US 9203425

I'll search for authoritative information on US 9,203,425, including any 2026 CAFC activity.

Let me check specifically for any 2026 appellate or litigation activity involving this patent.

Let me check the IPR numbers and which family patents were involved.

US Patent 9,203,425 B2 — Bibliographic Summary

I searched for the specific number 9203425, including 2026 CAFC activity. Here is what I found, with sources flagged.

Priority chain (from the specification): 14/662,343 is a continuation of 14/325,486 (filed Jul. 8, 2014, now US 9,031,417), which is a continuation of 13/280,371 (filed Oct. 25, 2011, now US 8,797,198), which is a continuation of 12/636,805 (filed Dec. 14, 2009, now US 8,044,835), which is a continuation-in-part of PCT/IL2008/000805 (filed Jun. 12, 2008), which claims priority to U.S. Provisional 60/943,559 (filed Jun. 13, 2007). Source: https://patents.google.com/patent/US9203425/en

Abstract (verbatim)

"A modulator device for converting digital data into modulation of an optical signal includes an electronic input for receiving an input data word of N bits and an electrically controllable modulator for modulating the intensity of an optical signal, the modulator including M actuating electrodes where M≧N. An electrode actuating device, most preferably a digital-to-digital convertor, operates actuating electrodes so that at least one electrode is actuated as a function of values of more than one bit of the input data word. According to an alternative, or supplementary, aspect of the invention, the set of electrodes includes at least one electrode having an effective area which is not interrelated to others of the set by factors of two. In one preferred implementation, a Mach-Zehnder modulator also provides phase modulation to give QAM functionality. Another implementation employs a semiconductor laser."

Plain-Language Overview of the Independent Claims

Important caveat: The claim text I retrieved runs through claim 28 and then truncates, although the patent states "(30)" claims. Claims 29 and 30 exist but I do not have their text, so I cannot characterize them. The four independent claims I can characterize are 1, 11, 15, and 24.

Claim 1 — "A circuit for converting a digital input data word… into a modulated signal"
A circuit with two parts: (a) an electrically controllable device carrying multiple actuating electrodes, hooked to an optical output port, that produces an optical signal modulated by a "binary actuation vector" (a set of on/off electrode voltages); and (b) an electrode actuating device that receives the multi-bit digital word and contains a digital-to-digital converter (DDC). The DDC supplies the binary voltages and presents the binary actuation vector to the electrodes so it represents the input data word. The key numeric limit: the number of actuating electrodes must be strictly greater than the number of bits in the input data word (M > N).

Claim 11 — "A modulator system for converting input digital data into analog modulation of an input optical signal"
An electronic input receives digital data. The system includes an electrically controllable optical modulator with multiple optical waveguide branches that carry the input optical signal, with multiple actuating electrodes disposed over those branches, plus an electrode actuating device. In response to the input data word, the actuating device supplies a binary actuation vector of binary electrode-actuation voltages. Again, M > N. The binary voltages modulate both the power and the optical phase delay of the input optical signal to produce the modulated optical output — this is the claim that captures the QAM/phase aspect.

Claim 15 — "An optical modulator for converting a digital input data word… into a modulated signal"
An electrically controllable device with multiple actuating electrodes provides an optical output modulated in response to binary electrode-actuation voltages applied digitally. M > N. An electrode actuating device receives the input data word and provides a binary actuation vector of multiple binary electrode-actuation voltages that actuate the electrodes, the vector representing the input data word and driving the electrodes.

Claim 24 — Method of converting a digital input data word into a modulated optical signal
Steps: (1) receive the digital input data word at an electrode actuating device; (2) pass an input optical signal through an electrically controllable optical modulator having multiple actuating electrodes disposed over at least one optical waveguide, where the number of electrodes is larger than the number of bits of the input data word; and (3) supply a binary actuation vector of binary electrode-actuation voltages to those electrodes responsive to the received data word, thereby modulating the power and optical phase delay of the input optical signal.

Observation on claim emphasis: The granted independent claims are narrower in framing than the specification's original summary. The specification's headline concept was "M ≥ N" plus "at least one electrode actuated as a function of values of more than one bit of the input data word." The granted independent claims instead emphasize (i) M strictly greater than N, (ii) binary actuation vectors / voltages applied digitally (direct digital driving), and (iii) in claims 11 and 24, concurrent modulation of both power and optical phase delay. Dependent claims 2–10, 12–14, 16–23, and 25–28 add features such as a semiconductor modulator, light generation device, phase-only/intensity-only/both modulation, PAM and QAM outputs, selection of an actuation vector from a set of available vectors, and differing electrode actuation patterns on first vs. second waveguide branches.

Litigation / 2026 CAFC Activity — Findings and a Non-Finding

• I found no 2026 CAFC docket proceeding that names U.S. 9,203,425 itself. I want to be explicit that I could not locate one, rather than infer one exists.
• There is significant 2026 CAFC activity in this patent family: Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd., Nos. 2024-1726, 2024-1727, 2024-1728 (Fed. Cir. April 13, 2026), a Rule 36 summary affirmance (per curiam — Moore, Mayer, Lourie) of three PTAB final written decisions in IPR2022-00575, IPR2022-00576, and IPR2022-01283. Source: https://www.cafc.uscourts.gov/opinions-orders/24-1726.RULE_36_JUDGMENT.4-13-2026_2674975.pdf
• Critically, the IPRs decided in that 2026 appeal involved U.S. 11,133,872 B2, not U.S. 9,203,425. The PTAB's own caption in IPR2022-00575/-00576 identifies "Patent 11,133,872 B2," and third-party coverage describes the cancelled patent as US11133872B2. US 11,133,872 is a later continuation in the same family (its specification lists US 9,203,425 as a parent). So the 2026 affirmance affects a family member, not the '425 patent directly.
• The Google Patents page for US 9,203,425 lists family-level litigation events (these are keyed to the family, not necessarily to the '425 patent itself): PTAB IPR2020-00122 filed (Not Instituted – Procedural); CAFC case 20-2047; D. Del. 1:21-cv-00295; E.D. Tex. 2:19-cv-00225. Treat these as family-level pointers, not confirmation that the '425 patent was the patent-in-suit.
• The E.D. Tex. action (2:19-cv-00225) asserted U.S. 10,033,465, 10,270,535, and 10,461,866 against Cisco — again family members, and numerous ex parte reexaminations (90/014,528; 90/014,607; 90/014,728; 90/014,527; 90/015,280; 90/019,333; 90/019,420) were run against those patents.

Explicit Uncertainties

1. Claims 29–30 text not available in the material I retrieved (truncated at claim 28). I cannot describe them.
2. No CAFC 2026 docket naming US 9,203,425 specifically was found; the only 2026 CAFC decision I located in this family is Cisco v. Ramot (24-1726 et al.), which concerns US 11,133,872 B2. If a 2026 appeal specifically involving the '425 patent exists, I did not find it and cannot confirm it.
3. Live USPTO Patent Center / PACER docket status was not directly queried (I relied on surfaced documents and patent databases); current legal status should be verified in USPTO Patent Center if it matters for your purpose.
4. The abstract and claim language above are quoted from the authoritative full text provided (https://patents.google.com/patent/US9203425/en) and the issued-patent PDF (patentimages.storage.googleapis.com/.../US9203425.pdf), which agree.

I'll research litigation involving US patent 9203425 across patent litigation databases.

