevance is at most (a) as background establishing the single-sensor-acquisition platform (Martin/Baeten) and (b) as optional, known pre-processing techniques recited in dependent claims 13 and 22–23 (the three Geco references) — i.e., potential §103 combination material, not §102 art.

5. "Most relevant prior art" — bottom line

• For the patent's primary (claim 1) subject matter: the closest art is the unnamed "adjacent-sensor spatial-derivative" proposal acknowledged in the specification, because it is the only cited concept operating on data from spatially separated sensors sampling the same wavefront. The patent distinguishes itself from it on the ground that a derivative incorporates sensor separation whereas the claimed spread measure does not — which is precisely why it does not anticipate.
• For the enabling-technology subject matter: the three 2000 SEG Expanded Abstracts (Martin ACQ 3.3; Baeten ACQ 3.4; ACQ 3.5) are the most relevant, since they establish single-sensor seismic acquisition — the necessary precondition for the claimed method.
• For the dependent pre-processing claims (13, 22–23): the most relevant citations are GB 2 309 082, US 5,971,095, and GB 2 337 591 (all Geco), which the patent itself names as usable adaptive noise/interference-attenuation techniques.

6. Record items that are not prior art (for completeness, clearly labelled)

The Google Patents page's "Cited By (15)" list — e.g., US 2004/0215396-type family siblings excluded; the actual forward citers are US 2006/0123104 A1 and US 2006/0136585 A1 (BMC Software), US 2006/0291329 A1 (James Martin), US 2007/0056795 A1 (David Hall), US 2008/0232195 A1 (Input/Output Inc.), US 2008/0270061 A1 (IBM), US 2009/0147621 A1 (Stian Hegna), US 2009/0154291 A1 (Ralf Ferber), US 2008/0270061-adjacent BMC filings, US 2010/0161577 A1 and US 2011/0238637 A1 and US 2011/0163222 A1 (BMC/Moser), US 2012/0075955 A1 (Timothy Dean), and granted US 8,203,906 B2 (WesternGeco) — are all post-2001 and therefore cannot be §102 prior art against US 7,003,402. I list them only to distinguish them from the prior art you asked for.

7. Recommended next step to harden this analysis

Because the front-page "References Cited" (examiner/applicant-cited U.S. and foreign patent documents) for application 10/479,098 was not present in the retrieved record, I cannot certify that no examiner-cited U.S. patents exist. To obtain the definitive §102 citation set, retrieve the USPTO PatentCenter file wrapper / granted-print front page for 10/479,098, or the corresponding EP 1,399,764 A1 search-report ("Documents Considered to be Relevant"), which would list any category X/Y art the examiner relied on. I have deliberately not substituted guessed patent numbers for that missing table.

Verification links used: https://patents.google.com/patent/US7003402/en; WO 2010/065778 A3 (WesternGeco context); GB2309082B / GB2337591B listings on the EP 0681193 record; US 5,971,095 PDF text (patentimages…/US5971095.pdf).

I'll ground this analysis by checking the prior-art references identified on the patent page (both the specification-cited references and the citation listings).

Let me check the prosecution-side prior art for the corresponding PCT/EP filings, which reveals what art examiners actually applied.

Obviousness Analysis — US7003402B2 (“Method of processing data”)

0. Threshold note on what the page actually supplies as prior art

Before the analysis, one necessary correction of the record, because it materially changes the §103 analysis:

• The page contains a “Prior art keywords” field (data, spread, sequence, measure, data sequence). That is Google Patents metadata, not a prior-art reference list.
• The “Cited By (15)” and “Families Citing this family (28)” tables are forward citations — documents that cite US7003402B2. Every entry there (e.g. US20080232195A1, US20090154291A1, US20090147621A1, US8203906B2, US7865322B2, US7539593B2, CN100347707C, CN100349012C) has a priority or publication date after the 2001‑06‑16 priority date and therefore cannot be §102/§103 prior art against this patent. They may be cited only as evidence of the level of ordinary skill as of a later date, if at all.
• The page does not reproduce a “References Cited / Patent Citations” section for US7003402B2.

Accordingly, the operative prior art must be drawn from (i) the references the specification itself identifies, and (ii) the applicant’s own admissions in the Background of the Invention. Both are legitimate and citable.

