US 11342998

US Patent 11,342,998 — Verification Summary

I confirmed the specific patent US 11,342,998 B2 (application US 17/481,904) via Google Patents and secondary litigation sources. No similar-number patents were substituted.

Bibliographic data

Abstract (verbatim from the patent)

"In a system for converting digital data into a modulated optical signal, an electrically controllable device having M actuating electrodes provides and optical signal that is modulated in response to binary voltages applied to the actuating electrodes. A digital-to-digital converter provides a mapping of input data words to binary actuation vectors for M bits and supplies the binary actuation vectors as M bits of binary actuation voltages to the M actuating electrodes, where M is larger than the number of bits in each input data word. The digital-to-digital converter maps each digital input data word to a binary actuation vector by selecting a binary actuation vector from a subset of binary actuation vectors available to represent each of the input data words."

Plain-language overview of the independent claims

The claims excerpt I have contains the full text of claim 1 and claim 16 as independent claims (the excerpt truncates mid-claim at claim 22 of 63). Both are drafted in near-identical structural form, differing in the direction of the input-value progression:

• Claim 1 (increasing input values): An optical modulation system with (a) an input for N digital input data bits, (b) an input optical signal, and (c) a modulator that modulates the input optical signal in response to the N bits, producing modulated optical outputs for transmission over optical fiber(s). A digital-to-digital mapping converts the N input bits into M digital output bits associated with voltage values, and the optical signal is modulated based on those voltages. The mapping specifies, for each possible digital input value, a corresponding digital output value. The defining limitation: within the mapping, for a first subset of successively increasing input values, the deltas between the numerical values of successive corresponding digital outputs decrease; and for a second subset of successively increasing input values, those deltas increase. In plain terms: the mapping is deliberately non-uniform/nonlinear — the step sizes between successive output codes shrink in some input ranges and grow in others — which is how the design compensates for the modulator's intrinsic cosine-like (MZI) nonlinearity and approximates a linear transfer function.

• Claim 16 (decreasing input values): Identical system structure (input for N bits, input optical signal, modulator producing fiber-transmittable outputs, digital-to-digital mapping of N input bits to M output bits/voltage values), with the same delta-based limitation stated for successively decreasing input values: deltas between successive digital outputs decrease for a first subset and increase for a second subset.

Uncertainty note: Because the supplied claims excerpt truncates at claim 22, I cannot fully verify the text of any additional independent claims (claims 23–63). The record indicates the '998 patent contains 63 claims, and litigation documents reference claim groups including claims 45–47, 49–54, 58, and 61–63, suggesting additional independent claims exist in the latter portion of the set. I do not have authoritative text for those and cannot responsibly summarize them.

CAFC / litigation status (2024–2026 dockets)

• PTAB IPR2022-01283 (Cisco Systems, Inc. et al. v. Ramot at Tel Aviv University Ltd.): filed July 20, 2022, challenging claims of US 11,342,998; institution Feb. 22, 2023; Final Written Decision issued Feb. 20, 2024 (APJ Monica S. Ullagaddi, with Ogden and Repko).
• CAFC Appeal 2024-1728 (Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.): appeal from IPR2022-01283, docketed April 23, 2024. This was consolidated/heard with appeals 2024-1726 and 2024-1727 (from IPR2022-00575 and IPR2022-00576).
• CAFC April 13, 2026 judgment: Per curiam (Moore, Chief Judge; Mayer and Lourie, Circuit Judges) — AFFIRMED under Fed. Cir. R. 36 (nonprecedential), entered April 13, 2026. This 2026 disposition affirmatively resolves the PTAB appeals in Ramot's favor.
• District of Delaware (Ramot v. Cisco, filed May 24, 2022, the issue date): Judge Gregory B. Williams, Nov. 12, 2024, found claims 45–47 and 49–54 indefinite, and declared claims 1, 4, 6–15, 58, and 61–63 not indefinite.
• CAFC case 26-1277: The Google Patents record lists a further Federal Circuit case number 26-1277 associated with this patent family, but my searches returned no docket detail for that specific number, so I cannot confirm its subject matter or status. Treat that as unverified.

Confidence notes

• Core bibliographic data (title, assignee, inventors, dates, abstract) is confirmed by both the supplied Google Patents full text and independent search results (Justia, RPX Insight, PatentLeaderboard, Patexia, CAFC opinion PDF).
• The April 13, 2026 CAFC Rule 36 affirmance is confirmed by the CAFC's own published judgment document (cafc.uscourts.gov), which is the most current authoritative litigation data available for this patent.

I'll search for litigation involving US Patent 11342998. Let me run several targeted searches.

I found one district court case. The patent text also references CAFC appeals and an IPR. Let me search for those specifically.

I'm building a clear picture. Let me search for the Delaware declaratory judgment cases and the CAFC appeals.

I've gathered sufficient information from live sources. Here is the known litigation involving US Patent 11,342,998 (US11342998B2, "Linearized optical digital-to-analog modulator," assigned to Ramot at Tel Aviv University Ltd.) — note that the "application number 11342998" hit in one search result (US-7223979-B1) is a different patent and is not relevant.

District Court Litigation

1. Ramot at Tel Aviv University Ltd. v. Cisco Systems, Inc.

• Plaintiff: Ramot at Tel Aviv University Ltd.
• Defendant: Cisco Systems, Inc.
• Jurisdiction: U.S. District Court for the Eastern District of Texas, Marshall Division (Judge Rodney Gilstrap)
• Case No.: 2:22-cv-00168-JRG
• Filing date: May 24, 2022 (12:23 a.m. CDT, the day the '998 patent issued)
• Nature: Infringement of the '998 patent via optical networking transceiver modules, line cards, and components (including Acacia products)
• Status/Outcome: Cisco's motion to transfer under the first-to-file rule was granted; the case was transferred to the District of Delaware on January 5, 2023 (docket reflects transfer; Patexia labels the termination "Dismissed - Other," which reflects the transfer). It was then effectively coordinated with the Delaware DJ action below.

2. Cisco Systems, Inc. and Acacia Communications, Inc. v. Ramot at Tel Aviv University Ltd.

• Plaintiffs: Cisco Systems, Inc. and Acacia Communications, Inc. (Cisco's subsidiary)
• Defendant: Ramot at Tel Aviv University Ltd.
• Jurisdiction: U.S. District Court for the District of Delaware (Judge Gregory B. Williams)
• Case No.: 1:22-cv-00674-GBW (related/coordinated with C.A. No. 21-1365-GBW, which concerns family patent 11,133,872)
• Filing date: May 24, 2022 (12:01 a.m. EDT — same day the patent issued). The court later noted the plaintiffs "abused the Declaratory Judgment Act by running to the courthouse at 12:01 a.m. on the day that Ramot's patent was released."
• Nature: Declaratory judgment of noninfringement of the '998 patent (accused products included Acacia AC1200/AC400, pluggable transceivers such as CIM8, CFP2, OSFP, QSFP-DD, DSP ASICs, silicon photonics, line cards, and routers)
• Status: Active/ongoing. On November 12, 2024, Judge Williams issued a claim-construction and indefiniteness ruling: claims 45–47 and 49–54 were held indefinite, while claims 1, 4, 6–15, 58, and 61–63 were found not indefinite (MLex; Studicata summary).

