US 8605794

Summary: U.S. Patent No. 8,605,794 (US8605794B2)

Verification note: I searched for the exact number "8605794" (the '794 patent). I deliberately excluded results for similar patent numbers in the same VideoLabs/Netflix/Unified Patents litigation cluster (e.g., 7,233,790; 7,440,559; 7,235,380; 8,291,236; 8,667,304), which are different patents. Sources: the full Google Patents text you supplied (authoritative for bibliographic data and claims), USPTO PTACTS documents, and CAFC/Justia docket records returned by search.

Bibliographic data

• Title: Method for synchronizing content-dependent data segments of files
• Patent/Application: US8605794B2; U.S. Application No. 11/918,276 (national phase of PCT/EP2006/060808)
• Inventors: Hermann Hellwagner, Jörg Heuer, Andreas Hutter, Michael Ransburg
• Original Assignee: Siemens AG (assignment recorded Jan. 21, 2009)
• Current Record Owner: VL Collective IP LLC (via chain: Siemens AG → Lough Corrib Intellectual Property Limited (2018) → VideoLabs, Inc. (2019) → VL IP Holdings LLC (2019) → VL Collective IP LLC (2019)). Google Patents lists the current assignee as VL Collective IP LLC / VL IP Holdings LLC.
• Filing date: March 16, 2006 (PCT); priority to German application DE 10 2005 016 866.3 filed April 12, 2005
• Issue date: December 10, 2013
• Status: Active; adjusted expiration November 2, 2028 (maintenance fees paid through 12th year, June 2025)

Abstract

"The synchronization of data segments is particularly necessary for the use of small screens. A method and a device are disclosed for synchronizing content-dependent first data segments of a first data file and content-dependent second data segments of a second data file, the first and second data segments being respectively output in sequence according to their chronological order in such a way that based on a predeterminable assignment rule for the first and second data segments, each second data segment is output together with a corresponding first data segment."

Plain-language overview of the independent claims (claims 1, 9, and 22 of 22)

Claim 1 (method): A method of synchronizing content-related segments from two files. First data segments from a first file and second data segments from a second file are output sequentially, in their chronological order, such that every second data segment is output together with an associated first data segment, where the pairing is governed by an assignment rule that assigns each second data segment to one first data segment. (In essence: pair up and co-output content-meaningful chunks of two files based on a predefined mapping, not on timestamps.)

Claim 9 (device): A synchronization device (apparatus claim) configured to do the same thing — sequentially output first and second content-related data segments in chronological order, with each second segment output together with an associated first segment based on an assignment rule assigning each second segment to a first segment. (Apparatus counterpart of claim 1.)

Claim 22 (method, marker-based): A method that (1) determines read-out orders for the first and second data segments using first and second content-related markers, (2) assigns each ordered second data segment to a corresponding ordered first data segment per an assignment rule, and (3) sequentially outputs the assigned pairs in chronological sequence. (Adds the content-marker-based ordering/random-access feature described in the specification, e.g., scene-change or speaker-change markers.)

Dependent claims 2–8, 10–21 add details such as: assignment after a fixed number of first segments (claims 2, 10), position-number-based assignment (claims 3, 11), multiple second files with a combined assignment rule (claims 4, 10, 11), media data (video/audio) (claims 5, 12, 15), gBSD/XSLT-based segment generation (claim 6), content-related markers (claims 7–8), XML metadata (claims 13–14), predefinable rule (claims 16–17), rule in a separate file (claims 18–19), and rule not based on a timestamp (claims 20–21).

CAFC / litigation status (2026 dockets)

• CAFC case 24-1890 — VL Collective IP, LLC v. Unified Patents, LLC: Appeal from IPR2022-01086 (PTAB Final Written Decision Dec. 18, 2023, holding claims 1, 3, 5, 9, 12, 13, and 15 unpatentable over Sonohara, U.S. 5,627,656; claims 20–21 were not shown unpatentable). Oral argument was held Jan. 9, 2026, and the Federal Circuit issued a nonprecedential opinion on Feb. 20, 2026 affirming the PTAB (rejecting VideoLabs' "assignment rule" and "content-related data segments" construction arguments; finding a timestamp-based argument forfeited). Sources: cafc.uscourts.gov opinion posting (02-20-2026-24-1890); vitallaw.com case report (Feb. 23, 2026); USPTO PTACTS notice of appeal.
• CAFC case 25-1525 (pending): Listed on the Google Patents litigation page as a CAFC case; per Patexia, it is the appeal of the Netflix IPR (IPR2023-00891), whose Jan. 7, 2025 Final Written Decision found all 22 claims unpatentable. As of the search date I found no CAFC decision in 25-1525 — treat its status as unconfirmed. (Do not confuse with CAFC 25-1132, a different VideoLabs v. Netflix appeal involving the '559 patent.)
• District court actions involving the '794 patent include VideoLabs v. Netflix (D. Del. 22-229, stayed), VideoLabs v. Roku (D. Del. 1:23-cv-01136), Starz Entertainment v. VL Collective (D. Del. 21-cv-1448), VideoLabs v. Hisense (E.D. Tex. 2:24-cv-00904); earlier cases against Apple (N.D. Cal. 3:23-cv-01307, dismissed with prejudice) and Amazon (W.D. Tex. 6:22-cv-00079, dismissed by stipulation).

Uncertainty caveats

• I relied on web search results (CAFC, USPTO PTACTS, Justia, Patexia, Vitallaw, PatSnap) rather than a direct PACER/USPTO database pull; docket-level details for 25-1525 could not be verified beyond the sources above.
• The PTAB's unpatentability findings are subject to appeal (24-1890 affirmed; 25-1525 pending), so the patent's enforceable claim set may change; the USPTO record still shows the patent as "Active."

Litigation Involving U.S. Patent No. 8,605,794 ("the '794 patent")

Below is a summary of the known litigation involving U.S. Patent No. 8,605,794 (owned by VL Collective IP LLC, successor-in-interest to Siemens AG), based on searches of court records, PTAB dockets, and the patent's litigation history. I have flagged where details could not be fully verified.

I. District Court Infringement Actions (VideoLabs/VL Collective as plaintiffs)

1. VideoLabs, Inc. et al. v. Netflix, Inc.

• Plaintiffs: VideoLabs, Inc.; VL Collective IP LLC
• Defendant: Netflix, Inc.
• Jurisdiction: D. Del. (Wilmington)
• Case No.: 1:22-cv-00229 (C.A. No. 22-229)
• Filing date: February 23, 2022 (original complaint). The '794 patent was added by First Amended Complaint filed May 6, 2022 (Count IV), along with the '878, '559, and '790 patents.
• Judges: Initially Colm F. Connolly; later Joel H. Slomsky
• Status: Netflix's §101 motion to dismiss the '794 count was denied (May 14, 2024 opinion). The case was stayed pending inter partes review. Netflix's IPR (IPR2023-00891) resulted in a Final Written Decision (Jan. 7, 2025) holding all 22 claims of the '794 patent unpatentable, which is the operative basis for the current stayed/effectively concluded posture of the district court case.

2. Starz Entertainment, LLC et al. v. VL Collective IP LLC et al.

• Plaintiff: Starz Entertainment, LLC et al.
• Defendants: VL Collective IP LLC et al. (VideoLabs entities)
• Jurisdiction: D. Del.
• Case No.: 1:21-cv-01448-JLH (Judge Jennifer L. Hall)
• Filing date: 2021 (exact filing date not verified in my searches)
• Status: A claim-construction (Markman) hearing and scheduling order (Dkt. 18, Feb. 10, 2022) are documented in IPR2022-01086 exhibits. This appears to be a declaratory-judgment-style action by Starz. Final disposition not confirmed in the records I retrieved; treat as pending/undetermined.

3. VideoLabs, Inc. et al. v. Roku, Inc.

• Plaintiffs: VideoLabs, Inc.; VL Collective IP LLC
• Defendant: Roku, Inc.
• Jurisdiction: D. Del.
• Case No.: 1:23-cv-01136-JHS (Judge Joel H. Slomsky)
• Filing date: October 11, 2023
• Status: Docket flags show "STAYED, CLOSED" — stayed (consistent with the VideoLabs/Netflix IPR pattern) and closed.

4. VideoLabs, Inc. et al. v. Hisense Co. Ltd. et al.

• Plaintiffs: VideoLabs, Inc.; VL Collective IP LLC
• Defendants: Hisense Co. Ltd.; Hisense Visual Technology Co. Ltd. (f/k/a Qingdao Hisense Electric Co. Ltd.); Hisense International Co. Ltd.; Hisense International (HK) Co. Ltd.; Hisense International (Hong Kong) America Investments Co. Ltd.
• Jurisdiction: E.D. Tex. (Marshall/Tyler)
• Case No.: 2:24-cv-00904-JRG-RSP (Judge Rodney Gilstrap; Magistrate Judge Roy S. Payne)
• Filing date: Early November 2024 (summonses returned executed Nov. 8–11, 2024)
• Status: Ongoing. A motion to stay pending settlement discussions was denied (Dec. 5, 2024); the case schedule was modified in March 2025.

5. VideoLabs, Inc. et al. v. Amazon.com, Inc. et al. (dismissed)

• Jurisdiction: W.D. Tex. (Austin), Judge Alan D. Albright
• Case Nos.: 6:22-cv-00079-ADA; 6:22-cv-01167-ADA
• Filing date: 2022
• Status: The '794 patent was previously asserted; joint stipulation of dismissal filed December 22, 2022. Dismissed.

