US 7336605

I'll search for this specific patent number and check for any litigation activity.

Let me verify the litigation details and check for any CAFC 2026 activity involving this patent.

I have solid results. Let me confirm whether there is any CAFC 2026 appeal specifically involving this patent, and identify other defendants asserting it.

US Patent 7,336,605 — Summary of Findings

Search scope note

I searched for the literal identifier 7,336,605 / US7336605B2 and did not substitute similar numbers. All results below correspond to that patent. Sources checked: Google Patents (full text supplied), Espacenet bibliographic record, Unified Patents litigation portal, CourtListener/Justia docket records, and govinfo.

Important caveat on dates: you set today's date as April 26, 2026, but my live searches returned docket activity dated June 2, 2026 (and later). I am reporting what the sources show and flagging the discrepancy rather than silently reconciling it.

Bibliographic data (literal)

Espacenet lists the inventor order as MOR Gal; BRUCKMAN Leon; MENAHEMI Zvi, and shows applicant names "CORRIGENT SYSTEMS, INC." and "CORRIGENT SYSTEMS, LTD." The assignment record filed May 13, 2003 names CORRIGENT SYSTEMS, LTD. as assignee. There is also a pre-grant assignment naming "Corrigent Corrigent Ltd" / a "CHANGE OF NAME" from Corrigent Systems Ltd. to Orckit-Corrigent Ltd. (effective 2009), then ORCKIT IP, LLC (2016), then Enhanced Data Streaming LLC (2025).

Abstract (as issued)

A method for establishing a connection with a guaranteed bandwidth for transmitting data over a logical link that includes a plurality of parallel physical links between first and second endpoints. A link bandwidth is allocated on each of the physical communication links so as to include a predefined safety margin, based on either a failure protection policy, or a measure of fluctuation that occurs in a rate of data transmission over the physical links, or both. A sum of the allocated link bandwidth over the plurality of the parallel physical links is substantially greater than the guaranteed bandwidth of the connection. The data are conveyed over the logical link by distributing the data for transmission among the physical links in accordance with the allocated link bandwidth.

Independent claims — plain language

The granted patent contains 18 claims, with two independent claims: claim 1 (method) and claim 9 (apparatus). Note: the pre-grant publication (US 2004/0228278 A1) had a different, broader claim set (26 claims; independents 1, 11, and 19), including claims expressly directed to a "protection policy." The granted claims are directed to the fluctuation-measure embodiment. I flag this because the difference between the published and granted claim scopes matters for infringement/invalidity analysis.

Independent Claim 1 (method):

1. Define a logical link made of several parallel physical links between two endpoints.
2. Determine a measure of fluctuation in the data transmission rate over those physical links when data is distributed among them — where the fluctuation is expressly attributed to non-uniform distribution of load among the parallel links in the logical link.
3. Allocate a link bandwidth on each physical link that includes a predefined safety margin based on that fluctuation measure, such that the sum of allocated link bandwidths across the parallel links is substantially greater than the connection's guaranteed bandwidth (i.e., deliberate over-allocation / headroom).
4. Convey the data by distributing it across the physical links in accordance with the allocated bandwidth.

Independent Claim 9 (apparatus):
The device counterpart of claim 1 — (a) a controller that receives the fluctuation measure (again tied to non-uniform load distribution among the parallel links) and allocates per-link bandwidth with a safety margin based on it, so the summed allocation exceeds the guaranteed bandwidth; (b) a distributor that determines how data is distributed across the links per the allocated bandwidth; and (c) data transmission circuitry that transmits per the distribution.

Key dependent claims: 802.3 link aggregation (2, 10); ATM/IMA (3, 11); PPP multilink (4, 12); hash function applied to frame headers to select a link (5, 13); allocation of no less than F·B/N where F derives from the fluctuation measure (6, 14); fluctuation measured as standard deviation (7, 15) or as the difference between average link utilization and the maximum link utilization (8, 16); and (17, 18) a main card + multiple line cards connected by traces, with concentrators, where the first and second physical links attach to different line cards.

Practical gist: rather than splitting guaranteed bandwidth evenly (B/N per link), the patent sets aside a statistical/over-provisioning margin per link so that traffic skew from hashing-based load distribution (and, in the broader disclosure, link or line-card failure) does not starve the guaranteed service.