The generic searches didn't return the specific patent. Let me search the specific case numbers and parties identified on the patent record.

Litigation Involving US Patent 9,203,425

Important threshold caveat. US 9,203,425 ("the '425 patent," Linearized optical digital-to-analog modulator, Ramot at Tel Aviv University Ltd.) is the root of a large continuation family. Ramot's enforcement activity appears to have been directed at later-issued continuations in that family — principally US 10,033,465 ("'465"), US 10,270,535 ("'535"), and US 10,461,866 ("'866") — rather than at the '425 patent itself. The litigation items below are recorded on the Google Patents "family has litigation" data for the '425 patent (https://patents.google.com/patent/US9203425/en). I flag where the search evidence shows a family member, not the '425 patent, was the actual asserted patent. I did not find any case in which US 9,203,425 was itself named as an asserted patent.

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, Eastern District of Texas (Marshall Division), Judge Rodney Gilstrap
• Case No.: 2:19-cv-00225-JRG
• Filing date: June 12, 2019
• Asserted patents: US 10,033,465 ('465), US 10,270,535 ('535), US 10,461,866 ('866) — all family members/continuations of the '425 patent. Accused products were Cisco networking equipment (e.g., NCS 4000 series, Nexus 400G, CFP2/QSFP56/QSFP-DD modules) using PAM4/16-QAM modulation.
• Status/outcome:
  • Cisco's first motion to stay (pending IPR) was denied without prejudice (Order, Feb. 4, 2020).
  • Following denial of the IPRs, Cisco filed ex parte reexaminations; the court granted a stay in January 2021 (2021 WL 121154) once all asserted claims stood rejected. See Dkt. 235.
  • The case was active into 2025–2026: a Joint Status Report (Dec. 11, 2025) and further docket activity in March–April 2026 (motion to withdraw counsel granted; discovery/scheduling/in limine motions denied as moot on Apr. 1, 2026).
  • No final judgment on the merits of the '425 patent (it was not asserted).
  • Docket: https://www.courtlistener.com/docket/15765854/ramot-at-tel-aviv-university-ltd-v-cisco-systems-inc/

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

• Plaintiff: Ramot at Tel Aviv University Ltd.
• Defendant: Acacia Communications, Inc.
• Jurisdiction: U.S. District Court, District of Delaware
• Case No.: 1:21-cv-00295
• Filing date: February 26, 2021
• Cause of action: Willful patent infringement (Nature of Suit 830 – Patent)
• Status/outcome: Closed (per RPX and the Ex Parte case tracker).
• Sources: https://litigation.rpxcorp.com/litigation/dedce-74725-ramot-at-tel-aviv-university-v-acacia-communications ; https://ai-lab-cl-prod.azurewebsites.net/case/dct/ded/1:21-cv-00295/ramot-at-tel-aviv-university-ltd-v-acacia-communications-inc

3. Court of Appeals for the Federal Circuit — Appeal No. 20-2047

• Listed parties/jurisdiction: U.S. Court of Appeals for the Federal Circuit (case listed on the '425 patent record under "US case filed in Court of Appeals for the Federal Circuit").
• Case No.: 20-2047
• Status: Listed on the Google Patents record as arising from the Ramot family litigation. I could not independently confirm from the search results the precise nature, parties, or disposition of this specific docket number. Given its 2020 vintage and the parallel E.D. Tex. case, it most plausibly relates to an appeal/mandamus in the Ramot v. Cisco or Ramot v. Acacia matters, but I state this only as an inference, not confirmed fact.

4. PTAB — Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.

• Petitioner: Cisco Systems, Inc.
• Patent Owner: Ramot at Tel Aviv University Ltd.
• Forum: Patent Trial and Appeal Board (PTAB)
• Case No.: IPR2020-00122 (listed on the '425 patent record; "Not Instituted – Procedural"). Related Cisco petitions IPR2020-00123 and IPR2020-00484 were filed against the parallel Ramot patents.
• Status/outcome: Institution denied under 35 U.S.C. §314(a) (Fintiv-based), Paper 15 (May 15, 2020). See Cisco Sys., Inc. v. Ramot at Tel Aviv Univ. Ltd., IPR2020-00122, Paper 15 (PTAB May 15, 2020). These IPRs challenged Ramot's continuation patents (the '465/'535/'866 family), not the '425 patent itself — so the IPR is a family litigation item relative to the '425 patent.
• Reference: https://portal.unifiedpatents.com/ptab/case/IPR2020-00122

5. Related post-grant (PTO) proceedings — not "litigation," but part of the same dispute

Cisco filed ex parte reexaminations against the asserted Ramot patents, which drove the E.D. Tex. stay:

• 90/014,527 ('535); 90/014,528 ('465); 90/014,606; 90/014,607; 90/014,608; 90/015,280 ('866); 90/019,333 ('465); 90/019,420 ('535).

6. Other/aggregated records

• Darts-IP "first worldwide family litigation" — the '425 patent record notes a "First worldwide family litigation filed" under family ID 39798167, indicating litigation on this patent family exists in more than one jurisdiction. I could not open the underlying Darts-IP record to enumerate specific non-U.S. cases.

Bottom line

• US 9,203,425 itself does not appear to have been asserted as a patent-in-suit in any case the search evidence surfaced. The known litigation is family litigation tied to the '425 patent on Google Patents.
• The core enforcement campaigns are:
  1. Ramot v. Cisco Systems — E.D. Tex. 2:19-cv-00225-JRG (filed June 12, 2019), stayed pending ex parte reexamination, still active on the docket into 2026.
  2. Ramot v. Acacia Communications — D. Del. 1:21-cv-00295 (filed Feb. 26, 2021), closed.
  3. Cisco's IPRs — IPR2020-00122 (and 00123, 00484), all institution-denied in 2020.
  4. CAFC Appeal No. 20-2047 — listed on the patent record; details unconfirmed.
• I could not confirm any ITC (Section 337) action involving the '425 patent or its family.

If you need the specific claims asserted, infringement/validity outcomes, or the CAFC 20-2047 disposition verified, those would require pulling the PACER dockets for 2:19-cv-00225 and 1:21-cv-00295 and the CAFC docket for 20-2047 directly, which I was unable to complete within this session.

I'll research the PTAB proceedings on this patent, starting with the one flagged in the patent's family page.

Key finding: IPR2020-00122 is on a different patent in the same family. Let me verify whether any AIA trial actually targets the '425 patent itself.

Now let me check the later 2022 IPRs and confirm whether any targets the '425 patent itself.

Proceedings overview

There are zero AIA trial proceedings on US 9,203,425 itself. The USPTO Open Data Portal (the canonical structured source in my instructions) returns no IPR, PGR, or CBM for this patent, and my searches found none either. The patent is Active, with 30 claims, none of which has ever been canceled, narrowed, or adjudicated at the PTAB. The one PTAB docket number that appears on this patent's Google Patents page — IPR2020-00122 — is a family-level link that does not challenge the '425; it challenges continuation patent US 10,033,465. That distinction is the single most important thing for a defendant to understand, because it means the "hardened by IPR" story belongs to the children of this patent, not to the '425.