PTAB (IPR) Proceedings

3. Cisco Systems, Inc. et al. v. Ramot at Tel Aviv University Ltd. et al.

• Petitioners: Cisco Systems, Inc. (et al.)
• Patent Owner: Ramot at Tel Aviv University Ltd. (et al.)
• Jurisdiction: Patent Trial and Appeal Board (PTAB)
• Case No.: IPR2022-01283 (challenging the '998 patent)
• Filing date: July 20, 2022
• Institution decision: February 22, 2023
• Final Written Decision: February 20, 2024
• Panel: Judges Monica S. Ullagaddi (author of FWD), Christopher L. Ogden, Jason M. Repko
• Status: Final Written Decision issued; appealed to the Federal Circuit (see below).

Federal Circuit Appeals

4. Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd. — Appeal Nos. 2024-1726, 2024-1727, 2024-1728

• Jurisdiction: U.S. Court of Appeals for the Federal Circuit
• Case No.: 2024-1728 (appeal from IPR2022-01283, the '998 patent IPR; consolidated with 2024-1726 and 2024-1727, which are appeals from IPR2022-00575 and IPR2022-00576 involving family patent 11,133,872)
• Filing date: 2024 (appeal docketed following the PTAB FWD)
• Status/Outcome: Judgment entered April 13, 2026, per curiam (Chief Judge Moore, Judges Mayer and Lourie): AFFIRMED under Fed. Cir. R. 36 (nonprecedential). The PTAB's final written decisions, including in IPR2022-01283, were affirmed.

5. CAFC Case 26-1277

• The Google Patents litigation record for this patent also lists a CAFC case, 26-1277 (filed 2026, likely an appeal from the Delaware DJ action 1:22-cv-00674). I was unable to independently verify the parties, docket, or status of this appeal within my search limits, so treat that entry as referenced in the patent record but unconfirmed.

Related (but distinct) family litigation — for context only

Earlier cases involve other patents in the same "Linearized optical digital-to-analog modulator" family (U.S. 10,270,535; 10,033,465; 11,133,872), not the '998 patent itself: Ramot v. Cisco, 2:19-cv-225-JRG (E.D. Tex., filed June 12, 2019; stayed pending ex parte reexamination); Ramot v. Acacia, 21-295-GBW (D. Del., filed Feb. 26, 2021; stayed pending reexam); and Cisco/Acacia v. Ramot, 21-1365-GBW (D. Del., filed Sept. 28, 2021, re the '872 patent). These should not be counted as '998 patent litigation.

Bottom line: US 11342998 has been involved in (i) Ramot's infringement suit against Cisco (E.D. Tex., transferred to D. Del.), (ii) Cisco/Acacia's DJ action in Delaware (with a partial-indefiniteness ruling in Cisco's favor in Nov. 2024), (iii) IPR2022-01283 at the PTAB (Final Written Decision Feb. 20, 2024, affirmed by the CAFC on April 13, 2026), and (iv) associated CAFC appeals (2024-1728 and a referenced 26-1277). The Delaware DJ action appears to remain the principal ongoing proceeding.

Proceedings overview

Web research confirms one AIA trial proceeding against US 11,342,998 — IPR2022-01283 (Cisco Systems, Inc. et al. v. Ramot at Tel Aviv University Ltd.) — which challenged all 63 claims, proceeded to a Final Written Decision (2024-02-20), and was affirmed by the Federal Circuit under Rule 36 on 2026-04-13. The USPTO ODP structured block in this prompt lists zero AIA proceedings for the patent — a data-lag discrepancy I flag explicitly: the ODP has not yet ingested IPR2022-01283, but the proceeding is real and confirmed by Patexia, the CAFC's own judgment docket, and PTAB paper citations. The bottom line for a defendant: the only IPR filed against this patent was lost by the petitioner at the CAFC, the patent owner prevailed on appeal, and there are no active PTAB proceedings — this patent is "hardened" against the art Cisco ran, though claims 45–47 and 49–54 were separately held indefinite in Delaware (a § 112 defense, not a PTAB cancellation).

IPR2022-01283 — Cisco Systems, Inc. et al. v. Ramot at Tel Aviv University Ltd.

• Type: Inter Partes Review
• Filed: 2022-07-20
• Status: Final Written Decision issued; appeal to the Federal Circuit affirmed — proceeding concluded, no claims confirmed canceled on the public record I can access (see caveat under FWD below)
• Judge panel: Monica S. Ullagaddi (author of the Final Written Decision), Christopher L. Ogden, Jason M. Repko
• Petition grounds: All 63 claims challenged under § 102/§ 103. I could not retrieve the complete ground-by-ground art list from my sources. A PTAB paper citation in a later decision confirms the petition relied on at least three references — one used in an earlier Office action rejection, a second listed in an IDS, and a third reference that had not been before the examiner (which is why the Board declined to exercise § 325(d) discretion). I will not name specific references because I cannot verify them.
• Institution decision: Instituted — 2023-02-22 (Paper 12). The Board declined to exercise discretionary denial under § 325(d), reasoning that because the combination applied a third reference not before the examiner, the petition presented a materially new ground.
• Final Written Decision: Issued 2024-02-20. ⚠️ Claim-level outcome not confirmed: my accessible sources (Patexia summary, CAFC judgment, secondary press) do not disclose the FWD's claim-by-claim table — I will not invent one. What is clear: Cisco appealed the FWD, which means the outcome was adverse to Cisco (a petitioner that wins cancellation does not appeal), and the CAFC then affirmed. Corroborating inference: in the parallel Delaware case (below), the district court treated claims 1, 4, 6–15, 58, and 61–63 as live and adjudicated their definiteness in November 2024 — conduct consistent with those claims having survived the IPR. That is an inference, not a confirmed FWD holding; pull Paper 40 (or the FWD paper number) from PTAB E2E for the authoritative claim table.
• Settlement / termination: No settlement indicated on the public record; the case ran to a Final Written Decision and appeal.
• Appeal: CAFC No. 2024-1728 (Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.), consolidated and heard with 2024-1726 and 2024-1727 (appeals from IPR2022-00575 and IPR2022-00576, which are IPRs Cisco filed against other patents in the same Ramot optical-DAC family). Argued by Angela M. Oliver (Haynes and Boone) for Cisco and Denise Marie De Mory (Bunsow De Mory) for Ramot. Disposition: AFFIRMED per curiam under Fed. Cir. R. 36 (nonprecedential), entered 2026-04-13 — panel of Chief Judge Moore and Circuit Judges Mayer and Lourie. Sources: CAFC judgment PDF, Justia mirror.
• Defensive value: For a defendant facing assertion of the '998 patent today, this proceeding cuts both ways and must be handled precisely. (1) Cisco — and parties in privity with Cisco — are estopped under § 315(e)(2) from re-running the § 102/§ 103 grounds raised (or reasonably available) in this IPR, and the art Cisco chose is now "tried and failed" at the CAFC; do not reuse it. (2) Because the Board's decision was affirmed, no claims are confirmed canceled; a defendant should not walk into a negotiation claiming "the patent was gutted in IPR" — that is false on the current record. (3) The strongest available invalidity ammunition is the Delaware indefiniteness ruling, not the IPR: on 2024-11-12, Judge Gregory B. Williams (D. Del.) held claims 45–47 and 49–54 indefinite for the ambiguous "wherein the N bits of the N bit digital input data word are mapped" limitation, while declaring claims 1, 4, 6–15, 58, and 61–63 not indefinite (MLex summary). If the demand letter cites claims 45–47 or 49–54, that is your lever.