6. VideoLabs, Inc. et al. v. Apple Inc. (dismissed)

• Jurisdiction: N.D. Cal.
• Case No.: 4:23-cv-01307-AMO (Judge Araceli Martínez-Olguín)
• Filing date: 2023
• Status: '794 previously asserted; joint stipulation of dismissal filed October 3, 2023. Dismissed.

Note: The Google Patents litigation list also shows a "California Eastern District Court" entry "3:23-cv-01307." The patent owner's own IPR mandatory notices list only the Northern District of California Apple case (23-cv-01307-AMO); the E.D. Cal. entry appears to be a data artifact of the same Apple case rather than a separate action. I found no independent evidence of an E.D. Cal. case.

II. Inter Partes Review (PTAB)

7. Unified Patents, LLC v. VL Collective IP LLC — IPR2022-01086

• Petitioner: Unified Patents, LLC
• Patent Owner: VL Collective IP LLC
• Filing date: Petition filed June 7, 2022; institution granted December 19, 2022
• Claims challenged: 1, 3, 5, 9, 12, 13, 15, 20, 21
• Final Written Decision: December 18, 2023 — claims 1, 3, 5, 9, 12, 13, and 15 held unpatentable as anticipated by Sonohara (U.S. Patent No. 5,627,656); claims 20 and 21 not shown unpatentable. Rehearing denied March 27, 2024.
• Status: Patent owner appealed; the Federal Circuit affirmed (see No. 9 below).

8. Netflix, Inc. v. VL Collective IP LLC — IPR2023-00891

• Petitioner: Netflix, Inc. (and Netflix Streaming Services, Inc.)
• Patent Owner: VL Collective IP LLC
• Filing date: Petition filed May 5, 2023; institution granted January 12, 2024
• Final Written Decision: January 7, 2025 — all 22 claims of the '794 patent held unpatentable.
• Status: Per PTAB docket, "Final Written Decision – Appealed." The Google Patents litigation list shows a corresponding Federal Circuit appeal (see No. 10 below).

III. Federal Circuit Appeals

9. VL Collective IP, LLC v. Unified Patents, LLC — No. 24-1890 (Fed. Cir.)

• Appellant: VL Collective IP, LLC
• Appellee: Unified Patents, LLC
• Jurisdiction: U.S. Court of Appeals for the Federal Circuit (appeal of IPR2022-01086)
• Filed: June 3, 2024 (docketed; notice of appeal dated May 29, 2024)
• Outcome: Affirmed in a nonprecedential opinion issued February 20, 2026 — the court affirmed the PTAB's claim constructions and its unpatentability findings as to claims 1, 3, 5, 9, 12, 13, and 15. Judgment entered the same day.

10. VL Collective IP, LLC v. Netflix, Inc. (or similar caption) — No. 25-1525 (Fed. Cir.)

• Jurisdiction: U.S. Court of Appeals for the Federal Circuit
• Filing date: 2025 (per Google Patents litigation listing, case 25-1525)
• Status: Reported on the patent's litigation history as a CAFC case; almost certainly the patent owner's appeal of the Netflix IPR2023-00891 Final Written Decision invalidating all 22 claims. I could not verify the caption, filing date, or current status from PACER/CAFC sources within my search limits, so treat these details as unverified.

Key takeaways

• The '794 patent has been asserted by VideoLabs/VL Collective against Netflix, Roku, Hisense, Amazon (dismissed), and Apple (dismissed), and is/was also the subject of a Starz action in Delaware.
• It has been challenged in two IPRs: IPR2022-01086 (Unified Patents; partial invalidation, affirmed by the CAFC on Feb. 20, 2026) and IPR2023-00891 (Netflix; complete invalidation of all 22 claims on Jan. 7, 2025, currently on appeal at the CAFC).
• Given the full invalidation of all claims in IPR2023-00891, the practical enforceability of the '794 patent is presently in question, subject to the pending Federal Circuit appeal.

Caveat: I was unable to independently verify the exact filing date and final disposition of the Starz case (1:21-cv-01448-JLH) and the details of CAFC No. 25-1525 within the search limits; those items are marked as unverified above.

Proceedings overview

**Two Inter Partes Reviews have been filed against US 8,605,794 — both ended in Final Written Decisions finding claims unpatentable (IPR2022-01086: claims 1, 3, 5, 9, 12, 13, 15 invalidated, claims 20–21 sustained, now affirmed on appeal; IPR2023-00891: all 22 claims invalidated, currently on appeal) — with zero settlements, zero institution denials, and zero proceedings still in trial, giving a defendant the strongest possible defensive posture: every claim of the '794 patent has been found unpatentable by the PTAB at least once, and the first set of invalidations has been affirmed by the Federal Circuit.

Data flag: The structured "PTAB proceedings on file" block states the USPTO Open Data Portal returns no AIA trial proceedings for this patent as of its most recent ingest. That is stale/incomplete — web search verified both proceedings below against primary sources (PTAB/PTACTS filings, the CAFC opinion in 24-1890, and case-summary databases). Treat the two proceedings below as the complete known set.

IPR2023-00891 — Netflix, Inc. et al. v. VL Collective IP LLC

(Most impactful first: this is the one that invalidated every claim.)

• Type: Inter Partes Review
• Filed: 2023-05-05
• Status: "Final Written Decision - Appealed" — the FWD (2025-01-07) found all 22 claims unpatentable; Patent Owner appealed to the Federal Circuit (No. 25-1525), which is pending as of 2026-08-31.
• Judge panel: Karl D. Easthom (author of the FWD), Jeffrey S. Smith, David C. McKone.
• Petition grounds: Netflix challenged all 22 claims of the '794 patent. (The specific prior-art references and statutory bases in Netflix's petition were not independently retrievable in my searches — the FWD on PTAB E2E/PTACTS is the authoritative source for the grounds; do not rely on the art list in this memo.)
• Institution decision: Instituted on 2024-01-12 for all challenged claims. Notably, the Board granted Patent Owner's motion for additional discovery (2024-03-28, Paper 26) into whether Netflix was a real party in interest ("RPI") in the earlier Unified Patents IPR (IPR2022-01086) — Patent Owner's theory being that Netflix directed Unified's IPR and should be dismissed/estopped. After the discovery, the Board rejected the RPI challenge, adopting the Samsung IPR reasoning that Unified's zone members are not RPIs of Unified's IPRs, and specifically found that Netflix was not time-barred when it filed (it filed 2023-05-05, within one year of the Delaware complaint adding the '794 patent on 2022-05-06). See PTACTS, IPR2023-00891, Order on RPI/Discovery issues (available via PTAB PTACTS docket for IPR2023-00891).
• Final Written Decision (2025-01-07): The Board held all 22 claims (1–22) unpatentable by a preponderance of the evidence. This covers the two independent method/device claims (1 and 9), the marker-based method claim 22, and every dependent claim — including claims 2, 4, 6, 7, 8, 10, 11, 14, 16–19, and 22 that were not challenged in the Unified IPR, and claims 20–21 that Unified failed to knock out. Sources: Patexia summary for IPR2023-00891; PTAB E2E/PTACTS case file.
• Settlement / termination: No settlement. Trial terminated 2025-01-07 by issuance of the FWD.
• Appeal: CAFC No. 25-1525 — Patent Owner's appeal of the FWD, filed 2025. As of 2026-08-31, no CAFC decision has been located; treat the appeal as pending. Do not confuse with CAFC 25-1132 (decided 2026-08-10), which affirmed a PTAB win for Netflix on the different '559 patent.
• Defensive value: This is the cleanest possible defense. The PTAB has found every single claim of the '794 patent unpatentable. Any demand letter or infringement theory built on any claim — including claims 20–21, which survived the Unified IPR — runs head-on into an on-point FWD. Until the CAFC disposes of 25-1525, the FWD is not "final" for issue-preclusion/certificate purposes, but it is more than enough to support a stay motion and to gut the value of the assertion.