The bottom-line defensive posture is therefore nuanced, and cutting against the patent on the family level while leaving the '425 untouched:

• On the '425 specifically: no estoppel, no institution decision, no claim construction by the Board. A defendant facing assertion of the '425 has the full prior-art field open — including art already tested against the sister patents.
• On the family: every IPR Cisco filed against the '425's continuations ('465, '535, '866) was denied institution under § 314(a) (Fintiv/NHK) — never decided on the merits. Ramot has never had a claim survive a PTAB Final Written Decision, but it has also never had one canceled. The "no IPR activity" needle for the '425 is real but is a sign the patent was never asserted (Ramot asserted the '465, '535, and '866 against Cisco, not the '425), not a sign the patent is bulletproof.

IPR2020-00122 — Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.

⚠️ Not a proceeding on US 9,203,425. This challenges US 10,033,465 B1 ("the '465 patent"), a continuation of the '425. It is listed here only because it is the docket the '425's Google Patents family page links to.

• Type: Inter Partes Review
• Filed: 2019-11-05
• Status: Not Instituted – Procedural (Unified Patents portal verbatim: "Not Instituted - Procedural / Appealed § 314(a)"). Gloss: the Board never reached the merits; it denied institution as an exercise of § 314(a) discretion.
• Judge panel: Not publicly identified in the sources I could retrieve.
• Petition grounds: pre-AIA § 103 obviousness. Primary reference US 7,277,603 (Roberts) combined with Ho (Phase-Modulated Optical Communication Systems, 2005) and/or US 7,609,935 (Burchfiel). The dispute centered on whether Roberts' DSP / "non-linear compensator" taught the claimed digital-to-digital conversion/mapping function. (Exhibit list from the Unified Patents docket: Ex.1005 Roberts, Ex.1006 Ho, Ex.1008 Burchfiel, plus dictionary and background references.)
• Institution decision: Denied — decision dated 2020-05-15. The panel relied on NHK Spring v. Intri-Plex and Apple v. Fintiv and reasoned that the E.D. Tex. trial (set for 2020-12-09) would precede the Board's projected FWD by roughly six months, that the parties had already invested heavily in claim construction and expert discovery, and that the petition raised the "same or substantially the same claims, grounds, arguments, and evidence as presented in the district court proceeding." Conclusion: "instituting would be an inefficient use of Board, party, and judicial resources."
• Final Written Decision: None — never instituted. No claim of the '465 (or any patent) was canceled or sustained.
• Settlement / termination: N/A (no trial).
• Appeal: Yes. Cisco appealed and, in the alternative, petitioned for mandamus challenging the NHK/Fintiv rule as an unlawful substantive rule under the APA. Consolidated at Nos. 2020-148, 2020-2047, 2020-2049 (Fed. Cir.). Disposition: appeals dismissed and mandamus denied, In re Cisco Systems Inc., nonprecedential order, 2020-10-30 (Dyk, Wallach, Taranto, JJ.) — holding the institution denials are unreviewable and refusing mandamus. CourtListener
• Defensive value (for the '425): None directly — it is a different patent. Its only relevance to the '425 is evidentiary: it shows Cisco tried to invalidate this patent family and was shut out on procedural grounds, not on the merits, which means neither side has a merits precedent to lean on.

IPR2020-00123 — Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.

⚠️ Not a proceeding on US 9,203,425. Targets US 10,270,535 (another continuation).

• Type: Inter Partes Review
• Filed: 2019-11 (Cisco filed its November 2019 wave together with the '465 petition)
• Status: Not Instituted – Procedural / § 314(a) denied
• Judge panel: Not publicly identified in retrieved sources.
• Petition grounds: pre-AIA § 103; same reference family — Roberts as primary, with Ho and Burchfiel. Ramot's Preliminary Response (bdiplaw.com PDF and the PTAB E2E preliminary response) argued all three Grounds failed because Roberts does not disclose or suggest the claimed digital-to-digital "converting," and separately that the art was cumulative under § 325(d).
• Institution decision: Denied — ~2020-05-15, same Fintiv/NHK theory as IPR2020-00122 (same trial date, same overlap analysis). Ramot also prevailed on § 325(d).
• Final Written Decision: None — never instituted.
• Settlement / termination: N/A.
• Appeal: Consolidated with IPR2020-00122 in the same Fed. Cir. order, Nos. 2020-148 / 2020-2047 / 2020-2049; dismissed / mandamus denied 2020-10-30.
• Defensive value (for the '425): None directly. Same as above — procedural denial only.

IPR2020-00484 — Cisco Systems, Inc. v. Ramot at Tel Aviv University Ltd.

⚠️ Not a proceeding on US 9,203,425. Targets US 10,461,866 B2 (another continuation).

• Type: Inter Partes Review
• Filed: ~2020-01 (the third and last of Cisco's November 2019–January 2020 wave)
• Status: Institution denied (Unified Patents docket caption; Bunsow De Mory press release: "decision issued today … denied institution of the third … IPR petition")
• Judge panel: Not publicly identified in retrieved sources.
• Petition grounds: pre-AIA § 103 — challenged claims 7, 8, 10–12, 19, 20, 22–24; Ground 1 claim 7, 8, 10–12 over Roberts + Ho; Ground 2 claims 19, 20, 22–24 over Roberts + Ho + US 6,760,111 (Mark). (Source: PTAB decision PDF, IPR2020-00484.)
• Institution decision: Denied — ~2020-08, again on § 314(a) / NHK-Fintiv (E.D. Tex. trial ahead of the Board's FWD, which was projected for August 2021). The panel also found Petitioner "provides little explanation of how Roberts' non-linear compensator 18 teaches or suggests the specific relationship recited in the claims," i.e., the merits were weak too.
• Final Written Decision: None — never instituted.
• Settlement / termination: N/A.
• Appeal: No (this denial does not appear in the 2020-148/2047/2049 consolidated appeal).
• Defensive value (for the '425): None directly. But note the substance: the Board expressed skepticism that Roberts teaches a digital-to-digital converter — a reasoning thread that a defendant asserting similar art against the '425 would have to overcome or distinguish, since the '425's claims recite the same DDC concept.

Second-wave IPRs (IPR2022-00575 / IPR2022-00576 / IPR2022-01283) — Cisco Systems, Inc. v. Ramot

⚠️ Not proceedings on US 9,203,425. These are the appeals decided at the Federal Circuit on 2026-04-13 and are not on the '465/'535/'866 either.

• Type: Inter Partes Review (three related proceedings, later-issued Ramot optical-DAC patents)
• Appeal: Consolidated at Nos. 2024-1726, 2024-1727, 2024-1728; AFFIRMED per Rule 36, entered 2026-04-13 — per curiam (Moore, C.J., Mayer and Lourie, JJ.). Counsel: Angela M. Oliver / Haynes and Boone for Cisco; Denise M. De Mory / Bunsow De Mory for Ramot. (CAFC Rule 36 judgment PDF)
• Which patents: I could not verify the target patent numbers of these three IPRs with the confidence required. A third-party litigation intelligence summary associates this appeal with the cancellation of US 11,133,872 B2 (the PatSnap litigation blog), but that association is not confirmed by a primary source I retrieved. Do not rely on it. Flagging as unverified.
• Defensive value (for the '425): Indirect. If a Rule 36 affirmance of a cancellation in this family is confirmed, it would show Cisco (as a serial petitioner against this family) eventually got traction on the later, separately-claimed continuations — which is a caution that the family's later issuances are more vulnerable than the '425. It does not cancel anything in the '425.