Litigation / docket findings

• Active district court case: Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG, U.S. District Court, Eastern District of Texas (Marshall Division), Judge Rodney Gilstrap. Filed October 10, 2025. The complaint asserts five patents, including 7,336,605 (the "'605 Patent"), alongside U.S. Patent Nos. 9,185,151 (reissued as RE50,398), 6,891,855, 7,660,234, and 7,032,135.
• Status: Cisco's motion to dismiss EDS's indirect/willful infringement claims (Dkt. 26) was DENIED in a Memorandum Opinion and Order signed June 2, 2026. The Unified Patents portal lists the case as Open, plaintiff classified as an NPE.

On CAFC 2026 dockets: I did not find any Federal Circuit appeal specifically involving 7,336,605. The case above remains at the district-court stage. I want to be explicit that this is a negative finding from the searches I ran — it is not proof that no appeal exists, and a CAFC appeal (if any) may not yet be indexed or may not name the patent number in searchable text.

Uncertainty / limitations

• I could not independently re-verify the docket beyond the June 2, 2026 order; later developments (claim construction, trial date, IPR petitions) are not reflected in the results I retrieved.
• The "current date" you provided (April 26, 2026) conflicts with the June 2026+ items my searches returned. I am reporting the search results as ground truth per your instruction, but you may want to confirm which date framing governs your analysis.
• Assignee data is stated "as listed" by Google Patents and the assignment records; the security agreement to WYE IP Capital LLC is a security interest, not an outright assignment.
• Claim-scope statements are based on the full granted text supplied and are reliable; the published-application claim set (26 claims) is distinct and should not be conflated with the granted 18-claim set.

Primary sources: https://patents.google.com/patent/US7336605/en · https://worldwide.espacenet.com/publicationDetails/biblio?CC=US&NR=7336605B2 · https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A25-cv-01038 · https://www.govinfo.gov/content/pkg/USCOURTS-txed-2_25-cv-01038/pdf/USCOURTS-txed-2_25-cv-01038-0.pdf · https://www.courtlistener.com/opinion/10893882/enhanced-data-streaming-llc-v-cisco-systems-inc/

I'll search for litigation involving US patent 7336605 specifically.

I found the key case. Let me search for any additional cases involving this patent.

Let me check for additional defendants and any earlier litigation involving this patent.

Litigation Involving U.S. Patent No. 7,336,605

Based on my searches (Unified Patents litigation portal, CourtListener/RECAP, Justia dockets, GovInfo, and Darts-IP family litigation data), here is what I can confirm about litigation specifically involving U.S. Patent No. 7,336,605 ("Bandwidth allocation for link aggregation").

Confirmed Case

Enhanced Data Streaming LLC v. Cisco Systems, Inc.

• Plaintiff: Enhanced Data Streaming LLC (EDS) — a patent assertion entity / NPE
• Defendant: Cisco Systems, Inc.
• Jurisdiction: U.S. District Court for the Eastern District of Texas, Marshall Division
• Case No.: 2:25-cv-01038-JRG (Judge Rodney Gilstrap)
• Filing Date: October 10, 2025 (Unified Patents lists the filing date as October 9, 2025; the complaint was filed on October 10, 2025 per the court's order)
• Asserted Patents: The complaint asserts five patents, of which the '605 Patent is one:
  • U.S. Patent No. 9,185,151 (reissued April 22, 2025 as RE50,398)
  • U.S. Patent No. 6,891,855
  • U.S. Patent No. 7,336,605 (the "'605 Patent")
  • U.S. Patent No. 7,660,234
  • U.S. Patent No. 7,032,135
• Status / Outcome to date: Open. The docket shows:
  • Cisco filed a motion to dismiss the indirect and willful infringement claims on December 19, 2025 (Dkt. 17).
  • EDS filed a First Amended Complaint on January 16, 2026 (Dkt. 22).
  • Cisco filed a renewed motion to dismiss the indirect/willful claims on January 30, 2026 (Dkt. 26).
  • On June 2, 2026, Judge Gilstrap issued a Memorandum Opinion and Order DENYING Cisco's motion to dismiss. The court found EDS adequately pled pre-suit knowledge (relying, inter alia, on a March 20, 2017 letter from prior owner Orckit IP, LLC, allegedly accompanied by a "Schedule of Orckit IP's Patent Portfolio" listing the asserted patents) and sufficient specific intent for post-suit induced infringement.
  • The case remains pending as of the available records.