IPR2022-01086 — Unified Patents, LLC v. VL Collective IP LLC

• Type: Inter Partes Review
• Filed: 2022-06-07
• Status: "Final Written Decision - Appealed" — the FWD (2023-12-18) found claims 1, 3, 5, 9, 12, 13, and 15 unpatentable; Patent Owner appealed, and the CAFC affirmed on 2026-02-20 (No. 24-1890, nonprecedential). With the appeal resolved in the petitioner's favor, this set of invalidations is now final.
• Judge panel: Karl D. Easthem, Jeffrey S. Smith, Patrick M. Boucher (Boucher authored the FWD). (Docket listings also show David McKone on the institution panel.)
• Petition grounds: Unified challenged claims 1, 3, 5, 9, 12, 13, 15, 20, and 21 as anticipated under § 102 by Sonohara, U.S. Patent No. 5,627,656.
• Institution decision: Instituted 2022-12-19 on all challenged claims.
• Final Written Decision (2023-12-18; reported at 2023 WL 8717013; Paper 37): The Board declined to construe "assignment rule" (rejecting Patent Owner's proposed negative limitation excluding "exact timing information"), adopted the specification's definition to construe "content-related … data segments" as segments having syntactical meaning within the data file, and held: "Based on the record before us, Petitioner has shown, by a preponderance of the evidence, that claims 1, 3, 5, 9, 12, 13, and 15 are unpatentable, but has not shown, by a preponderance of the evidence, that claims 20 and 21 are unpatentable." The Board's basis for sparing claims 20–21: Sonohara's "assignment rule" was "based on" two different timestamps, and dependent claims 20–21 require a rule not based on a timestamp. Patent Owner's rehearing request was denied (2024-03-27).
• Settlement / termination: No settlement. (Notably, Unified sought leave to terminate for budget reasons in August 2023 — before the FWD — which became part of Patent Owner's later RPI attack on Netflix; the Board in IPR2023-00891 found no evidence that request was made for Netflix's benefit.)
• Appeal: CAFC No. 24-1890 (filed 2024-06-03). Oral argument 2026-01-09; nonprecedential opinion 2026-02-20 affirming the PTAB. The CAFC rejected Patent Owner's arguments on: (1) the "assignment rule" construction — Patent Owner's new "timestamps" argument was forfeited because below it had argued only for exclusion of "exact timing information" (Conoco/Interactive Gift waiver rule); (2) the "content-related data segments" construction; and (3) the Board's consideration of Unified's first-time Reply argument. Opinion: CAFC 24-1890 opinion (PDF); docket: CourtListener 24-1890 docket; PTAB FWD available on PTAB E2E/PTACTS for IPR2022-01086.
• Defensive value: Independent claims 1 and 9 are dead — invalidated over Sonohara and the invalidation is now affirmed and final. Any infringement theory built on either independent claim is, at minimum, a Rule 11 problem for the claimant. Claims 20–21 survived this IPR but were subsequently invalidated in IPR2023-00891. The CAFC affirmance also locks in claim constructions (no "no-timestamps" limitation in "assignment rule"; "content-related data segments" = segments with syntactical meaning) that are binding on the patent owner in this litigation cluster.

Strategic summary

Claim status across both IPRs (as of 2026-08-31):

• Invalidated and final (certification pending but appeal rights exhausted): claims 1, 3, 5, 9, 12, 13, 15 (IPR2022-01086 FWD over Sonohara, affirmed in CAFC 24-1890 on 2026-02-20).
• Invalidated, on appeal (not yet final): all 22 claims (IPR2023-00891 FWD, 2025-01-07; CAFC 25-1525 pending).
• Sustained by any FWD: none, in the end. Claims 20–21 were sustained in the Unified IPR but were then found unpatentable in the Netflix IPR. Every other claim was invalidated in the Unified IPR outright or was never challenged there (claims 2, 4, 6, 7, 8, 10, 11, 14, 16–19, 22) — but the Netflix IPR swept up all of them.
• Untested by any IPR: none, as of the Netflix FWD.

In practical terms the '794 patent is, today, a patent with zero claims that have not been found unpatentable by the PTAB. The only thing keeping it on the register is the pendency of CAFC 25-1525 (and the mechanics of certificates of cancellation). Once 25-1525 resolves against the patent owner — or is abandoned — the Director's certificates will cancel every claim.

Estoppel landscape (§ 315(e)(2)): The statutory bar runs to each petitioner and its privies. Unified Patents is estopped from re-litigating (or having its privies litigate) any ground it raised or reasonably could have raised in IPR2022-01086; Netflix is similarly estopped as to IPR2023-00891. Critically for the broader litigation cluster, the Board in IPR2023-00891 rejected Patent Owner's RPI attack and found Netflix was not a privy of Unified in the '086 IPR — so Netflix's victory is not tainted by Unified's estoppel, and vice versa. For a new defendant (e.g., a company not already a party or privy), § 315(e)(2) imposes no bar: you may raise any § 102/§ 103 ground, including Sonohara and whatever Netflix used, and you are not time-barred until one year after service of the complaint (35 U.S.C. § 315(b)). That said, a new IPR is likely unnecessary: the better play is to ride the existing FWDs.

Pattern signals: This is a textbook coordinated defense of a broad, aggressively-litigated NPE patent. Unified Patents — a defensive aggregator — filed first (2022-06-07) on a strategic subset of claims; Netflix — an asserted defendant (D. Del. 22-229) — filed second (2023-05-05) and went for all 22 claims, explicitly not time-barred at filing. The patent owner (VideoLabs/VL Collective) litigated hard — discovery into RPI/privy relationships, rehearing requests, and two CAFC appeals — and lost every round: both FWDs, both rehearing denials, and the first CAFC appeal. The patent owner has continued asserting the patent in parallel (e.g., Hisense in E.D. Tex., stayed/closed cases against Roku, Apple, Amazon), but the PTAB record now gives every defendant in that cluster a common, decisive validity defense.

Recommended next steps

• If you are a defendant (or recipient of a demand letter) today: Lead with IPR2023-00891 — the FWD (2025-01-07, all 22 claims unpatentable; docket and FWD on PTAB E2E/PTACTS) — and with IPR2022-01086, which is final and affirmed: claims 1, 3, 5, 9, 12, 13, and 15 are unpatentable over Sonohara, affirmed, CAFC 24-1890 opinion. Move to stay any district court case pending 25-1525, and be prepared to argue issue preclusion on claims 1, 3, 5, 9, 12, 13, 15 immediately (the CAFC affirmance has made that FWD final). Any infringement theory resting on claim 1 or claim 9 is now affirmance-proof dead — demand letters citing them should be answered with a citation to 24-1890.
• Monitor the pending appeal: CAFC No. 25-1525 (appeal of IPR2023-00891) has no decision as of 2026-08-31. Check the CAFC docket/CourtListener for the briefing schedule and oral-argument notice; a decision is reasonably likely within the next term. If the CAFC affirms, every claim of the '794 patent is canceled; if it reverses, only claims 1, 3, 5, 9, 12, 13, 15 remain dead (and the surviving claims 2, 4, 6–8, 10, 11, 14, 16–22 would be back in play).
• Do not assume the USPTO record shows the full picture: the ODP-derived "proceedings on file" block is empty, but both IPRs exist and are documented in PTACTS, PTAB E2E, and the CAFC. If you are filing papers relying on these FWDs, pull the actual FWD documents from PTAB E2E (IPR2022-01086, Paper 37; IPR2023-00891 FWD) and cite them directly rather than relying on the ODP status field.
• Consider your own IPR only if needed: if you are a new defendant outside the 1-year § 315(b) bar and want an independent invalidation (e.g., to avoid any estoppel or finality gap), you can file on any § 102/§ 103 ground — including Sonohara — because neither prior petitioner is your privy. But the more efficient path is almost certainly stay-plus-FWD, not a third IPR.

I'll reconstruct the assignment chain for the '794 patent. I already have the Google Patents legal-events chain (Siemens → Lough Corrib → VideoLabs → VL IP Holdings → VL Collective IP, plus Praetor security interests). The critical missing piece is the USPTO Assignment Center records — especially the correspondents of record, which are the main NPE tell. Let me verify those records.

The direct searches for reel/frame records didn't surface assignment-center pages (those are often not indexed). Let me try alternate routes: the attorneys representing VideoLabs/VL Collective (the correspondents of record are likely among them), and mirrors of USPTO assignment data.

I've exhausted my search budget, but I have enough verified data to reconstruct the chain. Important transparency note up front: USPTO Assignment Center pages (assignment.uspto.gov / assignmentcenter.uspto.gov) are not indexed by web search, so I could not retrieve the correspondent of record for each recorded assignment from my available sources. The reel/frame numbers and conveyance types below come from the USPTO-derived legal-events table in the authoritative Google Patents record supplied with this patent (which mirrors the Assignment Center data), and from court/PTAB dockets. Where a data point (correspondent names) could not be verified, I say so rather than fabricate it.

Inventors

Pattern note: This is a textbook industry–university collaboration (2 Siemens CT researchers + 2 Klagenfurt academics) feeding Siemens' MPEG-21 standardization work — not the "all inventors depart within 12 months" pattern that often precedes a fire-sale. There is no evidence any inventor left Siemens around filing; the inventors' assignment to Siemens was executed 2007-11-02 to 2007-11-12 (reel 022169/0584), i.e., a confirmatory assignment timed to the US national-phase filing (Nov 2007), a standard practice rather than a red flag.

Original assignee

• Siemens AG (German multinational; headquartered Munich). Line of business: electrification, automation, digitalization, industrial manufacturing, and (at the relevant time) communications/networking equipment. This patent originated from Siemens Corporate Technology research on MPEG-21 bitstream adaptation for mobile multimedia streaming.
• Product evidence: No evidence found that Siemens shipped a product embodying the claims. The invention is a research/standardization-era patent (gBSD/BSD-based segment synchronization), consistent with Siemens CT's pre-competitive MPEG-21 work rather than a shipped product line.
• Current status: Operating — Siemens AG remains one of Europe's largest industrial companies; never in bankruptcy. The '794 patent was deliberately divested in 2018 (see timeline), i.e., a portfolio sale, not a fire-sale.

Assignment timeline

All reel/frame data below is drawn from the USPTO-recorded legal events as reflected in the Google Patents record for US8605794 (authoritative source supplied). Correspondent of record could not be retrieved from my sources — verify each entry at the USPTO Assignment Center search page (https://assignmentcenter.uspto.gov/ or https://assignment.uspto.gov/patent/index.html). Every entry below is a real recorded conveyance; the chain is complete.