Related non-AIA activity (ex parte reexamination — not PTAB trials)

After its IPR denials, Cisco pivoted to ex parte reexamination, which is a USPTO examining-corps proceeding, not an AIA trial. Per the parties' Joint Status Reports in Ramot v. Cisco, No. 2:19-cv-00225 (E.D. Tex.), at least eight reexaminations were filed on the '465, '535, and '866 patents (e.g., 90/014,526; 90/014,528; 90/014,606; 90/014,607; 90/014,608; 90/015,280; 90/019,333; 90/019,420). Office actions rejected all asserted claims, which prompted the district court to stay the case on 2021-01-13. None of the reexamination numbers I can identify targets the '425.

Strategic summary

Claim status of US 9,203,425. There is no claim that is CANCELED and none that is SUSTAINED-by-PTAB, because the '425 has never been before the Board. All 30 claims (including independent claims 1, 11, 15, and 24) remain UNTESTED. Nothing in the family's PTAB history narrows the '425. If you are being asserted the '425, you are litigating on a clean slate: no claim preclusion, no collateral estoppel, no Board claim construction to inherit.

Estoppel landscape. Because no IPR/PGR on the '425 ever reached institution, § 315(e)(2) estoppel does not attach to the '425 at all. The estoppel that exists is limited to Cisco as to the '465, '535, and '866 — and even there, Cisco escaped estoppel's practical bite because the Board denied institution (estoppel under § 315(e)(2) is generally understood not to arise absent a final written decision). The upshot for a current defendant: the entire prior-art field remains available against the '425, including Roberts, Ho, Burchfiel, and Mark — and, importantly, including art that the Board never substantively adjudicated. That is a double-edged fact worth internalizing: Ramot likewise has no IPR-driven prosecution-history disclaimers or Board findings on the '425 that it would have to defend; the '425's file history is an ordinary prosecution history.

Pattern signals.

1. Serial, energetic petitioner. Cisco filed at least six IPRs against this family — a November 2019/January 2020 wave (IPR2020-00122, -00123, -00484, all denied on Fintiv) and a 2022 wave (IPR2022-00575, -00576, -01283, which went to the Federal Circuit and were affirmed). Cisco also filed a 2021 Delaware declaratory-judgment action (No. 1:21-cv-01365) over US 11,133,872 and a 2022 DJ action over US 11,342,998. This is a defendant that exhausted every PTO avenue rather than settle.
2. No defensive aggregator on the '425. The Unified Patents PTAB Data attribution on the Google Patents page refers to Unified's data source, not Unified as petitioner. The real petitioner throughout is Cisco (a competitor-defendant, not a defensive aggregator). There is no RPX/Unified/Askeladden-style third-party IPR on the '425.
3. Patent owner litigates aggressively but has not faced a merits loss. Ramot (through Bunsow De Mory) won all three 2020 institution fights; the ex parte reexaminations forced a litigation stay, and the family remains a live portfolio. Ramot has separately hunted later continuations ('872, '998). The absence of an IPR on the '425 most plausibly reflects that Ramot chose to assert the continuations, not the '425 — not that the '425 is immune.

Caveat on the '425's own exposure. Because the '425 was never asserted, no third party has yet had the § 315(b) one-year clock or the incentive to build a petition against it. A defendant today should not read "zero PTAB proceedings" as "valid" and should not read "family got hit with six IPRs" as "the '425 is already invalid."

Recommended next steps

1. Correct the record first. If opposing counsel or a demand letter implies the '425 has been "upheld" or "challenged" at the PTAB, the precise truth is: no AIA trial has ever been instituted on the '425. Pull the USPTO PatentCenter / PTAB E2E record for the '425 (application 14/662,343) and confirm the "no proceedings" status; that is your baseline.
2. Attack grounds are wide open. With no § 315(e)(2) estoppel and no Board claim construction, a defensive IPR can be built fresh. Reuse the strongest art Cisco developed (Roberts, Ho, Burchfiel, Mark) but fix the defect the Board identified in IPR2020-00484 — the failure to explain how Roberts' DSP/non-linear compensator performs the claimed digital-to-digital mapping. The '425 claims make the DDC element central (claims 1 and 11 require M > N electrodes and a digital-to-digital converter; claim 15 recites the binary actuation vector), so the art must squarely meet that limitation.
3. Mind the Fintiv risk you'll now face in reverse. The 2020 denials were discretionary, and the PTAB's § 314(a) practice has since shifted (the June 2022 Director memo, and later changes). If you file, file early relative to any parallel case and consider a Sand Revolution-style stipulation not to press the same grounds in district court — that materially improved institution odds in the post-Fintiv era and directly counters the reasoning that killed Cisco's 2020 petitions.
4. Confirm the second-wave appeals before relying on them. The Rule 36 affirmance at Nos. 2024-1726/1727/1728 (2026-04-13) is confirmed; the target patents are not. Verify on PTAB E2E / the CAFC docket whether any canceled claim belongs to a patent you care about, and whether any of that reasoning (e.g., on the DDC limitation) transfers to the '425. A Rule 36 affirmance carries no precedential reasoning, so it will not collaterally estop you — but it may reveal what worked.
5. Watch the ex parte reexaminations. Outcomes in the '465/'535/'866 reexams can hand you examiner-adopted art and rejections for free. Track the Joint Status Reports in Ramot v. Cisco, No. 2:19-cv-00225 (E.D. Tex.), which have been filed regularly through at least 2025-12-11, including the '866 appeal awaiting a PTAB decision.

Sources: Google Patents – US9203425B2 · Unified Patents portal – IPR2020-00122 · In re Cisco Systems Inc., Fed. Cir. Nos. 2020-148/2047/2049 (2020-10-30) · IPR2020-00484 institution decision (PDF) · CAFC Rule 36 judgment, Nos. 2024-1726/1727/1728 (2026-04-13) · Joint Status Report, Ramot v. Cisco, No. 2:19-cv-00225 (E.D. Tex.)

Confidence flags: (a) The "no AIA trials on the '425" conclusion rests on the structured ODP block plus my searches — high confidence. (b) APJ panel identities for the 2020 denials — not retrieved; do not attribute names. (c) The patent targets of IPR2022-00575/-576/-01283 — unverified; the US 11,133,872 association comes only from a third-party litigation blog and should be confirmed before use.

I'll research the assignment chain and litigation history for US 9203425. Let me run several searches.

Let me dig into the assignment records and litigation details.

Let me look for the specific assignment reel/frame and any license records.

Ownership & Assignment Reconstruction — US 9,203,425 B2

Method note / data caveat: The USPTO Assignment Center (https://assignmentcenter.uspto.gov/) and the legacy Assignment Search (https://assignment.uspto.gov/patent/index.html) are JavaScript applications that could not be queried directly by the search tools available for this task. I therefore reconstructed the chain from the Google Patents legal‑events record embedded in the patent document, from PTAB/district‑court exhibits that quote the assignment instrument, and from court records. I could not verify reel/frame numbers or the correspondent-of-record, and I flag those fields explicitly below rather than inventing them. All statements about ownership are anchored to the court finding that "Ramot owns by assignment" the family patents.

Inventors

The three are named together in the assignment instrument quoted in the IPR record ("Inventor(s) Yossef EHRLICHMAN, Nazareth Ilit, ISRAEL; Ofer AMRANI, Tel Aviv, ISRAEL; Shlomo RUSCHIN, Herzliya, ISRAEL … Ramot at Tel‑Aviv University Ltd., Tel‑Aviv, ISRAEL; Assignment For Published Patent Application").