Admitted prior art (from US7003402B2’s own Background)

1. Conventional hard-wired group forming — sensors hard-wired into groups before digitisation; "the conventional process of merging the individual outputs of the sensors in each group is generally an ensemble averaging process, that yields a weighted average of the individual sensor outputs representing the estimate of the signal common to all the sensors in the group." This is a binding admission that computing a mean across an ensemble of spatially adjacent sensors was routine.
2. "It has been proposed to use measurements taken at adjacent sensors to estimate spatial derivatives of a wavefront." — admission of adjacent-sensor wavefront-difference art.
3. "Although generating a measure of the spread of a signal is known, it has not hitherto been possible to apply this to seismic data." — the applicant concedes the statistical operation is old; the alleged contribution is confined to its application to (then-new) point-receiver seismic data.

Specification-cited prior art (all published before the 2001‑06‑16 priority date)

Critical timing fact for §103: the enabling single-sensor acquisition platform was published on 1999‑11‑24 (GB2337592A) and described in the open SEG literature in August 2000 — i.e. roughly 18 months to two years before the 2001‑06‑16 priority date. The specification’s statement that "until the advent of the Schlumberger Q system, seismic acquisition systems did not have the acquisition bandwidth to make it possible" is therefore not supported as a date-based argument: the enabling art was public before the priority date.

---

1. Claim 1 — the broadest claim

Claim 1 requires only: determining, from a first data sequence at a first sensor and a second data sequence at a spatially separated second sensor sampling substantially the same signal, a sequence indicative of a measure of spread of the two sequences, thereby to provide an estimate of the initial uncertainty.

No sensor spacing is positive-recited beyond "spatially separated"; no particular spread metric; no normalisation; no grouping; no processing/control step.

Primary combination that renders claim 1 obvious

GB 2 337 592 (US 6,684,160) + GB 2 309 082 (US 5,971,095) + admitted conventional ensemble mean

• GB 2 337 592 supplies every structural/environmental element: two (indeed dozens of) hydrophones, spatially separated, sampling substantially the same up‑coming seismic wavefront, each digitised separately and individually accessible before group forming, with a ≤500 cm spacing chosen precisely because adjacent channels carry the same signal.
• GB 2 309 082 supplies the operational element and the express motive: it already processes "recordings prior to any group forming", i.e. it works on the individual traces, and it already computes an error measure across channels (the cost function J(t)) as the quantitative basis for its processing.
• The admitted conventional group-forming art supplies the mean across adjacent sensors ("estimate of the signal common to all the sensors in the group").

Why the combination is motivated (KSR factors):

1. Same field, same problem, same data. Both references are seismic noise-attenuation documents addressing the same streamer-borne and environmental noise; GB 2 309 082 is explicitly directed at pre‑group-forming data, which is exactly the data set GB 2 337 592 creates. This is the paradigm "combination of prior art elements according to known methods to yield predictable results."
2. The second quantity is not an independent invention. Having already computed the mean across an ensemble of co-located sensors (admitted art), computing the standard deviation / variance / max‑minus‑min of the same ensemble is the single most routine companion statistic in the field. If a POSITA can compute an average, the marginal step to a dispersion measure of the same sample set is arithmetic, not inventive (MPEP 2144.04; KSR v. Teleflex, 550 U.S. 398 (2007)).
3. Express motivation in the art. GB 2 309 082 and GB 2 337 591 both need a quantitative measure of how well the filter worked — the entire point of the cost-function/constrained-optimisation machinery. A residual/spread statistic is the natural, and the only cheap, way to measure that on data where no independent noise reference exists (GB 2 309 082 itself notes "a reference signal free of signal cannot be easily obtained in seismic data acquisition," so it generates reference channels).
4. Predictable result. Mean = common signal; dispersion about that mean = local noise/uncertainty. The specification itself adopts exactly this reasoning ("differences between data acquired at the two sensors are attributable primarily to local noise"), i.e. it states the result as an inference from geometry, not as a discovery.

Secondary/supporting combination

Martin et al. (2000) + GB 2 337 592 + conventional statistics. Martin et al. already disclose the Q‑Marine point-receiver streamer with 3.125 m spacing and post‑recording "dynamic group forming." A POSITA implementing dynamic (i.e. software, re‑definable) group forming would, as an obvious matter of programming, compute the spread of the samples entering each group — because the number of samples per group is now a processing parameter rather than a manufacturing fact, so the operator needs a statistic to choose it. That is a design incentive to compute the dispersion, not merely a design choice to do so.