• 2007-11-02 to 2007-11-12 (executed) / recorded 2009-01-21 — Reel 022169/0584

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Michael Ransburg, Hermann Hellwagner, Andreas Hutter, Jörg Heuer (the four inventors)
  • Assignee: Siemens Aktiengesellschaft
  • Correspondent: not retrievable from available sources
  • Context: Confirmatory inventor→employer assignment, executed ~18 months after the PCT filing and timed to the US national-phase entry; records Siemens as the original owner.

• 2018-10-01 (effective) / recorded 2018-10-09 — Reel 047103/0709

  • Conveyance: Assignment
  • Assignor: Siemens Aktiengesellschaft
  • Assignee: Lough Corrib Intellectual Property Limited (Ireland)
  • Correspondent: not retrievable from available sources
  • Context: Siemens divests the patent to an Irish IP-holding vehicle — the first step out of the operating company and into an assertion chain.

• 2019-10-15 (effective) / recorded 2019-10-16 — Reel 050733/0526

  • Conveyance: Assignment
  • Assignor: Lough Corrib Intellectual Property Limited
  • Assignee: VideoLabs, Inc. (California)
  • Correspondent: not retrievable from available sources
  • Context: Transfer-to-asserter — Lough Corrib passes the portfolio to VideoLabs, the entity that runs the licensing/enforcement campaigns.

• 2019-12-19 (effective) / recorded 2019-12-31 — Reel 051391/0919

  • Conveyance: Assignment
  • Assignor: VideoLabs, Inc.
  • Assignee: VL IP Holdings LLC (California)
  • Correspondent: not retrievable from available sources
  • Context: Internal reorg — VideoLabs moves assets up into a holding LLC (same "VL" family, same effective date as the next entry).

• 2019-12-19 (effective) / recorded 2019-12-31 — Reel 051392/0412

  • Conveyance: Assignment
  • Assignor: VL IP Holdings LLC
  • Assignee: VL Collective IP LLC (Delaware)
  • Correspondent: not retrievable from available sources
  • Context: Internal reorg completing the chain — VL Collective IP LLC is the entity that appears as co-plaintiff in every VideoLabs suit (Netflix, Roku, Apple, Hisense, TCL, Dell, Microsoft).

• 2019-12-04 (effective) / recorded 2020-02-07 — Reel 051748/0267

  • Conveyance: Security Interest (Security Agreement)
  • Assignor/Grantor: VL Collective IP LLC
  • Assignee/Secured Party: Praetor Fund I, a sub-fund of Praetorium Fund I ICAV (Ireland)
  • Correspondent: not retrievable from available sources
  • Context: Portfolio securitization / litigation financing — the Irish ICAV fund takes a security interest in the portfolio.

• 2020-03-24 (effective) / recorded 2020-03-31 — Reel 052272/0435

  • Conveyance: Security Interest (Security Agreement)
  • Assignor/Grantor: VL Collective IP LLC
  • Assignee/Secured Party: Praetor Fund I, a sub-fund of Praetorium Fund I ICAV
  • Correspondent: not retrievable from available sources
  • Context: Second, supplemental security interest — additional funding draw against the same portfolio.

• 2022-12-28 (effective) / recorded 2023-03-14 — Reel 062977/0325

  • Conveyance: Release by Secured Party
  • Assignor: Praetor Fund I, a sub-fund of Praetorium Fund I ICAV
  • Assignee: VL Collective IP LLC
  • Correspondent: not retrievable from available sources
  • Context: Security interest released (post-IPR-filing, after the portfolio's value collapsed) — title reverts cleanly to VL Collective IP LLC.

Non-assignment legal events on the record (not conveyances): US publication 2009-07-09; grant 2013-12-10; certificate of correction 2014-02-25; maintenance-fee payments 2017 and 2020 and 2025 (12th-year fee paid 2025-06-10, keeping the patent "Active" on the register despite the PTAB invalidations).

Timeline diagram

timeline
    title Ownership of US 8605794
    2005 : Priority filing in Germany
    2006 : PCT application filed
    2009 : Inventors assign to Siemens
    2013 : Patent issued
    2018 : Siemens to Lough Corrib IP
    2019 : Lough Corrib to VideoLabs
         : VideoLabs to VL IP Holdings
         : VL IP Holdings to VL Collective IP
    2020 : Security interest to Praetor Fund
    2022 : Unified Patents IPR filed
    2023 : Praetor releases security interest
    2025 : Netflix IPR invalidates all claims
    2026 : CAFC affirms Unified IPR win

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT

   • Reel 047103/0709 (2018-10-09): operating company Siemens AG → Lough Corrib Intellectual Property Limited (Irish holding vehicle, "Intellectual Property" name, no products).
   • Reels 050733/0526, 051391/0919, 051392/0412 (2019): onward into VideoLabs, Inc. → VL IP Holdings LLC → VL Collective IP LLC — Delaware/California LLCs whose entire business, per VideoLabs' own website (videolabsinc.com/licensing-programs), is acquiring and licensing portfolios ("VL Collective IP ... acquires patents originating from ... Samsung, LG, Ericsson, Panasonic, Siemens, HP, and Nokia ... comprised of more than 150 assets worldwide"). No product business exists anywhere in the post-2018 chain. This is direct, non-inferential evidence: the current assignee's own marketing describes a licensing-only vehicle.

2. Known asserter in the chain — PRESENT

   • Current assignee VL Collective IP LLC (and prior link VideoLabs, Inc.) is a documented high-frequency patent plaintiff: Netflix (D. Del. 22-229), Roku (D. Del. 23-1136), Apple (N.D. Cal. 23-1307), Amazon (W.D. Tex. 22-79), Hisense (E.D. Tex. 24-904), TCL (E.D. Tex. 25-161), Starz (D. Del. 21-1448), Dell, and Microsoft (E.D. Tex. 26-346). Both Unified Patents (IPR2022-01086) and Netflix (IPR2023-00891) challenged the '794 patent — Unified Patents is a defensive aggregator that specifically tracks high-frequency NPEs. An IAM profile of its litigation counsel (Bunsow De Mory LLP) explicitly describes "ongoing enforcement campaigns in video compression technologies" for VideoLabs. The entity appears on no product-manufacturer list of any kind.

3. Repeat correspondent across the chain — UNCLEAR (unverifiable)

   • I could not retrieve the correspondent of record for any reel/frame from my sources (Assignment Center pages are not web-indexed). What I can verify is recurring litigation counsel across the VL Collective portfolio: Reichman Jorgensen Lehman & Feldberg LLP (Christine E. Lehman, Brian C. Baran, Naveed S. Hasan — CAFC appeal 24-1890 and PTAB defense in IPR2023-00891 per the IPR docket and CourtListener party records) and Bunsow De Mory LLP (district-court enforcement). Recurrence of one law firm across many matters is a corroborating signal, but because I cannot tie a specific attorney to a specific assignment reel/frame, I mark this signal unclear rather than present. Pull the Assignment Center records to complete this check — the correspondent on reels 050733/0526, 051391/0919, and 051392/0412 is likely the same attorney and is the single best verification data point.

4. Cascading transfers — PRESENT

   • Lough Corrib → VideoLabs (2019-10-16) → VL IP Holdings (2019-12-31) → VL Collective IP (2019-12-31) — three chained transfers in under 3 months (and the entire post-Siemens chain completes in ~14 months, Oct 2018–Dec 2019). The last two transfers share the same effective date (2019-12-19) and same recording date (2019-12-31), with same-family "VL" names — consistent with a single coordinated reorg executed in one stroke. Reels 050733/0526, 051391/0919, 051392/0412.

5. Pre-litigation transfer — NOT PRESENT (but see securitization note)

   • The assertion chain completed 2019-12-31; the first complaint naming the '794 patent was the Netflix FAC in May 2022 (D. Del. 22-229) — more than 2 years later, so the strict 6-month pre-litigation-transfer signal is absent. However, the Praetor Fund I security interests (reels 051748/0267, 2020-02-07, and 052272/0435, 2020-03-31 — Praetorium Fund I ICAV is an Irish collective-investment vehicle, i.e., third-party litigation finance) show the portfolio was securitized to fund assertion ~2 years before the suits. That is a financing-for-litigation tell even though the title transfers themselves were earlier.

6. Bankruptcy fire-sale — NOT PRESENT

   • Siemens AG has never been in bankruptcy. The 2018 divestiture (reel 047103/0709) was an orderly, deliberate portfolio sale to an Irish holding company, not a §363 bankruptcy sale (contrast Kodak, Nortel, etc.).

7. Privateering — UNCLEAR

   • The pattern (Siemens → Irish holding → NPE → enforcement against streaming/CE companies) is consistent with privateering, but I found no SEC filing, press report, or license-back agreement evidencing that Siemens retained a revenue share or directed the assertions. VideoLabs' website frames it as "partnered with and acquired" portfolios. Without a documented revenue-share or control arrangement, I will not assert privateering.

8. Defensive aggregator (anti-NPE) — NOT PRESENT

   • The chain terminates at VL Collective IP LLC, an asserting entity. No transfer to RPX, Allied Security Trust, LOT Network, Unified Patents, or OIN ever occurred. The IPR victories by Unified Patents and Netflix are attacks on the patent by defensive actors, not ownership by them. (Inverse observation: the patent has been effectively neutralized by the PTAB regardless — all 22 claims invalidated in IPR2023-00891, and claims 1, 3, 5, 9, 12, 13, 15 invalidated and affirmed in IPR2022-01086 / CAFC 24-1890 — but that is a validity fact, not an ownership fact.)