Unusual-pattern check: No fire-sale precursor. The inventors' rights were assigned to the university's technology-transfer arm (Ramot), which is the standard academic assignment flow — not a mass departure followed by a portfolio dump. I found no evidence that any inventor departed TAU within 12 months of filing, and no evidence of a later inventor-held or spin-out interest. (I could not confirm current employment status of each inventor post‑2015.)

Original assignee

• Entity on the issued patent: Ramot at Tel‑Aviv University Ltd. (Israel).
• What it is: Ramot is the technology-transfer company / "Business Engagement Center" of Tel Aviv University. Per the Delaware court record: "Ramot is the Business Engagement Center of Tel Aviv University ('TAU') … provides the resources, as well as the business and legal frameworks for inventions made by TAU's faculty, students, and researchers, protecting the discoveries with IP." Its Rule 7.1 disclosure lists Tel Aviv University Economic Corporation Ltd. and Tel Aviv University as corporate parents.
• Does it ship a product embodying the claims? No. Ramot is a non-manufacturing IP/licensing organization. The technology itself (a linearized optical DAC / multielectrode MZM) is a genuine electro-optics invention from TAU's engineering faculty; no commercial product was launched by the assignee.
• Primary line of business: University IP management, licensing, and spin-out formation (Ramot reports holdings in 100+ startups; establishes ~15 companies/yr).
• Current status: Operating / active. Not acquired, not dissolved, not in bankruptcy. It is an active, currently litigating patent owner (see timeline).

Assignment timeline

Only one recorded assignment could be identified for this patent, plus the chain of continuation filings that seed the family. No post-issuance transfer to an LLC, an NPE, or a defensive aggregator was found.

• Execution date unverified / recorded 2015‑05‑03 — Reel/Frame NOT VERIFIED (could not query Assignment Center)

  • Conveyance: Assignment (inventor → assignee)
  • Assignor: Yossef Ehrlichman; Ofer Amrani; Shlomo Ruschin (all three, jointly)
  • Assignee: Ramot at Tel‑Aviv University Ltd.
  • Correspondent: Not verified — I could not retrieve the recording attorney/firm. (No recurrence can therefore be assessed; see Signal 3.)
  • Context: Internal/institutional normalization — standard academic inventor‑to‑university technology‑transfer assignment, not an arm's‑length acquisition.
  • Anchor: Google Patents legal event "2015‑05‑03 Assigned to RAMOT AT TEL‑AVIV UNIVERSITY LTD." and the PTAB IPR record quoting the executed assignment for the published application.

• No further recorded assignments found. No record of (i) a transfer to a "…IP/Holdings/Licensing/Ventures" LLC, (ii) a security agreement, (iii) a merger/change-of-name, or (iv) a defensive aggregation. Independent corroboration of continuing Ramot ownership: the Delaware court expressly found "Ramot owns by assignment the '872 and '998 patents, as well as U.S. Patent Nos. 10,270,535 (the '535 patent) and 10,033,465 (the '465 patent)" (Cisco Sys., Inc. v. Ramot at Tel Aviv Univ. Ltd., C.A. No. 22‑674, D.I. 51, Nov. 14, 2022). The '465 and '535 patents are in the same family as US 9,203,425.

Note on family structure (not a transfer): US 9,203,425 is the grandparent of a large continuation family (e.g., US 10,033,465; US 10,270,535; US 11,133,872; US 11,342,998; …). These were created by continuation filings, not by assignment. All remain in Ramot's hands.

Timeline diagram

timeline
    title Ownership of US 9203425
    2007 : Priority application filed by TAU inventors
    2014 : Ramot sues Cisco in EDTX
    2015 : Inventors assign patent rights to Ramot
         : US 9203425 granted
    2019 : Ramot sues Cisco over family patents
    2020 : Cisco IPRs denied institution
    2021 : Ramot sues Acacia in Delaware
         : Cisco and Acacia file DJ action
    2022 : Ramot sues Cisco over later family patent

Supporting litigation anchors (not ownership transfers): first suit Ramot v. Cisco, 2:14‑cv‑01018 (E.D. Tex.), filed 2014‑11‑05, voluntarily dismissed 2015‑02‑24; second suit 2:19‑cv‑00225 (E.D. Tex.), filed 2019‑06‑12 (asserting '535 and '465), stayed 2021‑01‑13; Ramot v. Acacia, 1:21‑cv‑00295 (D. Del.), filed 2021‑02‑26, stayed 2021‑09‑03; Cisco/Acacia DJ action 1:21‑cv‑01365 (D. Del.), filed 2021‑09‑28 re the '872 patent; Ramot v. Cisco, 2:22‑cv‑00168 (E.D. Tex.), filed 2022‑05‑24 re the '998 patent, transferred to D. Del. as 22‑674.

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. The only recorded assignment (recorded 2015‑05‑03) runs from the three inventors to Ramot at Tel‑Aviv University Ltd., a university technology-transfer organization with identified corporate parents (Tel Aviv University Economic Corporation Ltd. / Tel Aviv University). No "IP/Holdings/Ventures" LLC, no registered-agent address, no single-member Delaware/Texas shell appears anywhere in the chain.

2. Known asserter in the chain — NOT PRESENT. Neither the original nor current assignee (Ramot) matches any public NPE list (Acacia Research, Marathon, IV, IPNav, Wi‑LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, etc.). Unified Patents classifies Ramot as a "University" patent owner, and petitioner Cisco as the "Large Operating Company" (Unified Patents PTAB record, IPR2020‑00122). A university asserting its own faculty's patents is not a cascaded-LLC NPE pattern.

3. Repeat correspondent across the chain — UNCLEAR / cannot assess. Only a single assignment exists, so there is no recurrence to measure, and I could not retrieve the recording correspondent's name or firm from the sources available. No finding either way. (Litigation-counsel appearances — Bunsow De Mory for Ramot; Haynes & Boone for Cisco — are law-firm engagements in litigation, not assignment-recording correspondents, and do not bear on this signal.)

4. Cascading transfers — NOT PRESENT. No chained LLC assignments; the chain is one link deep (inventors → Ramot) and has remained static. No shared-correspondent-address cluster because there is no cluster.

5. Pre-litigation transfer — NOT PRESENT. The single assignment (recorded 2015‑05‑03) is the foundational inventor→assignee assignment, roughly four years before the June 2019 infringement suit and long before the 2022 suits. It is not a suit-enabling transfer to an asserter; Ramot is the original assignee, not a downstream venue-shopping assignee.

6. Bankruptcy fire-sale — NOT PRESENT. No Chapter 7/11 of the assignee; TAU/Ramot are solvent and actively litigating.

7. Privateering — NOT PRESENT. Privateering requires an operating company to hand patents to an NPE that asserts on its behalf. Here the owner is the original innovating institution and asserts its own patents; there is no operating-company→NPE hand-off.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. Ramot retains ownership and is an active asserting party, so the patent is not neutralized.

Verdict

Operating-company assertion — i.e., the original-assignee / innovator asserting its own patent; not an NPE chain.