Strategic summary

Canceled vs. sustained vs. untested. On the public record I can verify: no claims of US 11,342,998 are confirmed canceled by the PTAB. All 63 claims were challenged in IPR2022-01283; the FWD (2024-02-20) went against Cisco hard enough that it appealed, and the CAFC affirmed (2026-04-13). The most reasonable reading — corroborated by the Delaware court later treating claims 1, 4, 6–15, 58, and 61–63 as live — is that the Board sustained at least those claims as patentable. Claims 45–47 and 49–54 are effectively dead in the Delaware litigation via § 112 indefiniteness (a district-court holding, not a PTAB cancellation, and thus not an estoppel-generating event for anyone else). Claims 16–44, 48, 55–57, and 59–60 are, from my sources, untested or unreported — the Delaware order addressed only the asserted subset, and I could not confirm the FWD's treatment of the remainder. Any party preparing a validity attack must pull the FWD from PTAB E2E before characterizing any claim as dead.

Estoppel landscape. § 315(e)(2) bars Cisco and its privies from raising in district court any § 102/§ 103 ground they raised or reasonably could have raised in IPR2022-01283. A new defendant who is not in privity with Cisco faces no such bar and remains free to: (i) run § 102/§ 103 on different prior art never before the Board; (ii) raise § 101 (subject-matter eligibility — untouched by this IPR) and § 112 (indefiniteness, written description, enablement — only partially litigated in Delaware) theories; and (iii) cite the Delaware indefiniteness holding on claims 45–47 and 49–54 as binding precedent within that case. Practically, the cleanest prior-art runway for a new challenger is the digital-to-analog optical modulator art not used by Cisco — e.g., the Papuchon multi-electrode MZI line (U.S. Pat. No. 4,288,785) and the Yacoubian/Leven DAC literature the patent itself discusses in its Background — provided those were not already in Cisco's petition (verify against the FWD before relying on them).

Pattern signals. Cisco is a repeat challenger of the Ramot optical-DAC family: in addition to IPR2022-01283, it filed IPR2022-00575 and IPR2022-00576 against related family patents (both also affirmed against Cisco on 2026-04-13), and Cisco's earlier attempts to IPR two Ramot patents in the ~2020 timeframe were denied institution, with the Federal Circuit denying mandamus (Patently-O). The patent owner (Ramot) has defended every challenge through appeal and has yet to lose a final decision — an aggressive, well-resourced patent owner. Note on the "Unified Patents" appearance in the Google Patents feed: that is the data-licensing attribution for the PTAB case record, not a Unified Patents petition — do not confuse the data source with the petitioner. The CAFC case 26-1277 listed on Google Patents remains unverified (no docket detail retrievable); given the 2026 case number it may be a further appeal from the Delaware litigation, but I will not characterize it without a docket.

Recommended next steps

1. Pull the FWD before making any claim-level representation. Retrieve IPR2022-01283 (Final Written Decision, ~Paper 40) from PTAB E2E or the Unified Patents case portal and extract the claim-by-claim table. Everything in an invalidity memo turns on whether the Board canceled specific claims (e.g., the independent claims 1 and 16) or sustained them — and my sources do not disclose that table. Do not rely on my inference as fact.
2. If the demand letter cites claims 45–47 or 49–54, lead with Delaware. The 2024-11-12 indefiniteness ruling (Ramot v. Cisco / Cisco v. Ramot, D. Del., Judge Williams) is a public, citable holding that those claims cannot be enforced as written — quote the "wherein the N bits of the N bit digital input data word are mapped" reasoning (MLex).
3. No active PTAB proceedings exist, so there are no institution-deadline or trial-stage milestones to track. If you (a non-Cisco defendant) want your own IPR, watch the § 315(b) one-year bar running from service of the complaint, and build the petition on art not used by Cisco to avoid a déjà-vu denial and to sidestep any argument that the ground was already adjudicated. Given Ramot's 3–0 appellate record, a marginal petition is a poor investment — only file if you have genuinely stronger or materially different prior art, or a § 101/§ 112 theory that PTAB cannot reach (which points you back to district court rather than IPR).

I'll start by searching for the recorded assignment history and any ownership changes for US 11342998, plus cross-reference litigation/asserter databases.

The searches confirm Ramot owns the patent and is the direct litigant, but I need the specific USPTO assignment records (reel/frame, correspondents). Let me dig for those.

I've reached the tool-call limit for this task, so I'll answer from the evidence gathered across Google Patents, court records, Unified Patents, Patexia, and Justia. Here is the ownership-chain reconstruction for US 11,342,998.

Inventors

All three inventors are named on the patent and all three were affiliated with Tel Aviv University's School of Electrical Engineering at the time of the invention — confirmed by Ramot's own complaint in Ramot at Tel Aviv University Ltd. v. Cisco Systems, Inc., No. 14-1018-JRG (E.D. Tex., filed Nov. 5, 2014): "Each of the inventors of the Patents-in-Suit was affiliated with TAU's School of Electrical Engineering during the relevant time period, and each of the inventors assigned his rights to the Patents-in-Suit to Ramot."

Pattern note: No unusual departure pattern. All three inventors remain tied to the same assignee family ~18 years after the 2007 priority date — Ehrlichman, Amrani, and Ruschin continue to appear as inventors on Ramot filings through 2025 (e.g., US 19/274,676, filed 2025-07-21, published 2025-11-13; US 12,388,533, issued 2025-08-12; US 12,191,912, issued 2025-01-07). This is the opposite of the "inventors flee, then portfolio fire-sale" pattern.

Original assignee

• Entity: Ramot at Tel Aviv University Ltd. (Tel Aviv, IL) — the named assignee on the issued patent and the "Business Engagement Center" / technology-transfer arm of Tel Aviv University (per the D. Del. record in Cisco Systems, Inc. v. Ramot, No. 21-1365-GBW: "Ramot is the Business Engagement Center of Tel Aviv University ('TAU') and acts as the University's liaison to industry… protecting the discoveries with IP and working jointly with industry…").
• Products: Does not ship products embodying the claims. Ramot is a licensing/commercialization office (per its 2014 complaint: manages 2,200+ patents/applications worldwide; holds equity in 100+ startups).
• Primary business: University technology transfer and patent licensing.
• Current status: Active and operating; still the current assignee per Google Patents and Unified Patents ("Parent Company: Ramot At Tel Aviv University Ltd", original assignee = current assignee).