Verdict

NPE — high confidence.

Three-plus strong, independently documented signals converge: (1) shell-entity transfer — the patent left operating company Siemens AG (reel 047103/0709, 2018-10-09) and passed through three licensing-only LLC/Inc. vehicles (Lough Corrib IP → VideoLabs → VL IP Holdings → VL Collective IP; reels 050733/0526, 051391/0919, 051392/0412, 2019), whose sole business is portfolio licensing per VideoLabs' own website; (2) known asserter — VL Collective IP/VideoLabs is a high-frequency plaintiff (Netflix, Roku, Apple, Amazon, Hisense, TCL, Dell, Microsoft) that was itself targeted by the defensive aggregator Unified Patents in IPR2022-01086; and (3) cascading transfers + litigation-funding security interests — three chained transfers within 14 months followed by securitization to Praetor Fund I, an Irish ICAV fund (reels 051748/0267, 052272/0435, 2020). The only gaps are correspondent-of-record verification (Assignment Center pages unindexed; marked unclear) and privateering evidence (no SEC/press proof of a Siemens revenue share; marked unclear) — neither gap undermines the core conclusion.

Verification: run the same search at the USPTO Assignment Center (https://assignmentcenter.uspto.gov/) — searching patent number 8605794 will return all eight conveyances above (reels 022169/0584, 047103/0709, 050733/0526, 051391/0919, 051392/0412, 051748/0267, 052272/0435, 062977/0325); capture the correspondent of record on each to complete signal 3.

I'll verify the patent record and its citation list against the USPTO sources before analyzing the prior art.

Good — I've confirmed the key record. Now let me verify the IPR grounds and a few specific references, since the strongest prior art appears in the IPRs rather than the examiner's citation list.

USPTO record verification for US 8,605,794

I searched for the exact number 8605794 (excluding near-miss numbers in the same VideoLabs/VL Collective cluster). Confirmed record:

Sources: Google Patents US8605794B2, Justia claims, IPVerse IPR2022-01086, IPVerse IPR2023-00891.

Statutory framework: the application was filed 2006-03-16, so pre-AIA 35 U.S.C. § 102(a)/(b)/(e) governs the prior-art analysis (not § 102(a)(1)/(a)(2)). References published after 2005-04-12 can still qualify under § 102(e) via their earlier U.S. filing dates; I flag every such instance below.

Framing: there are TWO prior-art universes for this patent

1. The 20 patent citations + 5 non-patent citations of record — i.e., the references the examiner actually placed on the face of the patent. Filing-date metadata below is drawn from the authoritative Google Patents citation table supplied with this patent.
2. The prior art the IPRs actually relied on — which is a different set. Crucially, the reference that succeeded in invalidating the claims on § 102 (Sonohara, U.S. 5,627,656) is NOT on the face of the patent. It was found by Unified Patents. This is the single most important finding in this memo.

I build on, rather than repeat, the previously generated PTAB section — with one correction flagged in §C below.

A. Patent citations of record (examiner-cited), § 102 assessment

Tier 1 — Genuinely on-point synchronization references (best § 102 candidates)

1. US 5,598,352 — Cirrus Logic, Inc. — "Method and apparatus for audio and video synchronizing in MPEG playback systems"

• Filed 1994-09-30; issued 1997-01-28
• Discloses synchronizing audio and video elementary streams in MPEG playback.
• § 102 assessment: The best examiner-cited § 102 candidate for claims 1 and 9 (and hence their dependents) if read broadly enough to cover a non-timestamp pairing rule. In practice it aligns streams using MPEG system-layer timing (PTS/DTS-style), so it is far stronger as a § 103 reference and supports claims 20–21 being patentable over it. This is the reference most likely to have generated the "not based on a timestamp" negative limitation.

2. US 2001/0052943 A1 — Sone, Takurou — "Multimedia system with synchronization of music and image tracks"

• Filed 2000-06-02; published 2001-12-20
• Synchronizes music tracks and image tracks in a multimedia system.
• § 102 assessment: Claims 1, 5, 9, 12, 15 are the candidate set — pairing content-meaningful segments of a music file with segments of an image file is precisely the claimed architecture. Whether it anticipates turns on whether its pairing is expressed as an "assignment rule" rather than as timing alignment. Strong § 102 candidate / strong § 103 reference. Published well before 2005-04-12, so it is § 102(b) art.

3. EP 1 494 430 A2 — Samsung Electronics Co., Ltd. — "System and method for time synchronization between multimedia content and segment metadata"

• KR priority 2003-06-30; EP filing 2004-06-24; published 2005-01-05; granted as EP 1 494 430 B1 (2014-09-03); US counterpart US 7,428,696
• Discloses synchronizing multimedia content segments with segment metadata using first/second segment time information, computing reference time points and correcting a relative time point. [Verified: EPO PISE record, Google Patents EP1494430A2]
• § 102 assessment: Claims 1, 9, 13 are candidates (content segments vs. metadata segments, synchronized output). But because it is expressly timestamp/time-reference based, it cuts the other way on claims 20–21. This reference is almost certainly the "Shin" reference the examiner used during prosecution — the patent owner's own IPR presentation labels "Shin" as the "Prosecution Reference" that "adjusts timing information" and is timestamp-based, and describes exactly the Samsung disclosure. [Source: Patent Owner's IPR2023-00891 presentation, PTACTS]
• Note: I did not independently confirm that the "Shin" of the prosecution history is the inventor of US 7,428,696. Treat the Shin↔Samsung identification as a high-confidence inference, not a verified fact.

4. US 7,194,194 B2 — Matsushita Electric Industrial Co., Ltd. — "Method and an apparatus for reproducing bitstream having non-sequential system clock data seamlessly therebetween"

• Filed 1995-09-29; issued 2007-03-20
• Seamless reproduction across non-sequential bitstreams / discontinuous system-clock data.
• § 102 assessment: Relevant to claims 1, 2, 22 (segment ordering and jumping to non-sequential points). It teaches seamless transition between segments, not a "second-segment-to-first-segment assignment rule," so § 103 rather than § 102. Strong companion reference for claim 22.

Tier 2 — § 102(e)-only references (published after the 2005-04-12 priority date)

These four were cited for their earlier U.S. filing dates; they cannot be § 102(a)/(b) art because their publication postdates the '794 priority date. Any § 102 case must run through pre-AIA § 102(e) (U.S. filing date), and none of them is squarely on the "assignment rule."

5. US 2005/0155063 A1 — Sedna Patent Services, LLC — "Server-centric customized interactive program guide in an interactive television environment"

• Underlying filing 1999-04-15; published 2005-07-14 (post-priority)
• Server-side program-guide customization. § 102(e) candidate only; not anticipating any claim. Same family as US 6,904,610 below.

6. US 2006/0165379 A1 — Agnihotri, Lalitha A. — "System and method for generating a multimedia summary of multimedia streams"

• Filed 2003-06-30; published 2006-07-27 (post-priority)
• Segment-based multimedia summarization using content descriptors.
• § 102(e) assessment: Best § 102(e) candidate of this group for claims 7, 8 and 22 (content-related markers, scene-change selection, ordering segments for output). Anticipation of claims 1/9 is unlikely.

7. US 2005/0237952 A1 — Marconi Communications, Inc. — "Method and apparatus for conferencing with bandwidth control"

• Filed 2004-03-19; published 2005-10-27 (post-priority)
• Bandwidth-controlled conferencing. § 102(e) only; peripheral. At most context for claims 5, 12, 15 (media data).

8. US 2005/0286783 A1 — Matsushita Electric Industrial Co., Ltd. — "Moving image recording method"

• Filed 2004-06-24; published 2005-12-29 (post-priority)
• Moving-image recording. § 102(e) only; peripheral. At most context for claims 5, 12, 15.

Tier 3 — Cited, but not anticipating any claim (background / field / unrelated)

B. Non-patent citations of record (5)

C. The prior art that ACTUALLY invalidated the claims (not on the face of the patent)

This is the operative § 102 body of art, and it should not be confused with the citation list above. No examiner-cited reference ever invalidated a claim of the '794 patent; the IPR art did.

C-1. IPR2022-01086 (Unified Patents) — § 102 anticipation by Sonohara

US 5,627,656 — Sonohara et al.

• Filed/issued dates: not verified in my searches this session (issued 1997-05-13 per standard numbering era; verify before citing)
• Discloses reproducing image (video) data and sound (audio) data stored in a single file, with a header carrying "data identifying numbers" so that image/sound unit times bearing the same number are reproduced in synchronism; device includes reader 41 and main controller 43.
• § 102 claims invalidated: 1, 3, 5, 9, 12, 13, 15 — Final Written Decision 2023-12-18; affirmed by the Federal Circuit in No. 24-1890 (nonprecedential, 2026-02-20). Rehearing denied.
• Claims SURVIVING on § 102: 20 and 21 — the Board held Sonohara's assignment rule was timestamp-based, so the "not based on a timestamp" limitations were not met. (Those claims were later invalidated in the Netflix IPR.)
• The Board also declined to construe "assignment rule" and adopted the specification's definition of "content-related … data segments" (segments having syntactical meaning within the data file), a construction the CAFC affirmed.
• Source: PTACTS IPR2022-01086 documents, CAFC 24-1890 opinion, IPVerse case record.