Sources:

• Unified Patents litigation portal: https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A25-cv-01038
• Memorandum Opinion and Order (GovInfo): https://www.govinfo.gov/content/pkg/USCOURTS-txed-2_25-cv-01038/pdf/USCOURTS-txed-2_25-cv-01038-0.pdf
• CourtListener: https://www.courtlistener.com/opinion/[10893882](/patent/10893882)/enhanced-data-streaming-llc-v-cisco-systems-inc/
• Justia docket: https://docs.justia.com/cases/federal/district-courts/texas/txedce/2:2025cv01038/[241096](/patent/241096)
• Google Patents litigation link (case 2:25-cv-01038): https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A25-cv-01038

Earlier Enforcement Era (Not Confirmed as to the '605 Patent)

The Google Patents record for the '605 Patent indicates a prior enforcement period under the previous owner, Orckit IP, LLC (a subsidiary of Orckit-Corrigent). Orckit IP appears to have asserted a portfolio of its patents against networking companies around 2017–2018 (e.g., a February 1, 2018 suit by Orckit IP, LLC against Extreme Networks, referenced in Extreme's SEC filings; and the March 20, 2017 portfolio-notice letter from Orckit IP to Cisco referenced in the EDS v. Cisco order).

Caveat: I could not verify that the '605 Patent itself was specifically asserted in those earlier Orckit-era suits. The available sources confirm the '605 Patent was listed in Orckit IP's portfolio schedule, but I have no confirmed record of a prior complaint naming the '605 Patent as an asserted claim. I am flagging this as unconfirmed rather than asserting it as fact.

Summary

• Number of confirmed lawsuits naming the '605 Patent: One — Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG (E.D. Tex., filed October 10, 2025), currently pending (Cisco's motion to dismiss denied June 2, 2026).
• I found no other confirmed litigation specifically involving U.S. Patent No. 7,336,605.
• You should be aware that any assertion regarding earlier Orckit IP, LLC enforcement activity as it relates to the '605 Patent specifically is unverified and should be confirmed against the original complaints before being relied upon.

Note on identifiers: I have reported the patent number as 7,336,605, the case number as 2:25-cv-01038-JRG, and the related patents exactly as they appear in the sources (including the reissue RE50,398 and the other four asserted patents), without correction.

PTAB / AIA Trial Analysis — U.S. Patent No. 7,336,605

Structured source of record (provided): "The USPTO ODP API returns no AIA trial proceedings for this patent as of the most recent ingest."
Independent verification (this session): multiple targeted web searches (patent number in IPR/PGR/CBM contexts; petitioner-name searches; Orckit-era enforcement searches) surfaced no IPR, PGR, CBM, or derivation proceeding naming the '605 Patent. No proceeding number exists to report.

Contradiction check against prior sections: None. The "PTAB proceedings on file" block and the Litigation section are consistent — the '605 Patent has been asserted in litigation but never challenged at the PTAB.

Proceedings overview

Total AIA trial proceedings on U.S. Pat. No. 7,336,605: 0 (active: 0 · claims invalidated: 0 · claims sustained: 0 · settled: 0 · institution denied: 0).

Bottom line for a defendant: there is no PTAB record at all on this patent — no claims have been canceled, and no claims have been affirmatively upheld either. All 18 claims (independent claims 1 and 9, with dependents 2–8 and 10–18) are untested before the Board. You cannot point to a Final Written Decision and say "claim 1 is dead"; conversely, the patent owner cannot point to an FWD and say the claims survived a validity attack. The absence of any IPR after 20+ years of enforceability, followed by assertion in Enhanced Data Streaming LLC v. Cisco Systems, Inc., is itself informative: this is a patent that challengers have chosen not to attack at the PTAB, even though sibling Orckit patents in the same portfolio were attacked repeatedly and the '605 Patent was asserted against Cisco on 2025-10-10.