Assignment timeline

I could not directly query the USPTO Assignment Center (assignmentcenter.uspto.gov) from this environment, and no search result exposed reel/frame numbers or the correspondent of record for this patent. I therefore cannot responsibly state specific reel/frame values or attorney names — none were retrieved. Do not treat any reel/frame below as verified; they are not available in my retrieved sources.

What the record does show is that only one assignment event exists for this patent, and it is the routine inventor-to-assignee conveyance:

• Executed ~2007–2008 (original family) / recorded 2021-09-27 — Google Patents legal event, "Assigned to RAMOT AT TEL AVIV UNIVERSITY LTD."
  • Conveyance: ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)
  • Assignor: RUSCHIN, SHLOMO; AMRANI, OFER; EHRLICHMAN, YOSSEF (inventors)
  • Assignee: RAMOT AT TEL AVIV UNIVERSITY LTD.
  • Correspondent: Not retrievable from my sources. (Flag: no correspondent data available; cannot assess repeat-correspondent signal.)
  • Context: Standard inventor-to-employer-assignee conveyance recorded when the continuation (17/481,904) entered the USPTO system; the same inventors-to-Ramot chain underlies the whole continuation family back to the 2007 provisional.

No post-issuance assignments are recorded. The Delaware district court confirmed this directly in its Nov. 14, 2022 order: "Ramot owns by assignment the '872 and '998 patents" (No. 21-1365-GBW, D.I. 51). There is no evidence of any transfer to an LLC, holding company, licensing vehicle, or defensive aggregator at any point in this patent's life.

Timeline diagram

timeline
    title Ownership of US 11342998
    2007 : Priority filed by TAU inventors
    2011 : First family patent issued
    2021 : Continuation filed
         : Inventors assign to Ramot
    2022 : Patent issued
         : Cisco DJ suit filed same day
         : IPR filed by Cisco
    2024 : PTAB final written decision
         : Delaware definiteness ruling
    2026 : CAFC affirms PTAB

NPE / troll-pattern signals

1. Shell-entity transfer — Not present. The patent never left Ramot. No LLC, no "IP / Patents / Licensing / Holdings" vehicle, no registered-agent address, no single-purpose entity in the chain. The only conveyance is inventors → Ramot (recorded 2021-09-27 per Google Patents legal events).

2. Known asserter in the chain — Not present (with one caveat). Ramot is a repeat litigant of this family (E.D. Tex. 2014 and 2019; D. Del. 2021 and 2022; defendant in IPR2022-01283 and CAFC 24-1726/24-1727/24-1728) — so it is a high-frequency plaintiff in the broad sense. But it does not match any entity on the specified NPE lists (Acacia, Marathon, IV, Wi-LAN, Conversant, etc.), and it is a university TTO asserting its own faculty's patents, not a purchased-portfolio troll. Unified Patents' own portal tracks the litigation without flagging Ramot as a classic NPE.

3. Repeat correspondent across the chain — Unclear. The chain has only one link, so there is no recurrence to establish, and I could not retrieve the correspondent of record from the Assignment Center. Marked unclear solely due to missing data, not due to any positive finding.

4. Cascading transfers — Not present. Exactly one assignment, no chained LLCs, no rapid-fire transfers.

5. Pre-litigation transfer — Not present. The only assignment (recorded 2021-09-27) predates the first '998 litigation (May 24, 2022 — Cisco's D. Del. DJ action filed at 12:01 a.m. on issue day; Ramot's E.D. Tex. suit 2:22-cv-00168 filed at 1:23 a.m. the same day) by ~8 months, and it is the original inventors-to-Ramot conveyance, not a transfer to an asserting vehicle arranged for standing or venue.

6. Bankruptcy fire-sale — Not present. Ramot is solvent and operating; no bankruptcy anywhere in the chain.

7. Privateering — Not present. No evidence Ramot asserts on behalf of any operating company against competitors; Ramot asserts in its own name as TAU's TTO.

8. Defensive aggregator (anti-NPE) — Not present. Chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN.

Verdict

Insufficient data (only the original assignment exists; no reel/frame or correspondent details retrievable from my sources).

The assignment record shows a single, clean inventors → Ramot conveyance (recorded 2021-09-27 per Google Patents; "Ramot owns by assignment…" per D. Del. No. 21-1365-GBW D.I. 51), and zero post-issuance transfers. None of the eight troll-pattern signals are affirmatively present. The patent is held and directly asserted by a university technology-transfer office that practices no claims — a non-practicing entity in the broadest sense, but not a shell-NPE chain of the kind this analysis is designed to detect, and I lack the Assignment Center reel/frame data needed to test the correspondent-recurrence signal. For verification: https://assignmentcenter.uspto.gov/ (search Patent No. 11342998) and https://assignment.uspto.gov/patent/index.html.

Caveat on my confidence: I could not reach the USPTO Assignment Center directly and no retrieved source exposed reel/frame numbers or the correspondent of record. If the Assignment Center shows additional post-issuance recordings (e.g., a security interest or a transfer to a licensing entity), the verdict should be revisited. Based on everything retrieved — Google Patents legal events, the Delaware court's ownership finding, Unified Patents' family data, and the complete absence of any transfer in the litigation record — the working conclusion is that Ramot at Tel Aviv University Ltd. has owned this patent continuously since issuance, and the litigation posture is direct assertion by a university TTO, not a troll chain.

I was unable to retrieve the USPTO/Google Patents "References Cited" page for US 11,342,998 in full before hitting my search limit, so the list below is compiled from (i) the patent's own specification, which names the prior art it addresses, and (ii) the family record (the same citations carry through the continuation chain — e.g., the US 8,044,835 page lists the same non-patent citations). I flag every item's verification status explicitly rather than fabricate a complete examiner citation list.

Prior Art Analysis — US 11,342,998 B2 ("Linearized optical digital-to-analog modulator")

Methodology note

The USPTO database search for the literal number 11342998 returns the correct patent (US 11,342,998 B2, app. 17/481,904). Note that application number 11/342,998 belongs to an unrelated patent (US 7,223,979 B1, "Radiation Dosimeter System," Aerospace Corp.) — I have excluded that false positive per your instruction not to auto-correct identifiers.

I could not fully enumerate the examiner's formal "References Cited" section for the '998 patent. The references below are those affirmatively identified from the patent text itself (Background section) and from the family's non-patent citation record (Google Patents page of family predecessor US 8,044,835 B2). The IPR2022-01283 petition grounds were referenced in docket summaries but their specific asserted references were not retrievable in my searches; I flag that gap.