Justification (2–3 sentences): The sole recorded conveyance is the inventor→assignee assignment to Ramot at Tel‑Aviv University Ltd. (recorded 2015‑05‑03), and the Delaware court confirms Ramot "owns by assignment" the '425‑family patents — so the patent has not moved to any licensing shell, NPE, or aggregator, and every troll-pattern signal (1, 2, 4, 5, 6, 7, 8) is absent. The asserting party is the original innovating institution's technology-transfer arm, litigating against a real industry practitioner (Cisco/Acacia) on technology TAU developed. Caveat for precision: Ramot is a non-manufacturing university TTO, so it does not itself "ship products embodying the claims"; the accurate characterization is "original-assignee/university assertion," which sits in the non-NPE / non-defensive side of the taxonomy rather than the troll side.

Verify at: https://assignmentcenter.uspto.gov/ (search patent number 9203425) and https://assignment.uspto.gov/patent/index.html — I recommend confirming the exact reel/frame, the execution date, and the correspondent of record for the 2015‑05‑03 assignment, which I could not retrieve here.

I'll search for the patent's cited prior art references, using both the patent number and reference names.

The search returned a lot of noise from other documents that merely contain the digit string "9203425". Let me target the actual citation list for US 9,203,425 directly.

Let me try to retrieve the citation table from Justia and the patent's own front page.

Prior-Art Analysis — US 9,203,425 B2

Scope and method note (read first). The authoritative full text I was given for US 9,203,425 B2 does not include the front-page "(56) References Cited" / IDS table. My searches attempted to retrieve that table from Google Patents, Justia and USPTO materials, but a large fraction of the results were false positives — unrelated documents that merely happen to contain the digit string "9203425" as a foreign application number (e.g., French application FR 9203425 / Bio Mérieux; UK application GB 9203425D0; Mexican MX 9203425; PCT WO 9203425-A). Per the operating rule, I treat those identifiers literally and have excluded them; none relates to this patent.

Accordingly, this analysis is grounded in (a) the prior art the specification itself cites in its Background (authoritative full text supplied), (b) patent citations that surface on the faces of family members in the same specification (I did not retrieve the '425 face itself), and (c) family-level search-report data. I flag every point of uncertainty rather than fill gaps.

Governing law caveat. Earliest priority is June 13, 2007, so pre-AIA 35 U.S.C. § 102 applies. For a reference to anticipate under § 102 it must disclose all elements of a claim, "arranged as in the claim," in a single reference. Several references below are relevant mainly as § 103 art, not as anticipatory art.

References attributable to US 9,203,425 (or its family)

Note on timing: References #5 and #6 (2003, 2004) predate the earliest priority date (2007) by more than one year; #1–#4 well predate it. All are available as § 102(b) statutory-bar art if they qualify as printed publications/patents.

§ 102 assessment — can any single reference anticipate?

Bottom line: none of the identified references appears to anticipate the granted independent claims. The granted independent claims (1, 11, 15, 24) each contain a numeric limitation the cited art lacks, and most contain a structural limitation the cited art also lacks:

• Every independent claim requires the number of actuating electrodes to be strictly greater than the number of bits of the input data word (M > N). The '425 specification's Background expressly states that Papuchon's device used a conventional power-of-two sequence and that Leven/Chen's device uses "a single modulator for every 2 bits." Neither maps an N-bit word onto more than N binary electrode-actuation voltages. On the face of the record available to me, #1–#6 do not disclose M > N.
• Claims 1, 11, 15 and 24 require a binary actuation vector / digital-to-digital conversion in which the vector represents the data word — i.e., a non-one-to-one mapping of bits to electrodes. Papuchon (#1, #4) maps bits one-to-one to power-of-two electrodes, and Yacoubian (#5) drives one modulator per bit. Neither discloses the DDC mapping.
• Claims 11 and 24 require modulation of both power and optical phase delay (the QAM aspect). The cited art is directed to intensity/amplitude conversion; the phase/QAM teaching does not appear in #1–#6.

Consequently, #1–#6 are best characterized as § 103 prior art (teachings a skilled artisan might combine) rather than anticipatory art. Specifically:

• #1 / #4 (Papuchon) disclose the multi-electrode MZI modulator recited in claims 1, 4–7, 15, 18–20 — but with electrode lengths in a factor-of-two series and a one-to-one bit map, and without a DDC or M > N.
• #2 (Chen) and #6 (Leven) disclose a high-speed optical DAC architecture relevant to claims 1 and 15, but explicitly suffer from the non-linearity the '425 patent cures.
• #3 (HP '292) discloses an optical DAC relevant to claims 1 and 15; I could not verify its disclosure with enough specificity to assess M > N.
• #5 (Yacoubian) discloses electro-optic DAC using per-bit modulators, relevant to the multi-electrode/multi-modulator concept of claims 1 and 15.

Important additional point: the '425 Background's own characterizations of Papuchon, Yacoubian and Leven/Chen function as applicant admissions about the prior art, usable under § 102(a)/(b) and § 103 to establish the state of the art — independent of whether the documents were formally listed as citations.

Explicit uncertainties and non-findings

1. I did not retrieve the (56) References Cited table for US 9,203,425 itself. The references above are assembled from the specification's Background and from family-member faces/search reports. The '425 front page may list additional U.S. patents, U.S. patent-application publications, or foreign documents that I have not captured; I cannot rule those out.
2. A "Non-Patent Citations (4)" list appears on parent US 8,044,835, of which I could positively identify Leven (2004), Papuchon (1980) and Yacoubian (2003); the fourth entry was not retrieved and is unconfirmed.
3. Title of US 7,212,292 B2 was not fully captured (the search snippet truncates the assignee string to "Hewlett-Packard Development Company, L…" and its title). I have not verified its exact title and therefore do not assert one.
4. I did not query USPTO Patent Center / the actual PTO-892 in application 14/662,343, so the examiner-cited references for the '425 patent are not confirmed by me. What appears in a WEST search history for a sibling case (15/298,327) — e.g., the query string "9203425.pn." returning 12 hits — is a search log, not a citation list, and I do not treat it as citations.
5. No element-by-element comparison against full reference texts was possible. My § 102 conclusions rest on the characterizations in the '425 specification and on family documents, not on full-text review of each reference; a definitive anticipation analysis requires the complete text of each reference and a full claim chart.
6. The current-task header states April 26, 2026, while the system date given to me is 2026-09-24; the prior-art analysis above is unaffected by this discrepancy.

If you can supply the patent's actual (56) References Cited table (or the PTO-892 from application 14/662,343), I will produce a complete, citation-by-citation claim chart.

I'll ground this analysis in the prior-art references identified on the patent record and the specification's own background section, plus verify the family's post-grant proceedings that may reveal additional art.

Let me verify the identities of the IPR-grade references (Roberts, Taraschuk, Wright) so I cite them accurately rather than by surname only.

Obviousness Analysis — U.S. Patent 9,203,425 B2 under 35 U.S.C. § 103

Scope note. This builds on the bibliographic, claim-characterization, and litigation sections already produced. I do not repeat the claim summaries; I refer to claims by number and rely on the element mapping established there. One date discrepancy to flag: the task header states "Current Date: April 26, 2026," while the session header states 2026-09-24. Both post-date the April 13, 2026 CAFC judgment discussed earlier, so nothing turns on it, but I note it for the record.