Anticipating-readings argument (worth flagging even though the chart is §103)

The specification describes the measure of spread as including "a simple difference between maximum and minimum amplitudes within the ensemble." For a two‑sensor ensemble, max − min is the absolute channel difference |S₁ − S₂|. GB 2 337 592 expressly filters each hydrophone output "using as further input the digitized output of at least one nearby hydrophone," which in a two‑channel noise‑cancellation scheme necessarily forms a difference/estimate of the local noise. If the Board/EDPO were to construe "measure of spread" to read on |S₁ − S₂|, claim 1 is arguably anticipated under §102(b). Confidence: moderate — the abstract of GB 2 337 592 (as retrieved) does not literally recite "difference," so this is inference from the filtering step, not an express disclosure.

---

2. Claims 2–6 — mean, standard deviation, normalisation

• Claim 2 (determine common signal) — the arithmetic mean over an ensemble: admitted prior art verbatim ("generally an ensemble averaging process, that yields a weighted average").
• Claims 3, 4 (element-wise commonality / spread) — the element-wise mean and element-wise dispersion of corresponding time samples: pure arithmetic implementation of claims 1–2. Obvious.
• Claims 5, 6 (spread normalised by |common signal|) — literally the coefficient of variation, σ/|μ|, a textbook statistic predating 1900. This is the strongest §103 claim set in the patent.

Two independent lines of motivation for normalising:

1. Within the field: seismic traces are non‑stationary (the specification says so: "Because seismic traces are not time-stationary sequences, it is desirable to normalise the measure of spread"). Normalising a signal‑dependent noise measure by an amplitude estimate is the standard practice in the industry’s "relative noise" / noise‑to‑signal‑ratio QC metrics, and is what GB 2 309 082 and GB 2 337 591 effectively do when they scale/weight by input power ("white-noise gain … weighted by the input power").
2. Outside and analogous fields: the ISO Guide to the Expression of Uncertainty in Measurement (GUM, first published 1993) defines "Type A" evaluation of uncertainty as the statistical analysis of a series of observations, with standard uncertainty expressed from the experimental standard deviation, normalised by the mean where a relative uncertainty is required. Applying a metrological standard uncertainty construct to repeated, spatially over‑sampled seismic observations is an analogous use in a field of endeavour that the patent itself concedes reaches beyond seismics ("acoustic, ultrasonic or electromagnetic data"). Precedent analogy in a different field, for a known problem, is a KSR-sanctioned route to obviousness. Confidence in the GUM identification: high on substance, moderate on the precise edition year (1993 first edition, 1995 corrected reprint).

---

3. Claims 7–14 — selection of ensembles and notional grouping

• Claim 7 (select two or more sequences from a plurality, determine their spread) — selecting a subset of adjacent channels is precisely what conventional hard‑wired grouping did mechanically, and what GB 2 337 592 does computationally ("filtering … uses as further input the digitized output of at least one nearby hydrophone"). Obvious.
• Claims 8–10, 13 (common signal of selected sequences; grouping into ensembles) — the admitted ensemble averaging step, performed in software rather than hard wire. The specification itself concedes equivalence: "this set of average sequences (with appropriate weighting) is equivalent to the data output from a conventional, analogue-grouped system."
• Claims 11, 12 (normalisation of the selected-ensemble spread) — same as §2 above, applied to the selected ensemble.
• Claim 14 (repeat for another ensemble) — looping over a spatial array. Routine.
• Notional rather than fixed grouping — this is presented as the invention’s novelty, but it is the necessary and immediate consequence of the prior art hardware. Once GB 2 337 592/Martin et al. remove the hard wiring and retain individual traces, "grouping" can only be a processing construct. A limitation that is the inevitable by‑product of known prior art practice does not confer patentability. No reference teaches away from re‑defining groups; the absence of a capability is not a teaching away (In re St. Jude line: scepticism/criticism is required, and none is here).

---

4. Claims 15–17 — outputting the sequences

Outputting σⱼ and the mean as traces/visual display, or storing them, is the routine output step of any seismic QC tool. GB 2 309 082 already produces filtered output for storage/redisplay; seismic processing systems of the era routinely displayed QC panels (FIG. 2 of the patent itself shows traces as a merely illustrative convenience). Obvious; a display/storage limitation adds no patentable weight where the underlying computation is known.

---

5. Claims 18–19 — standard deviation and variance

Expressly the two most common dispersion measures; §2 above. Claim 19’s stated rationale in the specification ("ease of use of the variance in multi‑trace combinations") is itself an optimisation/design-choice justification, i.e. the classic §103 "obvious to try among a finite number of identified, predictable solutions" situation.