A. U.S. Patent References (identified from the '998 specification)

1. US 4,288,785 A — Papuchon et al.

• Full citation: M. Papuchon et al., "Optical digital-to-analogue converter," US Patent 4,288,785, filed (assumed 1979/1980), granted September 15, 1981 (assignee: Thomson-CSF). The specification describes it as "proposed many years ago by Papuchon et al."
• Brief description: The foundational multi-electrode Mach-Zehnder interferometer (MZI) digital-to-analog converter. A single MZI modulator has a segmented electrode whose sectioning lengths follow a conventional power-of-two digital sequence (e.g., 0.5, 0.25, 0.125, 0.0625), with each electrode driven by one bit of the digital input word. The patent expressly criticizes this design: it "did not solve the non-linearity problem, and thus suffered from severe limitation in the dynamic range, and subsequently the attainable resolution" — the cosine transfer function of the MZI corrupts linearity when electrodes are driven directly by the raw bits.
• § 102 anticipation potential: This is the closest structural prior art to the '998 patent's basic system (N-bit input → M actuating electrodes on an MZI → intensity-modulated optical output). It would potentially anticipate any claim that recites only the generic structure: input for N digital bits, an electrically controllable intensity modulator with M actuating electrodes, and electrode actuation in response to the input word (e.g., the structural preamble of claims 1 and 16, and possibly dependent claims 4, 5, 7–11 that add only "electrodes driven by voltages," "waveguide branches joined," "phase/amplitude modulation," "MZI-based modulators," and "optical signal source").
• However, it does NOT disclose the defining limitation of claims 1 and 16 — the digital-to-digital mapping in which deltas between successive digital output values decrease over a first subset of input values and increase over a second subset (i.e., a non-uniform, linearity-compensating code mapping). Because § 102 anticipation requires every limitation to be present, US 4,288,785 does not anticipate independent claims 1 or 16. Nor does it disclose the DDC-based "at least one electrode actuated as a function of more than one bit" feature (independent claims in the original filing, claims ~18–22 of the issued set), the non-power-of-two electrode lengths (the "at least one effective area not interrelated by factors of two" independent claim), or the QAM / RZ / return-to-zero features (claims 12–13, 58, 61–63 territory).

2. US 7,061,414 B2 — Chen et al. (IBM)

• Full citation: Y. K. Chen et al., "Optical Digital-To-Analog Converter," US Patent 7,061,414 B2, filed ~2004, granted June 13, 2006 (IBM). The '998 specification cites it as the patent covering the Leven et al. design ("also the subject of U.S. Pat. No. 7,061,414 entitled 'Optical Digital-To-Analog Converter' to Y K Chen et al.").
• Brief description: An optical DAC employing one MZI modulator per every two bits (2-bit-per-modulator segmentation). The '998 patent characterizes it as "highly nonlinear; it yields only 3.8 effective bits for a 6 bit design."
• § 102 anticipation potential: Same analysis as Papuchon: it is an MZI-based segmented-electrode optical DAC and would cover the generic structural limitations, but it does not teach the variable-delta digital-to-digital mapping of claims 1/16, the M>N excess-electrode linearization, or the non-power-of-two electrode optimization. It does not anticipate the independent claims; at most it could anticipate a dependent claim limited to "a plurality of waveguide branches whose outputs are joined" (claim 6) or "MZI-based modulator" (claim 11) if those claims did not incorporate the mapping limitation — but as drafted (dependent on claims 1/16), they inherit the mapping limitation and are not anticipated.

3. US 8,044,835 B2 — Ehrlichman et al. (Ramot) — family member, not prior art

• For completeness: US 8,044,835 (the direct ancestor in this continuation chain, priority 2007-06-13) is in the same family and is not § 102 prior art to the '998 patent (same inventive entity, continuity, and no § 102(a)(2)/102(b) bar against a continuation claiming the same priority date). It appears in the "cited by" records only because Google's citation clustering is family-aware. Do not treat it as an anticipating reference.

B. Non-Patent Literature (verified from the family's citation record)

4. Papuchon et al., Electronics Letters 1980 (NPL)

• Full citation: M. Papuchon et al., "4-bit digital driven integrated amplitude modulator for data processing" (recorded on the family page as "Bits Digital Driven Integrated Amplitude Modulator for Data Processing"), Electronics Letters, February 14, 1980, vol. 16, no. 4, pp. 142–144.
• Brief description: The journal publication corresponding to US 4,288,785 — a multi-electrode MZI amplitude modulator driven by a 4-bit digital word with power-of-two electrode lengths.
• § 102 anticipation potential: As NPL published in 1980, it is prior art under § 102 (pre-AIA and AIA alike for this claim set, given the 2007 priority date). Same analysis as item 1: anticipates generic structural limitations only; not the variable-delta DDC mapping of claims 1/16.

5. Yacoubian et al., IEEE Photonics Technology Letters 2003 (NPL)

• Full citation: A. Yacoubian et al., "Digital-to-analog conversion using electrooptic modulators," IEEE Photonics Technology Letters, vol. 15, pp. 117–119, January 2003.
• Brief description: Proposes one MZI modulator per bit for optical DAC (a parallel-bank architecture rather than a single segmented MZI). The '998 specification cites it as a "much more complex" alternative that failed to solve the nonlinearity problem efficiently.
• § 102 anticipation potential: The one-modulator-per-bit architecture does not disclose the '998 claims' single-modulator/M-electrode structure with a DDC mapping; it does not anticipate claims 1/16 or any of the single-modulator dependent claims. At most it is relevant to § 103 obviousness combinations (and was evidently considered by the examiner as such).

6. Leven et al., LEOS 2004 (NPL)

• Full citation: A. Leven et al., "A 12.5 Gsamples/s optical digital-to-analog converter with 3.8 effective bits," Lasers and Electro-Optics Society (LEOS) 2004, 17th Annual Meeting of the IEEE, vol. 1, pp. 270–271, November 2004.
• Brief description: A 6-bit optical DAC with 3.8 effective bits using a per-2-bits MZI scheme (the counterpart paper to US 7,061,414). Highly nonlinear per the '998 patent.
• § 102 anticipation potential: Same as item 2 — no disclosure of the variable-delta mapping; does not anticipate claims 1/16. Relevant only to § 103.

C. References asserted in litigation (context)

7. IPR2022-01283 (Cisco v. Ramot)

• The PTAB petition (filed July 20, 2022; instituted Feb. 22, 2023; Final Written Decision Feb. 20, 2024, panel Ullagaddi/Ogden/Repko; affirmed by the Federal Circuit April 13, 2026 under Rule 36) challenged claims of the '998 patent. I could not retrieve the specific prior-art grounds/references asserted in the petition or FWD within my search limits — the docket summaries I obtained list the case, panel, and claims-challenged fields but not the reference-by-reference grounds. Do not treat any specific reference as "the IPR ground" unless you pull the FWD PDF from the PTAB API. The CAFC affirmance means the claims that were instituted survived on the grounds asserted, which is consistent with the view that the variable-delta DDC-mapping limitation was the differentiator.

8. Delaware DJ action (Cisco/Acacia v. Ramot, 1:22-cv-00674-GBW)

• The Nov. 12, 2024 indefiniteness ruling (claims 45–47, 49–54 indefinite; claims 1, 4, 6–15, 58, 61–63 not indefinite) is about claim definiteness, not § 102 anticipation — noted here only to distinguish the statutory analyses.

Bottom line on § 102 exposure

The single most relevant prior art is US 4,288,785 (with its 1980 Electronics Letters counterpart). It is the only reference disclosing the exact structural skeleton of the '998 claims (single segmented MZI, N-bit input, M electrodes, intensity output). The reason it does not anticipate the independent claims is the digital-to-digital mapping with non-monotonic deltas (claims 1 and 16) — a feature no located reference discloses, which is also why the IPR challenge was resolved in Ramot's favor and affirmed on appeal.