Important: Sonohara is absent from the 20-item citation list on the face of US 8,605,794. It was independently located by the petitioner. It is therefore the single most valuable prior-art reference for this patent notwithstanding the examiner never citing it.

C-2. IPR2023-00891 (Netflix) — § 102 anticipation by Comps, § 103 with Wan and Ahn

Per a secondary case-summary source (ExParte AI Lab case summary, IPR2023-00891) the petition ran three grounds:

• Comps teaches synchronizing different data types (sound, pictures, text) held in separate tracks of a multimedia document using non-temporal "SYNCH1/SYNCH2/SYNCH3" commands — the petitioner mapped those commands to the claimed "assignment rule," and argued Comps expressly states the commands "are not temporal commands … but are instead dependent on a specific event," thereby satisfying claims 20–21's "not based on a timestamp" limitation.
• Wan supplies XSLT generation of presentation scripts (claim 6) and a scene description stream with content-related markers (claims 7, 8, 22).
• Ahn supplies XML-encoded metadata (claim 14).
• Result: FWD 2025-01-07 held all 22 claims unpatentable; on appeal at CAFC 25-1525 (pending per the record as of my search).
• The PTACTS excerpts confirm the substance: "Comps teaches an 'assignment rule' that is not based on a timestamp … Comps makes clear that its SYNCH commands 'are not temporal commands … but are instead dependent on a specific event,' and they are not 'present in the tracks at regular time intervals.'" (PTACTS IPR2023-00891 documents)

Verification flags (do not cite these without pulling the FWD/exhibits):

• The patent numbers 7,386,782 (Comps), 2004/0024898 (Wan), and 2004/0098398 (Ahn) come from an AI-generated secondary summary, cross-checked only in part against PTACTS quotations of "Comps" and "Wan." The PTACTS record references Comps as EX1004 and Bederson's declaration as EX1003, which is consistent with Comps being a patent exhibit, but I could not verify the numeric identifiers directly. Pull the FWD and EX1004 from PTAB E2E/PTACTS before relying on them.
• Filing/publication dates for Comps, Wan, and Ahn were not retrievable within my search budget. All three must predate 2005-04-12 (Comps as a patent, Wan/Ahn as pre-AIA § 102(e) applications).
• Discrepancy with the previously generated PTAB section: that section stated Netflix's prior-art grounds "were not independently retrievable." Partially resolved above — but note the earlier section's implicit assumption that IPR2022-01086 relied only on Sonohara is now in tension with the IPVerse case listing for IPR2022-01086, which also names US 6,654,933 to Abbott et al. I could not verify what role, if any, the Abbott reference played in the '086 IPR (the FWD excerpt I retrieved addresses only Sonohara). Treat the Abbott reference as an unresolved data point and verify against the '086 FWD (Paper 37).

D. Ranking: most relevant prior art for US 8,605,794

E. Bottom line

• The examiner's citation list for US 8,605,794 is broad but shallow: 20 patent references and 5 NPL items, of which roughly four (US 5,598,352; US 2001/0052943 A1; EP 1 494 430 A2; and the NPL Li et al. and BBL papers) bear genuinely on the synchronization claims, and the rest are coding-, transport-, delivery-, display-, or unrelated-art background. None of the examiner-cited references was ever used to invalidate a claim.
• The prior art that does anticipate the claims sits outside the face of the patent: US 5,627,656 (Sonohara) — which invalidated claims 1, 3, 5, 9, 12, 13, 15 (final, CAFC-affirmed) — and Comps (reported US 7,386,782, optionally with Wan US 2004/0024898 A1 and Ahn US 2004/0098398 A1), which invalidated all 22 claims in the Netflix IPR.
• The claim set that survived the first IPR — claims 20 and 21 ("assignment rule … not based on a timestamp") — was defeated precisely because Comps showed synchronization commands that are expressly "not temporal commands." That is the reference any future challenger or defendant should lead with, alongside Sonohara.

Uncertainties I am flagging rather than papering over: (i) filing/issue dates for Sonohara and the numeric identifiers/dates for Comps, Wan, and Ahn are unverified against primary documents; (ii) the Shin ↔ Samsung EP 1 494 430 identification is an inference; (iii) the role of US 6,654,933 (Abbott) in IPR2022-01086 is unresolvable on my current evidence; and (iv) NPL-2 (Devillers) may bear a date at or after the 2005-04-12 priority date — verify before treating it as § 102(b) art.

I'll ground this §103 analysis in the actual prior-art references of record. Let me verify the key references (especially Sonohara, the Samsung EP reference, and the non-patent literature).

Let me identify the second reference Netflix used (Abbott) and the other grounds asserted against the '794 patent.

Obviousness Analysis Under 35 U.S.C. § 103 — U.S. Patent No. 8,605,794

Prepared: April 26, 2026
Patent: US 8,605,794 B2 ("the '794 patent") — Method for synchronizing content-dependent data segments of files
Assignee of record: VL Collective IP LLC
Critical date: April 12, 2005 (DE 10 2005 016 866.3); PCT filed March 16, 2006 — pre-AIA § 103(a) applies, and KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), governs the motivation-to-combine inquiry.

This memo builds on the previously generated summary, litigation, PTAB, and assignment sections. It does not repeat their content; where those sections established a fact (e.g., the CAFC affirmance in No. 24-1890, the Netflix FWD invalidating all 22 claims), I cite it rather than restate it.

I. Legal framework and the two claim-construction rulings that drive everything

Three construction points control the § 103 analysis, and all three are now settled as against the patent owner:

Practical consequence: after KSR and after the CAFC affirmance, the patent's asserted point of novelty collapses to a single binary question — does the art teach or suggest (i) pairing content-meaningful chunks of two files and (ii) doing so by a stored mapping rather than by a clock? Every reference below answers yes, and at least one (Comps) answers yes expressly without timestamps, which is why Comps independently disposes of the only claims that survived the Unified IPR.

II. The prior-art universe of record

A. References cited on the face of the '794 patent (the "Prior Art section")

Patent literature (19 citations): US 4,569,026 (Best); EP 0 753 954 A2 (General Instrument); US 5,598,352 (Cirrus Logic); US 5,623,424 (Toshiba); US 5,640,421 (Sony); US 5,661,523 (Matsushita); US 6,052,390 (HP); US 2001/0052943 A1 (Sone); US 6,597,861 B1 (Pioneer); JP 2004-056777 (Matsushita); EP 1 494 430 A2 (Samsung); US 2005/0008240 A1 (Banerji); US 6,904,610 B1 (Sedna); US 2005/0237952 A1 (Marconi); US 2005/0286783 A1 (Matsushita); US 2006/0165379 A1 (Agnihotri); US 7,099,848 B1 (Listen.com); US 7,194,194 B2 (Matsushita); US 7,443,367 B2 (Sharp).

Non-patent literature (5 citations):

• [1] Panis, G. et al., "Bitstream Syntax Description: A Tool for Multimedia Resource Adaptation within MPEG-21," Signal Processing: Image Communication, Vol. 18, No. 8, Sept. 2003, pp. 721–747 — the gBSD/BSD reference.
• [2] Devillers, S. et al., "Metadata streaming use case and requirements," ISO/IEC JTC 1/SC 29/WG 11 M11638 (Jan. 2005).
• [3] Hellwagner, H. et al., "Report of CE on Bit-Stream Adaptation in Constrained and Streaming Environments," M11706 (Jan. 2005).
• [4] Burnett, I. et al., "Bitstream Binding Language — A Grammar for Digital Item Transport," M11611 (Jan. 2005) — BBL.
• Li et al., "Multimedia Segment Delivery Scheme and its performance for real-time synchronization control," IEEE ICC '94, pp. 1734–1738.

B. References asserted in the two IPRs (not on the patent's face — never substantively considered during prosecution)

• Sonohara, U.S. Patent No. 5,627,656 — Motion picture reproducing apparatus in synchronism with sound signal (issued 1997). Track and data identifying numbers in a file header synchronize image and sound unit times. Primary reference in IPR2022-01086.
• Abbott, U.S. Patent No. 6,654,933 (as identified in the third-party IPR summaries; the Abbott family also includes U.S. Pat. No. 7,272,780) — System and method for media stream indexing and synchronization. Separate video and audio "atoms," segmented into GOPs and audio frames, correlated by index files keyed to atom-relative byte positions.
• Comps, U.S. Patent No. 7,386,782 B2 (Alcatel; PCT/FR01/02844, WO 02/23912, published Mar. 21, 2002) — Method for synchronizing a multimedia file. Separate tracks for sound / pictures / text, coordinated by non-temporal SYNCH commands. Primary reference in IPR2023-00891.
• Wan, US 2004/0024898 A1 — XSLT-generated presentation scripts; content markers / scene boundaries.
• Ahn, US 2004/0098398 A1 — XML-source metadata converted into a synchronization format.
• Shin and Rosenau — the references the applicant distinguished during prosecution on the ground that they "consider time information for synchronization, rather than using an assignment rule."

Verification flags. The Comps / Wan / Ahn ground attributions for IPR2023-00891 come from a third-party machine summary of Ex. 1007 (Exparte AI Lab), not from the FWD itself. The substance of Comps is confirmed against the reference's own text and figure descriptions and against PTACTS expert-declaration excerpts. The exact number for Abbott (6,654,933) should be confirmed against the IPR exhibit cover page (EX1004) before being cited in any filing. I have not independently re-verified the Abbott number from the patent's own front page.