Critical independent fact that changes the defensive calculus regardless of PTAB: per the patent's legal-status record, the '605 Patent expired on 2025-12-01 ("Expired – Lifetime"; "Adjusted expiration 2025-12-01"). (Treat the status field as an assumption, not a legal conclusion, per the source's own disclaimer.) A patent that has expired cannot be the basis for prospective injunctive relief, and any recovery is backward-looking only. IPR of an expired patent remains legally available, but the practical payoff is limited to neutralizing past damages.

Proceedings

There are no proceedings to template. Rather than fabricate proceeding numbers or outcomes, the section below documents the adjacent-family PTAB activity that a defendant will inevitably encounter when researching the '605 Patent — carefully labeled as not proceedings on the '605 Patent.

(No proceeding) — No petitioner has ever challenged U.S. Pat. No. 7,336,605 at the PTAB

• Type: —
• Filed: —
• Status: No AIA trial proceeding on file (USPTO ODP, as provided).
• Judge panel: —
• Petition grounds: —
• Institution decision: —
• Final Written Decision: —
• Settlement / termination: —
• Appeal: —
• Defensive value: Neither side has PTAB ammunition on this patent. Any validity fight on the '605 Patent will be fought on the pleadings/discovery record in the district court under the clear-and-convincing standard, not on an IPR record.

Adjacent context (NOT proceedings on the '605 Patent — do not cite as such)

The following are sibling Orckit-family patents whose PTAB history is often surfaced by automated patent tools when you search the '605 family. They are different patents and their outcomes have no estoppel or preclusive effect on the '605 Patent:

The single most important cross-reference for a defendant: the primary reference relied on against the sibling '740 patent is "Bruckman" — U.S. Patent Application Publication No. US 2004/0228278 A1, titled "Bandwidth Allocation for link Aggregation," naming Bruckman, Mor, and Menahemi — i.e., the pre-grant publication of the '605 Patent itself (application 10/436,516, filed 2003-05-13). In the '740 proceedings, the Board found Bruckman-based grounds to present "compelling merits." That reference is not prior art to the '605 Patent (it is the same application), but it tells you exactly how the Office and petitioners have treated this specification when it appears as art against other claims — a data point, not a ground.

Strategic summary

Claim status — all 18 claims are UNTESTED. Nothing has been canceled and nothing has been sustained. Claims 1–18 of the '605 Patent retain the statutory presumption of validity intact and unprejudiced by any PTAB record. If you are a defendant, there is no shortcut: you cannot move to dismiss or for summary judgment on the strength of an FWD, because none exists.

Estoppel landscape — clean slate. Because no IPR/PGR was ever instituted, 35 U.S.C. § 315(e)(2) estoppel is inapplicable to anyone on this patent. No petitioner, real party in interest, or privy is barred from raising any § 102/§ 103 ground in district court. Conversely, this also means the patent owner has never had to narrow the claims via a Board-driven disclaimer or amendment. The full field of prior art cited of record remains available as a starting point (front-page citations include US 5,933,422; US 6,032,194; US 6,370,121; US 6,562,231; US 6,626,157; US 6,678,241; US 6,757,286; US 6,763,025; US 6,801,506; US 6,826,147; US 6,826,158; US 6,922,394; US 6,934,259; US 7,158,486; US 7,184,402; EP 1 052 808 A1; WO 00/74318 A1), plus the 15 non-patent citations (IEEE 802.1D/802.1W, IEEE 802.3 Clause 43 (2002), RFC 2892, RFC 1990, RFC 3031, Dziong et al., IMA Strategic Technologies Group (2001)).

Timing — this is the live issue. The patent expired 2025-12-01, six weeks after EDS filed suit (2025-10-10). Two consequences: (i) any IPR petition now would be against an expired patent (still permissible, but the remedy sought is past damages, and claims are construed under Phillips); and (ii) the § 315(b) one-year bar runs from service of the complaint, which I have not independently confirmed — if Cisco was served in or about late October 2025, its § 315(b) window closes in or about late October 2026, i.e., imminently. I flag the service date as unverified.