Confidence caveats: (1) I could not confirm the complete official examiner citation list for the '998 patent — there may be additional U.S. patents cited on the face of the grant that I did not surface (e.g., segmented-electrode MZI or optical-DAC art from the 2000s). (2) The specific IPR2022-01283 grounds are unverified. (3) Claims 23–63 text was not available to me, so my claim-specific statements apply to claims 1–22 (full text) plus the litigation-identified claims (45–47, 49–54, 58, 61–63), which I know only by number and subject-matter characterization from the Delaware ruling.

Obviousness Analysis Under 35 U.S.C. § 103 — US 11,342,998 B2

Prepared: April 26, 2026 | Analyst role: Technical patent analyst | Statute: 35 U.S.C. § 103 (pre‑AIA, applicable to this family's June 13, 2007 priority date)

I. Executive summary

US 11,342,998 claims a linearized optical digital-to-analog modulator whose independent claims (1 and 16) are defined primarily by a digital-to-digital mapping whose output-code step sizes ("deltas") are deliberately non-uniform: decreasing across a first subset of successive input values and increasing across a second subset. In functional terms, the claim is a mathematical fingerprint of a lookup-table predistortion that inverts the cosine-squared transfer function of a Mach-Zehnder interferometer (MZI) so that optical output approximates a straight line over the full dynamic range.

The closest prior art — US 4,288,785 (Papuchon), with its 1980 Electronics Letters counterpart — supplies the entire structural skeleton (segmented-electrode MZI DAC, N-bit input, M electrodes, power-of-two electrode lengths, intensity output) but drives the electrodes directly from the input bits, which the '998 specification itself admits produced a severe nonlinearity/dynamic-range limitation. The later art (Yacoubian 2003; Chen US 7,061,414 / Leven 2004) confirms the problem was known and unsolved but takes it in different architectural directions (parallel MZIs per bit; one MZI per two bits).

Bottom line up front: A colorable § 103 case can be built — the problem (MZI cosine nonlinearity) was notorious, and predistortion was an admitted "common solution" in the modulator art, making a digital implementation of that predistortion a plausible KSR-style "obvious to try." However, on the current adversarial record — IPR2022-01283 challenged all 63 claims, the PTAB's Final Written Decision (Feb. 20, 2024) went against petitioner Cisco (which appealed), and the Federal Circuit affirmed (Apr. 13, 2026, Rule 36) — at least one § 102/§ 103 attack has already failed at the Board and on appeal. Because the specific grounds Cisco ran are unverified from my sources, I cannot state that the combinations analyzed below were the ones rejected; but any new challenger must treat the claim's delta-limitation as the hill to take, and should build on art materially different from whatever Cisco ran. My overall confidence that claims 1/16 would fall under § 103 on the listed references alone is low-to-moderate; dependent claims that add QAM/vector-modulation or non-power-of-two electrode-length features are progressively harder to reach with this record.

II. Legal framework applied

Under 35 U.S.C. § 103 and Graham v. John Deere Co., 383 U.S. 1 (1966), the inquiry is:

1. Scope and content of the prior art;
2. Differences between the prior art and the claims at issue;
3. Level of ordinary skill in the pertinent art; and
4. Secondary considerations of non-obviousness (commercial success, long-felt need, failure of others, teaching away, unexpected results).

Under KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), obviousness may be shown by: combining prior-art elements according to known methods with predictable results; substituting one known element for another; and "obvious to try" where a finite number of identified, predictable solutions exists and the skilled artisan has a reasonable expectation of success. Rigid "teaching-suggestion-motivation" formalism is disfavored; common sense and design incentives may supply the reason to combine. A claimed feature that is an inherent consequence of an otherwise obvious combination need not be separately disclosed in the references (Ex parte Levy, 17 USPQ2d 1461 (Bd. Pat. App. & Interf. 1990); In re Kratz, 592 F.2d 1169 (CCPA 1979)).

III. Person of ordinary skill in the art (PHOSITA)

A PHOSITA for this patent would be a person with ~4–6 years of experience in integrated photonics / optical communications, holding at minimum an M.S. (more typically Ph.D.) in electrical engineering, physics, or applied optics, with working knowledge of:

• Mach-Zehnder interferometer transfer functions (cos² response) and electro-optic phase modulation;
• segmented/multi-electrode modulator design and voltage-length product (Vπ·lπ) tradeoffs;
• mixed-signal design: DAC/ADC architectures, binary weighting, lookup tables, and digital predistortion;
• optical transmitter architectures (intensity modulation, and at least familiarity with QAM/coherent formats by 2007).

The 2006–2007 timeframe matters: digital predistortion of nonlinear analog front-ends was mature (RF power-amplifier linearization, cable-TV optical links), and the '998 specification itself concedes "in practically all present systems signals are processed digitally."

IV. Claim scope — what must be shown obvious

IV.A. Independent claims 1 and 16 (full text available)

Technical reading of the delta limitation. If the target output is a uniform linear ramp Uᵢ = i/2ᴺ and the modulator transfer is T = cos²((π/2)·Σ BⱼLⱼ), then the electrode code required for linearity is Cᵢ = (2/π)·arccos(√Uᵢ). Successive code deltas Δᵢ = Cᵢ₊₁ − Cᵢ are proportional to 1/√(i(2ᴺ−i)) — large at low codes, shrinking toward mid-scale, and growing again toward the top of the range. That is precisely "deltas decrease over a first subset, then increase over a second subset." The limitation is therefore not an arbitrary design detail: it is the necessary signature of any mapping that predistorts a cosine-squared modulator for full-range linearity. This cuts both ways for § 103: (a) it makes the limitation inherent in the known predistortion concept applied to an MZI (favoring obviousness if the DDC concept is obvious); but (b) none of the listed references discloses a DDC or any non-direct mapping, so the challenger must rely on motivation/inherency reasoning rather than an explicit teaching.

IV.B. Dependent claims 2–22 (full text available)

• Claims 2, 17 — delta behavior in the reverse input direction (decreasing inputs): a trivial bidirectional extension of claims 1/16; real mappings are defined for all inputs.
• Claims 3, 18 — a third subset where deltas "remain the same": permits a plateau; broad catch-all.
• Claims 4–5, 19–20 — electrodes "driven based on / directly responsive to" the voltage values: inherent to any electrode-driven modulator (Papuchon).
• Claims 6, 21 — two waveguide branches, first/second sets of voltages, joined outputs: the standard push-pull MZI structure.
• Claims 7, 22 — "phase modulation, amplitude modulation, or phase and amplitude modulation": the "or" makes amplitude-only sufficient (Papuchon's mode of operation).
• Claims 8–10 — outputs vary in intensity / amplitude / phase: inherent in an MZI driven by electrode voltages.
• Claim 11 — MZI-based modulator: Papuchon.
• Claim 12 — outputs "corrected for non-linearities" of the modulator or of the optical fiber.
• Claim 13 — mapping "determined based on a pattern for actuating drive voltages that alters a linearity of an optical response": restates the independent-claim concept in functional terms.
• Claim 14 — subsets may be "non-contiguous": permissive breadth; nearly any real mapping satisfies or can be designed to satisfy this.
• Claim 15 — input optical signal provided by an optical source: conventional.