Governing law. The '425 patent's claims are entitled, at the earliest, to a June 13, 2007 priority date (Prov. 60/943,559). The claims therefore arise under pre-AIA 35 U.S.C. § 103(a) — which is the ground the family's IPRs actually pleaded ("pre-AIA 35 U.S.C. § 103(a)"). The controlling framework is Graham v. John Deere (scope/content of prior art; differences; PHOSITA level; secondary considerations) as refined by KSR Int'l v. Teleflex, 550 U.S. 398 (2007) (predictable combinations; "known technique to improve similar devices"; "design incentive").

1. The "Prior Art" content on the record

The prior-art material available for this patent falls into two tiers, and the distinction matters:

Tier 1 — art cited on the face of the '425 patent / in its own background (Prior Art date 2007-06-13; prior-art keywords: electrode, actuating, binary, modulator, actuation):

• Papuchon et al., U.S. Pat. No. 4,288,785 ("Light intensity modulator with digital control and a digital-analog converter comprising such a modulator," Thomson-CSF). https://patents.google.com/patent/[US4288785](/patent/US4288785)
• Yacoubian & Das, "Digital-to-analog conversion using electrooptic modulators," IEEE Photon. Technol. Lett., 15(1):117–119, Jan. 2003.
• Leven et al., "A 12.5 Gsample/s optical digital-to-analog converter with 3.8 effective bits," LEOS 2004, 1:270–271, the subject of Chen et al., U.S. Pat. No. 7,061,414 ("Optical digital-to-analog converter"). https://patents.google.com/patent/[US7061414](/patent/US7061414)

Tier 2 — art that Cisco actually ran against the family sibling U.S. 11,133,872 (IPR2022-00575/-00576/-01283), surfaced in the search (full written decisions affirmed per curiam, Fed. Cir. 24-1726/-1727/-1728, Apr. 13, 2026):

• Roberts, U.S. Pat. No. 7,277,603 B1, "Integrated optical waveform modulation" (Nortel; inventor Kim B. Roberts). I confirmed this is the reference the petitions call "Roberts (EX1005)." https://patentimages.storage.googleapis.com/cb/6d/c8/9bb5b225414b33/[US7277603](/patent/US7277603).pdf
• Taraschuk (EX1006), described by Cisco as "a linearizer for an optical modulator that maps an input signal to a larger bit output signal" (6-bit in → 8-bit out). I could not confirm the Taraschuk patent/publication number — I will cite it by name and flag the identifier as unverified.
• Wright (EX1009) — relied on in IPR2022-01283; number unverified.

Provenance caveat (important): Tier-2 references were applied by the Board to the '872 patent's claims, not to the '425 patent. Their persuasiveness against '425's specific claim wording (e.g., "binary actuation vector," "optical output port," "M strictly greater than N") is my analysis, not a Board holding. I flag every place where I extrapolate.

2. Level of ordinary skill (PHOSITA)

A POSITA here is a person with an M.S. (or equivalent experience) in electrical/optical engineering and 2–3 years in integrated-optics modulator design, familiar with: Mach-Zehnder modulator transfer functions (the cos² response), multi-electrode "digital" modulators (Papuchon), photonic DACs (Yacoubian; Leven/Chen), and transmitter-side digital predistortion/look-up-table linearization (Roberts; Taraschuk). This is the level the family's own IPR expert (Dr. Blumenthal) applied and that the patent's own specification presupposes when it says a DDC "may readily be implemented using commercially available high-speed ASICs."

3. What the primary references disclose

Papuchon (US 4,288,785) is the closest art and is admitted in the '425 specification as the origin of the multi-electrode MZI DAC concept. It discloses essentially the entire device architecture of the independent claims except the two limitations at issue:

• A two-arm (Mach-Zehnder) interferometer DAC driven by an n-bit control word, with binary control voltages ("two states: state 0 … and state 1 … at a predetermined value v common to all the signals");
• Multiple actuating electrode sets disposed on/along the waveguide arms (symmetric pairs E₂ᵢ/E₃ᵢ with a common central ground electrode E_M — i.e., electrodes on both branches);
• A photodetector on the output guide for the optical→electrical DAC output;
• Explicit recognition of the linearity problem, including a proposal to "associat[e] therewith a feedback loop for linearizing its response";
• What it does NOT disclose: it uses power-of-two electrode lengths with direct bit-to-electrode mapping, so M = N; and it has no digital mapping/DDC in which an electrode is actuated as a function of more than one input bit.

Roberts (US 7,277,603) supplies precisely the missing pieces:

• A digital signal processor 34 that "receives an input data signal x(m) and produces two multi-bit sample streams V_L(n) and V_R(n) representative of the desired phase modulation to be applied to each branch of an MZ modulator 4";
• A "non-linear compensator 18" whose multi-bit sample values "can be computed taking into account non-linearities of … the MZ modulator 4, such that … the output of the MZ modulator 4 closely matches the target E-field modulation" — i.e., a transmitter-side digital pre-distortion engine, functionally a DDC;
• Logic circuits 46 that map each multi-bit stream to a 7-bit digital drive signal S_x(n) on a multi-bit parallel bus, "each line of which is connected to a respective electrode 40 of a dual-branch MZ modulator." The petitions characterize this as converting "x(m) data (containing between two and six bits of data, N>1) to two 7-bit S_x(n) signals." That is literally more electrodes driven than input bits (M > N) — and with differing signals on the two branches;
• The I(n)/Q(n) computation framework, indicating complex-amplitude (QAM-capable) modulation.

Taraschuk supplies an express look-up-table linearizer mapping a 6-bit input to an 8-bit output — an independent, explicit teaching of the "map N bits to M > N bits to linearize an MZ modulator" concept.

Yacoubian (2003), Leven (2004)/Chen (US 7,061,414) establish the field's trajectory toward higher effective bits and disclose binary electrical drive of electro-optic phase modulators with coherent summing to synthesize analog output (Leven/Chen: "If the control voltages are now configured such that each of them can have two different levels … 2^i output current levels are realizable") — supporting the phase/power and PAM/QAM dependent claims.

4. The obviousness case against the independent claims

4a. Claim 1 (circuit; M > N; DDC presenting a binary actuation vector)

Primary combination: Papuchon + Roberts. Papuchon provides the claimed device/platform; Roberts provides the electrode-actuating device, the DDC-equivalent non-linear compensator, the M > N drive, and the per-branch patterns. The two are in the same field (optical communications / photonic DACs), address the same problem (MZ non-linearity), and Roberts computes its drive streams for exactly the kind of MZ modulator Papuchon built. The combination is the KSR paradigm of "applying a known technique (digital predistortion) to a known device (multi-electrode MZ DAC) ready for improvement to yield predictable results."

Alternative/clarifying combination: Papuchon + Taraschuk for the M > N and mapping limitations, and Papuchon + Roberts + Taraschuk for the full set (Taraschuk's LUT being the concrete form of the DDC).

4b. Claim 11 (modulator system; waveguide branches; binary voltages modulate power and optical phase delay)

Papuchon's mechanism is phase-delay modulation in the branches whose interference produces power modulation — so a POSITA reads the "power and optical phase delay" language on Papuchon's own electro-optic phase-shift operation. Roberts expressly generates streams "representative of the desired phase modulation to be applied to each branch" and drives a dual-branch MZ modulator whose output E-field (amplitude and phase) tracks a target. Yacoubian and Leven/Chen confirm coherent/phase-domain photonic DACs are known. The M > N limitation is again met by Roberts (7-bit drive vs ≤6-bit input).