III. The independent claims, element by element

Claim 1 (method)

Claim 9 (device)

Substantively identical to claim 1 in apparatus form. Comps expressly discloses "a device for synchronizing data in a multimedia file containing at least one track … and at least one synchronization command in each track," having "first means for reading the data of each track" — a § 112 ¶ 6-style mapping to the claimed "synchronization device." Sonohara discloses "file reproduction processing device 8" with reader 41 and main controller 43.

Claim 22 (marker-based ordering)

Adds: (a) determining read-out order using first and second content-related markers; (b) assigning per the rule; (c) sequentially outputting the assigned pairs. See § IV.E below.

IV. Combinations that render the claims obvious

A. Combination 1 (strongest): Comps alone, or Comps in view of Sonohara — claims 1–22

Comps is the single most dangerous reference because it satisfies the only limitation that defeated Unified's case (claims 20–21).

• "Assignment rule." Comps's SYNCHi commands are a stored mapping in each track that causes a specified second-file event to be output together with a specified first-file event. That is, verbatim, "an assignment rule for assigning each one of the content-related second data segments to one of the content-related first data segments." The SYNCH1 / SYNCH2 / SYNCH3 set is the combined rule of claim 4.
• "Not based on a timestamp" (claims 20–21). Comps says it outright: "The synchronization commands are not temporal commands, as in the prior art, but are instead dependent on a specific event. Thus the fields SYNCHi are not present in the tracks at regular time intervals." This is a direct, textual hit on claims 20–21 — precisely the gap that let those claims survive Sonohara.
• Claim 2 ("further assignment carried out after a number of sequentially succeeding first data segments"). Comps inserts the SYNCH command only "before each event-related command characterizing an important event"; intervening "unimportant" events (fields D₁, JPEGD₀/₁/₂, TEXTD₀/₁ — waiting times and skipped events) fill the interval. That is the claim-2 pattern of one association, then a number of intervening first segments, then the next association. Sonohara supplies the same pattern more explicitly: uniform 1-second unit times, with image #1-02 synchronized to sound #2-02 at a regular interval.
• With Sonohara added, the combination additionally provides the explicit numeric position mapping that claims 3 and 11 require ("the number representing a position of the first data segment within the chronological sequence") — Sonohara's data-identifying numbers are ordinal positions within each track (#1-01, #1-02, …; #2-01, #2-02, …).

Motivation to combine (KSR factors 1–4):

1. Same field. Both address synchronization of separately-stored media component streams for playback (Sonohara: image/sound file reproduction; Comps: "multimedia files … portable systems such as mobile radio terminals, pocket computers").
2. Same problem, expressly articulated by the references themselves. Sonohara's own background states the defect the invention solves: "the file control is intricate because the audio data and the image data are read out independently from the two files. Namely, it is necessary to adjust timings for synchronism … a load imposed on a microprocessor is increased. In addition, the formation of programs for controlling the two files is intricate." Comps solves exactly that problem. A POSITA reading Sonohara is therefore affirmatively directed to a multi-file architecture with an explicit cross-file synchronization mechanism (Comps's SYNCH commands or Sonohara's identifiers) — this is the "reference criticizes the very thing the invention improves, and points to the fix" motivation that KSR and In re Fulton treat as strong.
3. Mutual motivation from Comps's side. Comps's background states the failure mode of multi-track systems — separate microprocessors "drifting" due to oscillator differences and divergent operating systems, so that "the text of a phrase will be displayed before the sung phrase is heard." The obvious remedy is a deterministic identifier/position mapping of the kind Sonohara and Abbott supply. Comps (structural, event-driven) + Sonohara (numerical, position-driven) is a textbook KSR "combination of known elements according to known methods."
4. Predictable result / reasonable expectation of success. Both are fully enabled, deterministic, software-implementable rules for aligning segments. Nothing in either reference is unpredictable, and the combination yields no more than the expected sum of the parts (Anderson's-Black Rock). There is no teaching away: Sonohara does not disparage multiple files (it starts from them), and Comps does not disparage identifiers.

B. Combination 2: Sonohara + Comps (mirror configuration, for claim-drafting robustness)

If a tribunal were to hold that Comps's SYNCH commands do not map cleanly onto "each" segment (VideoLabs's best Comps argument was that intervening D₁/JPEGD/TEXTD fields mean not every content-related segment is output "together" with a partner), the mirror combination cures it: Sonohara as the primary reference supplies "reading, in order, said image data and said sound data … and at the same time synchronously outputting" for every unit-time pair, and Comps is cited for the proposition that a POSITA would have implemented this over two distinct files with a non-temporal stored rule — which is what converts an otherwise timestamp-based teaching into claims 20–21.

This is exactly how the two petitions together cover the field: the CAFC-affirmed Sonohara FWD takes claims 1, 3, 5, 9, 12, 13, 15; Comps takes claims 20–21 (and, in the Netflix FWD, everything else).

C. Combination 3: Abbott + Comps (or Sonohara) — claims 1, 3, 5, 9, 12, 13, 15, 20, 21

Abbott is independently strong on the two hardest limitations:

• Two distinct files. Video atom vs. audio atom is a cleaner "first data file / second data file" mapping than Sonohara's single composite file — the ground on which the Board rejected Unified's Sonohara/Ground mapping (VideoLabs persuaded the Board that Sonohara synchronizes segments within one file; see the POPR excerpt: "the use of a single data file instead of two is central to its alleged contribution to the art").
• Rule not based on a timestamp (claims 20–21). Abbott assigns audio frames to GOPs by atom-relative byte positions ("each of these index numbers points to the same atom-relative byte position that is the beginning of audio frame"), with no clock. That is a non-timestamp assignment rule.
• Rule in a separate file (claims 18–19). Abbott's index files ("a base atom index file," "an auxiliary atom index file") are maintained separately from the media atoms — a literal read on claims 18–19, which were not reached in either IPR.

Important caveat (this is the weakness to manage): in IPR2022-01086 the Board found Petitioner did not adequately show that Abbott discloses "each of the second data segments output together with an associated one [first segment]" — because Abbott's audio frames A1–A5 all map to the same byte position (the beginning of A1), a playback that honored the rule literally "would be repetitive and asynchronous" and the "video data would not be output 'sequentially' or 'chronologically.'" That is an evidentiary failure of one petition, not a holding that the reference is inapt for § 103. A new § 103 challenge should be built differently — with an expert declaration explaining (i) that the index-file mapping is a design choice of granularity (the reference elsewhere maps audio boundaries "that most closely match the Groups of Pictures boundaries"), and (ii) an explicit KSR rationale for selecting a one-to-one frame-to-GOP correspondence so that output is sequential — rather than relying on a bare anticipation mapping. Abbott's own stated purpose ("synchronizing media data," "minimize a time offset between audio and video data") supplies the rationale.

D. Combination 4: Any of the above + Panis [1] (gBSD) and/or XSLT — claim 6

Claim 6 (gBSD or XSLT segment generation) is the easiest claim in the patent to invalidate, because the specification itself concedes it:

• The '794 spec. states the first fragmentation description file "uses a gBSD syntax … being based on the XML language" and is "virtually identical with FIG. 9 from [1]" — i.e., the applicant admitted Panis [1] discloses the claimed gBSD segmentation.
• The spec. further states that "the extensible stylesheet language transformation (XSL Transformation or XSLT) represents another option for describing the fragmentation rules" and points to [1].
• Document [2] (Devillers) and [3] (Hellwagner) — the applicant's own MPEG-21 work products — describe gBSD-based adaptation for streaming and constrained environments, which is the '794 patent's stated field.

Under MPEP § 2129 and In re Fout, an applicant's statement in the specification that a technique is known is a binding admission usable as prior art. Motivation: gBSD was purpose-built for MPEG-21 resource adaptation and streaming (Panis, abstract); a POSITA assembling a synchronizer for MPEG-2/4/H.264 media (the exact codecs named at '794 spec. 4:55–62) would use the standardized descriptor it already had to delineate segment boundaries. Combining a known segmentation-description tool with a known synchronizer is the paradigm predictable combination.

E. Combination 5: Any of the above + Wan (US 2004/0024898) and/or BBL [4] / EP 1 494 430 A2 — claims 7, 8, 22

• Claim 7 (content-related marker determines which segments output first): Wan discloses content markers within a scene-description stream defining temporal relationships and scene boundaries. BBL [4] (cited in the '794 background as "a method by which a synchronization can be performed") uses content-descriptive addressing. EP 1 494 430 A2 (Samsung, cited on the face of the '794 patent) describes segment metadata with segment IDs, key-frame positions and group intervals (TV-Anytime) — i.e., content markers used to locate and reproduce a selected segment.
• Claim 8 (scene change / speaker change): Wan's scene boundaries; EP 1 494 430's TV-Anytime group types expressly include "shots," "highlights/events," "insertionPoints" — i.e., scene-level content markers. The '794 spec. identifies precisely "a scene change in an image sequence or a change of speaker in a speech sequence" as the marker's use.
• Claim 22 (markers order the segments, then the rule pairs them): the combination of Comps's SYNCH rule with Wan's or EP 1 494 430's markers yields the claim-22 sequence directly. This is the ground the PTAB credited in IPR2023-00891 (Comps + Wan for claims 6–8 and 22).