---

6. Claims 20–21 — processing the common signal; controlling that processing on the basis of the spread

GB 2 309 082 + GB 2 337 591 + the patent’s own cited art render this obvious:

• Both references are adaptive processors whose coefficients are adjusted by minimising a computed error/objective function.
• Weighting data by a noise/uncertainty estimate is ordinary practice in least‑squares seismic inversion and stacking; the specification concedes it: "Often these are guessed or, alternatively, simply omitted, treating all data as of equal reliability."
• Motivation: substituting a measured weight for a guessed one is the definition of an obvious improvement with a predictable benefit. There is no new mechanism — no new inversion, no new optimiser.
• Claim 21 (feed the spread back to control the processing) is expressly the feedback loop already present in the adaptive-filter art (GB 2 309 082: "the filtered output signal may be reprocessed in an iterative manner … fed back to at least one of the reference signal and filtering means"). That is a feedback/control loop, and the spread statistic is an obvious candidate control signal.

---

7. Claims 22–24 and 31 — pre-processing and repeated pre-processing

• Claims 22, 31 (pre‑process the data before computing the spread) — the specification itself points to GB 2 309 082, US 5,971,095 and GB 2 337 591 as the pre‑processing: "Noise attenuation and common signal estimation may be performed by … adaptive noise and interference attenuation techniques such as the techniques described in UK patent No 2 309 082, U.S. Pat. No. 5,971,095, or UK patent GB 2 337 591." Where the patentee expressly identifies the prior art as performing a step, the combination is at least strongly prima facie obvious.
• Claim 23 (vary pre‑processing parameters on the basis of the computed spread, and repeat) — an iterative parameter search minimising a measured criterion. Adaptive filter optimisation is exactly this (GB 2 337 591; GB 2 309 082). The only change is the choice of criterion; choosing a physically meaningful, cheaply computed error criterion is routine.
• Claim 24 — note: it depends on claim 31, which depends on claim 7. I record this dependency chain literally and do not correct it. Its scope is therefore tied to the pre‑processing + ensemble‑selection combination, which is obvious for the reasons in §§3 and 7 taken together.

---

8. Claims 25–27 — seismic data and single-sensor acquisition

Obvious over GB 2 337 592 / US 6,684,160 and Martin et al. (2000), both of which disclose the point‑receiver acquisition itself, before the priority date. Claim 27 ("acquiring each seismic data sequence using single sensor seismic data acquisition") recites the already-public Q‑Marine architecture (3.125 m spacing, separately digitised hydrophones) — the patent itself describes this architecture as known art in its Background. A claim whose only additional element is admitted prior art cannot be saved by the earlier claims’ spread computation.

---

9. Claims 28–30 — apparatus and storage medium

• Claim 28 is the apparatus counterpart of claim 1 ("means for determining …"), obvious on the identical rationale.
• Claim 29 (programmable data processor) and claim 30 (storage medium containing a program) are routine implementations. GB 2 309 082, GB 2 337 591 and GB 2 337 592 all contemplate digital processing of digitised traces; programming a general‑purpose processor to compute a mean and standard deviation is not inventive. Storage-media claims of this type add nothing beyond the method they encode.

---

10. The patentee’s best rebuttals, and why they are weak

---

11. Bottom line

Overall: On the prior art available on this page, US7003402B2 presents a strong prima facie case of obviousness across all 31 claims, driven by one central proposition: the patent claims a standard statistical descriptor (dispersion of a set of nominally co-located measurements) applied to a data set that the same corporate family had already made publicly available 18–22 months before the priority date, to obtain the predictable result of a noise/uncertainty indicator. The patent’s own specification supplies the two admissions that carry most of the weight — that the spread measure as such is known, and that ensemble averaging over adjacent sensors was already the industry norm.

Caveats I want stated explicitly:

1. The page as provided does not contain US7003402B2’s own examiner‑cited reference list; I could not retrieve the EPO/UK search report for WO2002103397A1 / EP1399764A1 / GB2376527B from the sources reached. A definitive invalidity position should be re‑run against the actual ISR/EPO search‑report citations, which are not in evidence here.
2. My retrieval of GB 2 237... / GB 2 337 591 / GB 2 337 592 was via family records (EP0873528, EP1084426, WO1999060421, WO1999060423), not from the face of a US7003402B2 "References Cited" table. Publication dates used (1997‑07‑16, 1999‑11‑24, 1999‑12‑01, 2000‑05‑03, 2000‑07‑12) are taken from those records and should be verified against the registers before any filing is made.
3. Where I have characterised GB 2 337 592 as "forming a difference," that is my inference from its filter‑using‑nearby‑hydrophone language, not a quotation. I have flagged it as such rather than presenting it as express disclosure.
4. Forward‑citation documents in the page’s "Cited By"/"Families Citing" tables are not prior art to this patent and I have not used them as such.