4c. Claim 15 (optical modulator; binary voltages "applied digitally"; M > N; actuating device provides the binary actuation vector)

Same mapping as claim 1. "Applied digitally / direct digital driving" is disclosed by Papuchon's binary multi-electrode drive and by Roberts' parallel-bus digital drive of individual electrodes (no mediating analog DAC between DSP and electrodes in the '603 logic-circuit path).

4d. Claim 24 (method; M > N; modulate power and phase delay)

Method counterpart of 4a+4b; obvious over Papuchon (receiving word; passing light through MZ; binary drive) in view of Roberts (digital mapping to a larger number of electrodes; per-branch phase control).

5. Dependent claims

• Cl. 2 / 16 (semiconductor modulator): Papuchon (LiNbO₃), Yacoubian (EO polymer), Chen '414 (InP/GaAs/LiNbO₃) — semiconductor/EO modulators conventional.
• Cl. 3 / 17 (light generation device): routine implementation choice (LED/laser DACs well known).
• Cl. 4 / 18 (phase-only), Cl. 5 / 19 (intensity-only): Papuchon (intensity); Roberts (phase).
• Cl. 6 / 20 (both phase and intensity): Roberts' complex/E-field output; Chen '414 coherent summation.
• Cl. 7 (DDC selects among available actuation vectors): the ordinary operation of a look-up-table DDC (Taraschuk).
• Cl. 8 / 22 / 27 (PAM): inherent in Papuchon's multi-level intensity output.
• Cl. 9 / 21 / 28 (QAM): Roberts' I/Q sample streams and Chen/Leven's coherent complex modulation.
• Cl. 10 / 13 / 14 / 23 (differing actuation patterns on first vs. second branch): Roberts discloses different V_L(n)/V_R(n) and S_x(n) per branch — directly met.
• Cl. 12 (PAM enabling QAM): Roberts/Yacoubian/Leven.
• Cl. 25–28: method analogues of the above.

6. Motivation to combine — articulated (KSR)

1. Same field, same problem. Papuchon and Roberts are both optical-communications modulator art; both expressly confront the MZ modulator's non-linear input→output transfer function.
2. The specification's own admissions. The '425 background concedes the cos² non-linearity problem, that "analog pre-distortion" is one of the "common solutions," and that a fast multi-bit DAC is needed — i.e., the inventor admits the pre-distortion/linearization goal and technique were known.
3. Papuchon itself points to linearization ("feedback loop for linearizing its response"; reduced-range biasing).
4. A recognized deficiency in the art. Leven reports only 3.8 effective bits for a 6-bit design — a documented, quantified shortfall that gives a concrete reason to improve linearity/ENOB.
5. Predictable result. Replacing Papuchon's fixed bit-to-electrode wiring with a look-up-table/non-linear-compensator mapping to a redundant electrode set is the known remedy for a known non-linearity, with a reasonable expectation of success.
6. Combinability / no change in principle of operation. Roberts' drive streams are designed for a multi-electrode dual-branch MZ modulator of Papuchon's type; only the mapping changes.
7. Motivation to use M > N specifically. Extra electrodes give the extra degrees of freedom needed to place output levels closer to the ideal line (and Taraschuk expressly uses a larger-bit output to linearize).

7. Where the obviousness case is weakest (and Ramot's likely rebuttal)

I want to be candid about the soft spots rather than present a one-sided case:

1. The M > N crux. Neither Papuchon nor Yacoubian (one MZM per bit) nor Leven/Chen (one modulator per bit/two bits) inherently teaches M > N for a single multi-electrode MZI. M > N comes from Roberts and Taraschuk, and Roberts' drive signals are multi-bit digital rather than the "binary actuation vector" the '425 claims recite. Ramot argued in the family prosecution that Taraschuk "only discusses mapping an M-bit digital signal into an N-bit parallel digital signal, where M can be less than N." If a fact-finder accepts that framing, the M > N limitation loses its cleanest mapping and the case leans harder on Roberts' 7-bit-vs-≤6-bit drive.
2. "Binary" vs. multi-bit tension. The '425 claims recite a binary actuation vector (on/off per electrode). Roberts' inputs to the electrodes are binary (each electrode on the shared bus; "each active electrode receives the same voltage"), so this is defensible — but a POSITA could read Roberts as a multi-level drive, which the patent's own spec treats as a different option ("switching between different distinct voltage levels"). This is a genuine § 103-vs-§ 102-style fit issue.
3. No Board holding on these exact claims. As flagged in §1, the Board's decisions (and the Fed. Cir. Rule 36 affirmance) went to the '872 patent. No institution was reached on '425 itself (IPR2020-00122 was denied on Fintiv/procedural grounds — not on the merits), so there is no adjudicated obviousness ruling on '425's claims to cite.
4. Priority-date exposure cuts both ways. If Ramot's 2007 priority holds, post-2007 art (e.g., the inventors' own Ehrlichman 2008 papers, which Cisco sought to admit) is not prior art. If priority fails for any claim, those papers become prior art — a separate validity theory Cisco litigated in E.D. Tex. and that I have not independently resolved.
5. Secondary considerations. The specification's commercial framing (CATV, wireless base stations, defense) could be argued as long-felt need / industry adoption, though I have no evidence (licenses, praise, copying) on this record to support a nexus.

8. Bottom line

• Strongest § 103 theory (independent claims 1, 11, 15, 24): Papuchon (US 4,288,785) in view of Roberts (US 7,277,603), optionally further in view of Taraschuk. Papuchon supplies the multi-electrode MZ DAC with binary drive, electrodes on both branches, and an optical output port; Roberts supplies the digital (DDC-equivalent) non-linear compensator, the per-branch multi-bit drive, M > N, and complex/phase modulation; Taraschuk supplies an express look-up-table N→M>N linearizer mapping. Motivation is grounded in the same field, the same non-linearity problem, Papuchon's own linearization suggestion, the spec's admission that pre-distortion is "common," and the art's quantified ENOB shortfall.
• Dependent claims 2–10, 12–14, 16–23, and 25–28 fall readily over these references (and Yacoubian 2003 / Leven 2004 / Chen US 7,061,414 for the phase, PAM, and QAM features).
• Principal vulnerability of the case: the "binary actuation vector with M strictly greater than N" combination is the crux; it is met cleanly by Roberts but not by the classic Papuchon/Yacoubian/Leven references, and no tribunal has yet adjudicated it against '425's specific claim wording.

Verification flags. (1) "Taraschuk" and "Wright" reference numbers are unconfirmed — I have only Cisco's petition narrative; treat the identifiers as unverified. (2) The Tier-2 references were applied to U.S. 11,133,872, not '425. (3) Claims 29–30 of '425 remain unavailable in the retrieved text (per the earlier section) and are not analyzed. (4) The '425 priority-date question is unresolved on this record.

Sources: https://patents.google.com/patent/[US9203425](/patent/US9203425)/en · https://patents.google.com/patent/US4288785 · https://patents.google.com/patent/US7061414 · https://patentimages.storage.googleapis.com/cb/6d/c8/9bb5b225414b33/US7277603.pdf · IEEE Photon. Technol. Lett. 15(1):117–119 (Yacoubian) · LEOS 2004, 1:270–271 (Leven) · PTAB IPR2022-00575/-00576/-01283 petition materials (ptacts.uspto.gov) · Fed. Cir. 24-1726 (Apr. 13, 2026).