Net: claims 2–15 and 17–22 rise and fall largely with claims 1 and 16. The only additional structural meat is the dual-branch push-pull drive (claims 6, 21) and the amplitude/phase option (claims 7, 22) — both standard MZI features by 2007.

IV.C. Claims 23–63 (text not available)

I have only fragmentary knowledge: Delaware held claims 45–47 and 49–54 indefinite and found claims 1, 4, 6–15, 58, 61–63 not indefinite (Nov. 12, 2024). From the specification's feature list and the QAM/RZ/semiconductor-laser embodiments, the latter claims likely cover QAM constellations (≥16 points), return-to-zero minimum-amplitude operation, semiconductor light-generating devices (laser/LED with M electrodes and threshold electrode), M=N embodiments, and non-power-of-two electrode effective areas. I cannot responsibly perform a claim-by-claim § 103 analysis of claims 23–63 without their text; my comments on QAM/RZ/laser claims below are expressly conditional.

V. Prior art inventory (from the Prior Art section)

Additional evidentiary material (not itself prior art, but the patentee's own admission of the state of the art, usable against Ramot as an admission): the '998 specification states that (i) MZI attenuation "varies as the cosine of the phase difference"; (ii) "the common solutions for this problem are either the biasing of the device to a quasi linear regime coupled with reducing the modulation range to reduce distortion, or use of an analog pre-distortion circuit to feed the modulator"; and (iii) "in practically all present systems signals are processed digitally." These admissions establish that the problem (cosine nonlinearity) and one known solution class (predistortion) were in the art.

Verified gap: I could not retrieve the specific references/grounds asserted in IPR2022-01283. The Patexia docket confirms all 63 claims were challenged and the FWD was appealed, but not the ground-by-ground art list. Do not assume any combination below was or was not the litigated ground until the FWD is pulled from PTAB E2E.

VI. Primary § 103 combination

Combination C1: Papuchon (A1 and/or B1) + the nonlinearity-recognition art (A2/B2/B3) + the admitted predistortion solution, implemented digitally

Reference mapping to claim 1/16 elements:

VI.A. Motivation to combine — the narrative

1. The problem was explicit and pressing. A1/B1 (1980–81) established the segmented-electrode MZI DAC but, per the '998 specification's own account (and consistent with the historical record), its "power-of-two digital sequence … did not solve the non-linearity problem, and thus suffered from severe limitation in the dynamic range, and subsequently the attainable resolution." By 2003–2004, B2 (Yacoubian) and A2/B3 (Chen/Leven) were still wrestling with the same problem — B2 proposes "much more complex" one-MZI-per-bit hardware; A2/B3 concede only 3.8 effective bits for a 6-bit design. A PHOSITA reading this art cluster in 2007 would understand: (i) the MZI-DAC idea is sound, (ii) its cosine response is the obstacle, and (iii) nobody had yet fixed it cleanly within a single-modulator architecture. That is a textbook "design need or market pressure" under KSR.

2. The solution class — predistortion — was admitted to be known. The '998 specification concedes that feeding the modulator with a pre-distortion circuit was a "common solution" to MZI nonlinearity. A PHOSITA seeking to improve Papuchon's DAC would naturally reach for the known remedy for MZI nonlinearity — predistortion — rather than a novel one. This supplies the missing reason to alter Papuchon's direct drive.

3. Digital implementation was the obvious (indeed, nearly forced) embodiment. The '998 specification concedes that "in practically all present systems signals are processed digitally." By 2007, implementing a predistortion function as a digital lookup table / code mapping was routine (the RF-PA and CATV linearization arts did exactly this). Critically, in Papuchon's architecture the MZI itself is the DAC, and the electrode drives are inherently binary (each electrode on/off at a common voltage v). There is no analog domain between the input word and the modulator in which to insert a conventional analog predistorter — the only place to predistort is before the electrodes, i.e., in the digital domain, mapping each N-bit input word to a different M-bit electrode pattern. A PHOSITA who (a) keeps Papuchon's single-modulator architecture, (b) wants full-range linearity, and (c) accepts the known predistortion remedy, is led with near-logical necessity to a digital-to-digital converter between the input word and the electrodes. That is the "finite number of identified, predictable solutions" scenario KSR treats as obvious-to-try.

4. The alternative routes in the art reinforced, rather than undermined, the incentive. B2 (parallel MZIs per bit) and A2/B3 (one MZI per two bits) demonstrated that the field was actively seeking any workable linearization of an optical DAC; their hardware complexity (B2) and poor ENOB (A2/B3) would motivate a PHOSITA to seek a simpler fix to the original Papuchon architecture — namely, smarter mapping rather than more hardware.

VI.B. Reasonable expectation of success

A PHOSITA would have a strong expectation that a lookup-table mapping chosen to match the known cos² transfer function would linearize the output: for each desired output level Uᵢ, choose the electrode pattern whose cos² output is closest to Uᵢ (a straightforward, computationally trivial search over 2ᴹ patterns). The '998 specification's own optimization (its Eq. 5: B̂ᵢ = Dec2Binᴹ((2/π)·arccos(√Uᵢ))) is a textbook least-squares/inverse-function fit that any competent engineer could implement without invention. The spec's FIG. 2A→2B result (a dramatic linearity improvement from mapping alone, with unchanged Papuchon electrode lengths) confirms that the fix is a simple code reassignment — precisely what a PHOSITA would predict from inverting a known transfer curve.

VI.C. The delta limitation — inherency argument

Even though no listed reference spells out "deltas decrease in a first subset and increase in a second subset," that pattern is mathematically forced by the goal of linearizing a cos² response over the full dynamic range (see § IV.A: Δᵢ ∝ 1/√(i(2ᴺ−i)), U-shaped). Under the doctrine that a limitation inherent in an otherwise-obvious combination need not be expressly taught (Ex parte Levy; Kratz), if the DDC itself is obvious, the delta structure follows as a matter of course. This is the single most important analytical point in the § 103 case for claims 1/16 — the claims effectively read on any full-range cosine-linearizing code mapping, i.e., on the predictable product of applying the known predistortion remedy to the known Papuchon device.