Motivation: Wan and EP 1 494 430 address the same "find and play the right segment" problem the '794 patent's markers solve, in the same field, and both are cited on the face of the '794 patent; the combination is a mere substitution of one known addressing scheme (byte offsets / identifiers) for another (content markers), which KSR treats as obvious absent a teaching away or unexpected result.

F. Combination 6: Any of the above + EP 1 494 430 A2 (Samsung) and/or Ahn (US 2004/0098398) — claims 13, 14

Claims 13–14 extend the method to metadata, in particular XML-encoded metadata. EP 1 494 430 A2 is directly on point: it synchronizes multimedia content with segment metadata received over different channels, and expressly states that "TV-Anytime Metadata … expressed in an XML format" defines the metadata for a segment and a segment group. Ahn supplies a second, independent teaching of synchronizing XML-source metadata with multimedia content. Motivation: the '794 spec. itself contemplates XML-encoded metadata (data file "also known as media data MID" with XML metadata MTD), and document [2] (Devillers) is the applicant's own "metadata streaming" use-case paper — establishing that combining XML metadata synchronization with media synchronization was an identified, active field need. The substitution of an XML rule document (the '794 patent's Fig. 4 StreamingInstructions document) for a binary index file is an obvious format choice under KSR.

G. Combination 7: Any of the above + Sone (US 2001/0052943 A1), EP 0 753 954 A2, or Li et al. — claims 4, 10, 11

Claims 4, 10 and 11 require more than one second data file synchronized by a combined assignment rule.

• Comps literally discloses this: one first file (Track 1, sound) plus two second files (Track 2, pictures; Track 3, text), all coordinated by the combined set {SYNCH1, SYNCH2, SYNCH3}. This is claim 4 without any combination at all.
• Sone ("Multimedia system with synchronization of music and image tracks") and EP 0 753 954 A2 (high-rate isochronous data in MPEG-2 streams) provide independent corroboration.
• Li et al. (ICC '94) teaches segment-based delivery with synchronization control — supporting the proposition that segment-level, multi-stream synchronization was well known in the art a decade before the priority date.

Motivation: adding a third data type (e.g., text/subtitles, or a second audio track) to a two-track synchronizer is the classic KSR "combination of known elements according to known methods" with no change in principle of operation, and yields no unexpected result.

V. Why the applicant's prosecution distinction does not survive the full prior-art universe

The prosecution record shows the applicant distinguished Shin and Rosenau on the ground that "these references consider time information for synchronization, rather than using an assignment rule for synchronization." (Board FWD, as quoted in CAFC 24-1890.) That argument builds an implicit "no-timestamps" limitation into claim 1 — and it fails for three independent reasons:

1. The claim language was never amended to say it. Only dependent claims 20–21 recite "not based on a timestamp." The Board refused the negative limitation, and the CAFC held the contrary argument forfeited. Under KSR and In re De Blauwe, claims are given their broadest reasonable construction consistent with the specification, not the applicant's attorney argument.
2. Comps is not a "time information" reference. Comps states that its commands "are not temporal commands, as in the prior art." So the very distinction the applicant drew over Shin/Rosenau does not distinguish Comps.
3. Abbott is not a "time information" reference. Abbott assigns by byte position, not by clock.

VI. Claim-by-claim mapping summary

Legend: S = Sonohara '656; C = Comps '782; A = Abbott '933 family.

VII. Secondary considerations

No secondary consideration saves any claim:

• No unexpected results. The '794 specification alleges none, and none was argued in either IPR. The patent's own Figure 1 example (A=2 → pair every second video segment) is a routine engineering choice within the disclosed design space (In re Aller).
• No long-felt need / failure of others. The field was crowded and the need was met repeatedly and early (Best 1986; Comps 2000/2002; Abbott 1998/2003; Sonohara 1997) — the opposite of a long-felt but unsolved need.
• Industry praise / licensing. The VideoLabs licensing and enforcement campaign (Netflix, Roku, Apple, Amazon, Hisense, TCL, Dell, Microsoft — see the litigation section) is not probative of non-obviousness absent a nexus between the licensing and the claimed subject matter (In re GPAC; Ormco Corp. v. Align Tech.). Given that the PTAB has held every claim unpatentable at least once, any nexus theory would be untenable.
• Commercial success. No evidence of a Siemens product embodying the claims (see the assignment section).

VIII. Robustness assessment and the honest counter-arguments

Where the art is strongest:

1. Comps is the single most complete reference. It alone reaches claims 20–21 — the only claims that survived Sonohara — because it is expressly non-temporal.
2. Claim 6 is effectively self-invalidating: the specification admits gBSD/XSLT segmentation is known and that its own description file is "virtually identical with FIG. 9 from [1]."
3. Claims 18–19 (rule in a separate file) were never reached in the Unified IPR and are squarely met by Abbott's index files.
4. Claims 4, 10, 11 are met by Comps's three-track architecture alone.

Where a defendant must be careful:

1. Do not lead with Abbott for claim 1. The Board accepted VideoLabs's argument that a literal application of Abbott's A1–A5→A1 mapping would be "repetitive and asynchronous," defeating the "each … output together" and "sequentially"/"chronological" limitations. Frame Abbott as a § 103 reference with an explicit rationale (granularity selection to match GOP boundaries, per Abbott's own "most closely match" language), not as an anticipation reference for claim 1.
2. Comps's disputed-subset issue. VideoLabs argued below that Comps maps only a subset of events and skips others, so not "each" content-related segment is output "together." The Board found this "may have some merit" for the JPEG mapping but held the record sufficient at institution as to the text fields (TEXT1/TEXT2) synchronized with notes Nf₃/Nf₄. A new challenge should therefore ground the claim-1 mapping on the concurrently-output important events (TEXT₁↔Nf₃, TEXT₂↔Nf₄), not on JPEG framings.
3. Sonohara is unusable against claims 20–21. The Board found Sonohara's number/duration header information "correspond[s] to a timestamp." Pair Sonohara (or Comps) with a genuinely non-temporal reference (Comps itself, or Abbott's byte positions) for those claims.
4. Verify the reference identifiers. Comps = U.S. 7,386,782 B2 ✔ (confirmed against the front page). Sonohara = U.S. 5,627,656 ✔ (title and claim text confirmed). Abbott's number (6,654,933) and the Wan/Ahn application numbers are taken from third-party IPR summaries and should be confirmed against PTAB exhibits EX1004/EX1006/EX1008 before filing.

IX. Bottom line

There is no claim of the '794 patent that survives a properly framed § 103 challenge.

• Comps (U.S. 7,386,782) — alone or with Sonohara — renders claims 1–21 obvious, and because its SYNCH commands are expressly "not temporal commands," it reaches claims 20–21 that defeated Unified's Sonohara-only attack.
• Abbott (U.S. 6,654,933 family) renders claims 1, 3, 5, 9, 12, 13, 15, 18, 19, 20 and 21 obvious when its index-file mapping is presented with a KSR rationale rather than as anticipation.
• Panis [1] (gBSD) plus XSLT plus the applicant's own admissions render claim 6 obvious as a matter of the specification's own text.
• Comps + Wan (and/or EP 1 494 430 A2 / BBL [4]) render claims 7, 8 and 22 obvious via content markers and scene-change identification.
• EP 1 494 430 A2 (Samsung) + Ahn render claims 13–14 (XML metadata) obvious.
• Comps + Sone / EP 0 753 954 A2 render claims 4, 10, 11 (plurality of second files, combined rule) obvious.

The Federal Circuit's February 20, 2026 nonprecedential affirmance in No. 24-1890 has already fixed the constructions that make these combinations work ("assignment rule" unconstrained; "content-related data segments" = syntactically meaningful segments), and the Netflix FWD of January 7, 2025 (all 22 claims unpatentable, on appeal in No. 25-1525) already applied them across the full claim set. A defendant or demand-letter recipient should not need a third IPR: the existing record, plus a Comps-based § 103 theory for claims 20–21, is dispositive.

Sources cited

• Full text of US 8,605,794 B2 (Google Patents) — supplied as authoritative.
• CAFC No. 24-1890, opinion of Feb. 20, 2026.
• PTAB FWD, IPR2022-01086, Unified Patents, LLC v. VL Collective IP LLC (Dec. 18, 2023); PTAB FWD, IPR2023-00891, Netflix, Inc. v. VL Collective IP LLC (Jan. 7, 2025) — via PTACTS/PTAB E2E.
• Sonohara, U.S. Pat. No. 5,627,656 — Google Patents.
• Comps, U.S. Pat. No. 7,386,782 B2 — Google Patents; US 2004/0098365 A1.
• EP 1 494 430 A2 (Samsung) — EPO Global Patent Index; U.S. counterpart US 7,428,696.
• Abbott, U.S. Pat. No. 7,272,780 / 6,654,933 family — SumoBrain.
• Willkie Farr & Gallagher, release on the Netflix IPR victory (Jan. 9, 2025).

Caveats: reference numbers for Abbott, Wan, and Ahn derive from third-party IPR summaries rather than the primary exhibits and are flagged for verification. The Comps/Wan/Ahn ground structure is as summarized from a machine-generated index of Netflix's Ex. 1007 and should be confirmed against the FWD itself before being cited in any filing.