VI.D. The counter-narrative (why C1 may fail)

1. No listed reference teaches a DDC. Papuchon's entire design is direct binary weighting; Yacoubian and Chen/Leven respond to the nonlinearity with architecture changes, not mapping changes. A challenger must therefore rely on general knowledge and the patentee's own admissions to supply the "insert a digital predistorter" step — an argument that invites the response that the specification's mention of "analog pre-distortion" actually distinguishes the claimed digital-to-digital approach, and that no reference suggests predistorting a segmented-electrode DAC (as opposed to predistorting the analog drive of a single modulator).
2. Possible teaching away. B2 and A2/B3, knowing of Papuchon's failure, abandoned the single-modulator architecture. A fact-finder could find that the skilled artisan's design path led away from repairing Papuchon and toward multi-modulator schemes — undermining the "obvious to try the DDC fix on Papuchon" theory.
3. The delta limitation is claimed structurally, not just functionally. Although I argue it is inherent, a patent owner will stress that the claim requires the mapping to actually be implemented with that delta behavior, and that the prior art's direct binary mapping has monotonic power-of-two deltas — the opposite of the claimed non-uniform pattern. The contrast between Papuchon's uniform 2⁻ʲ step structure and the claim's deliberately warped step structure is the patent owner's cleanest § 103 differentiator.
4. The adversarial record. IPR2022-01283 (all 63 claims challenged; FWD Feb. 20, 2024; CAFC affirmed Apr. 13, 2026) demonstrates that some § 102/§ 103 attack failed at the Board and on appeal. If Cisco's grounds resembled C1 (Papuchon-family art + a motivation theory), the Board's rejection is direct evidence that C1-type reasoning did not persuade. Because the grounds are unverified, I flag this as a caution rather than a finding.

VII. Secondary § 103 combinations

Combination C2: Papuchon (A1/B1) + Chen/Leven (A2/B3) — electrode-segmentation design freedom

• What it adds over C1: A2/B3 show that an MZI DAC's electrodes need not follow Papuchon's strict binary sectioning — Chen/Leven deliberately section electrodes into 2-bit groups — and they quantify the linearity deficit (3.8 ENOB). This teaches the PHOSITA that (i) electrode segmentation is a design variable, not a fixed law of nature, and (ii) linearity is the metric that matters.
• Use: supports obviousness of dependent-claim features directed to modified electrode arrangements, and reinforces the incentive prong of C1. It does not by itself teach the DDC or the delta structure.
• Weakness: Chen/Leven's own solution was more hardware, not smarter mapping; no express suggestion to add a code-mapping stage.

Combination C3: Papuchon + standard numerical optimization (for non-power-of-two electrode lengths and/or M>N features)

• The '998 spec's own optimization (Eqs. 6–9) is a least-squares linear solve for electrode lengths given a chosen mapping — the kind of routine numerical optimization any PHOSITA would apply to minimize RMSE. If claims directed to "at least one effective area not interrelated by factors of two" (or to excess electrodes M>N) are in play (claims 23–63 territory; text unverified), a challenger would argue that optimizing L and adding degrees of freedom (extra electrodes) are obvious design refinements once the linearization goal is set.
• Weakness: No listed reference discloses non-power-of-two electrode lengths or M>N; the motivation is generic optimization, which PTAB/courts sometimes reject as hindsight absent a specific teaching or problem in the reference. This ground is weaker than C1 for claims 1/16 and is only relevant to dependent claims.

Combination C4: For QAM / vector-modulation claims (conditional)

If claims 58/61–63 (and others) cover QAM with ≥16 constellation points via dual-arm electrode sets (the spec's Example IV), none of the listed references is a QAM/vector modulator — all are single-dimension intensity DACs. Reaching those claims would require importing general QAM-transmitter knowledge or additional art (dual-drive MZM QAM transmitters, I/Q electro-optic modulators) not present in the Prior Art section. On this record, the QAM claims are the least vulnerable to § 103. Same conclusion for return-to-zero and semiconductor-laser/LED claims if present — the listed art is all external-modulator, intensity-only, NRZ-oriented.

VIII. Dependent claims 2–22 — incremental analysis

IX. Secondary considerations (Graham factor 4)

On the available record:

• Long-felt need: The '998 spec itself closes its Background with "There is therefore a need for a digital to analog converter which would offer improved linearity of response without sacrificing efficiency or dynamic range" — evidence that the problem was long-recognized (1980 Papuchon → 2007). Whether the solution met with market success is unproven: no commercial product, license revenue, or industry adoption evidence was located. Ramot is a university TTO that does not practice the claims.
• Failure of others: The Yacoubian (B2) and Chen/Leven (A2/B3) papers are evidence that others tried and failed to linearize optical DACs cleanly — this cuts for non-obviousness (it shows the problem resisted the field's known approaches) but also for obviousness (it shows active, sustained motivation — the KSR design-need prong). Net: neutral-to-favorable for the patent owner.
• Teaching away: Arguable from B2/A2/B3 abandoning single-MZI architectures (favorable to patent owner); rebuttable because abandoning one architecture does not teach away from fixing it with a known remedy.
• Adversarial outcome: The PTAB FWD and CAFC affirmance in Ramot's favor is real-world evidence that at least one § 102/§ 103 attack failed — though Rule 36 affirmance gives no reasoning and the grounds are unverified. This is the strongest secondary indicator of non-obviousness available and must temper any confident obviousness conclusion.
• Unexpected results / praise / licensing: None located.

X. Candid assessment and confidence levels

Most important caveats:

1. IPR grounds unverified. I could not retrieve the art Cisco ran in IPR2022-01283. Before any new § 103 campaign, pull the FWD (PTAB E2E) and confirm which references/grounds were rejected — estoppel and res judicata-adjacent concerns aside, re-running a failed combination is futile, and the Board's reasoning will reveal exactly why the delta-limitation survived.
2. Claim text gap. My claim-specific analysis is rigorous only for claims 1–22 (full text). Claims 23–63 require their text before any responsible § 103 conclusion.
3. § 103 vs. other statutes. The delta limitation's essentially mathematical/functional character makes claims 1/16 arguably more vulnerable to § 101 (abstract idea / purely numerical mapping characteristic) and § 112 (indefiniteness of "deltas between numerical values," demonstrated by Delaware's partial indefiniteness ruling) than to § 103. A defendant should weigh those avenues alongside any § 103 attack.
4. New-art runway. If a fresh § 103 attack is pursued, it should use art not in the listed set and (once verified) not in Cisco's petition — e.g., analog/digital predistortion literature for MZI and RF modulators, other segmented-electrode MZI DAC papers, and 1990s–2000s optical transmitter linearization patents — to supply an explicit teaching of code-mapping linearization rather than relying solely on motivation/inherency.

XI. Conclusion

On the prior art identified in the Prior Art section, the most plausible § 103 case is C1: combine Papuchon's segmented-electrode MZI DAC (US 4,288,785 / Electronics Letters 1980) with the field's demonstrated, long-unmet need to linearize it (Yacoubian 2003; Chen US 7,061,414 / Leven 2004) and the admitted "common solution" of predistortion, implemented digitally as a digital-to-digital mapping — the delta-decrease/increase structure of claims 1/16 being the inherent mathematical consequence of inverting a cos² transfer for full-range linearity. The motivation narrative is genuinely strong under KSR; the evidentiary gap — no listed reference discloses any non-direct mapping, and the only adversarial § 103 test on record (IPR2022-01283 → CAFC 2024-1728, affirmed Apr. 13, 2026) was lost by the petitioner — is equally genuine. My honest bottom line: claims 1 and 16 would survive a § 103 challenge built only on the listed references more likely than not, and the dependent QAM/vector claims (claims 23–63 territory) are substantially safer still; a successful § 103 theory would need materially stronger, explicitly mapping-teaching prior art than this record contains.