Pattern signals. (1) Same-petitioner risk: Cisco has demonstrated willingness to file IPRs against Orckit-family link-aggregation patents (IPR2023-00401), so a Cisco petition on the '605 Patent would fit an existing pattern — but none has been filed to date. (2) No defensive aggregator: no Unified Patents (or similar) filing is on record for this patent; the Google Patents litigation entry reflects a court case, not a PTAB case. (3) Patent-owner posture: Orckit IP, LLC has actively litigated portfolio-wide (including a 2018-02-01 suit against Extreme Networks in Düsseldorf over EP 1 958 364 B1, and the 2017-03-20 / 2018-07-11 portfolio-notice letters to Cisco), and Orckit-family patents have been carried to Federal Circuit appeal — so an owner-side appeal culture exists even though it has never touched the '605 Patent.

Recommended next steps

If you are defending against assertion of the '605 Patent:

1. Do not represent to a court or a counterparty that any claim of the '605 Patent has been invalidated. No PTAB proceeding exists. Any statement to the contrary would be false and sanctionable. The correct statement is: "U.S. Patent No. 7,336,605 has never been subject to an AIA trial proceeding; all 18 claims are untested at the Board."
2. Lead with expiration. Confirm the 2025-12-01 expiry against the USPTO Patent Center and the maintenance-fee record (12th-year fee paid 2019-08-01 as a small entity) before relying on it. If confirmed, the exposure is past damages only — focus discovery and damages analysis on the recovery window and non-infringing alternatives rather than injunctive defenses.
3. If you are considering an IPR, move now. Confirm the actual § 315(b) service date. If it is in October 2025, the petition deadline is in or about October 2026 — days away as of today (2026-09-28). Note that Director Squires's 2025-10-16 memorandum centralizes institution with the Director, and the October 2025 proposed rules (not yet effective) would channel parallel § 102/§ 103 arguments; account for both in strategy. For an expired patent, weigh whether the past-damages exposure justifies the petition cost.
4. Read the sibling-petition record for art framing, not for estoppel. The Bruckman-based challenges to U.S. Pat. No. 7,545,740 (IPR2023-00401, IPR2024-00026, IPR2024-01237) show how the Board has evaluated link-aggregation claim limitations and how it has treated § 325(d) arguments on cumulative art in this family. That is useful for framing obviousness combinations against claims 1 and 9 — nothing more.
5. Track the district court. Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG (E.D. Tex.) is pending; Cisco's Rule 12(b)(6) motion was denied 2026-06-02 (Dkt. 48), and pre-suit knowledge was sustained on the strength of the 2017-03-20 Orckit IP portfolio letter. Docket watch: https://docs.justia.com/cases/federal/district-courts/texas/txedce/2:2025cv01038/[241096](/patent/241096) . Trial-stage milestones will drive any stay motion.

If no PTAB activity is what you were hoping to find: it is. There are zero AIA trial proceedings on U.S. Patent No. 7,336,605 as of 2026-09-28. The absence is a signal — well-asserted, portfolio-held patents in active litigation often attract IPRs, and this one has not.

Sources

• USPTO PTAB E2E / PTAB proceedings search: https://ptacts.uspto.gov/ptacts/
• USPTO PTAB Decisions (FWDs and institution decisions): https://developer.uspto.gov/ptab-web/#/search/decisions
• Unified Patents litigation portal (case 2:25-cv-01038): https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A25-cv-01038
• EDS v. Cisco Memorandum Opinion and Order (2026-06-02): https://www.govinfo.gov/content/pkg/USCOURTS-txed-2_25-cv-01038/pdf/USCOURTS-txed-2_25-cv-01038-0.pdf
• Google Patents record for US 7,336,605 (legal status / expiry field): https://patents.google.com/patent/US7336605/en
• Sibling-patent petition record (U.S. Pat. No. 7,545,740): IPR2023-00401, IPR2024-00026, IPR2024-01237

Caveats I am obligated to state plainly: (a) I could not confirm the date Cisco was served with the complaint, so the exact § 315(b) bar date is unverified; (b) the expiration date 2025-12-01 comes from the patent's legal-status field, which the source itself flags as an assumption rather than a legal conclusion; (c) my searches were web-based and are not a substitute for a PTAB E2E / Docket Alarm / Lex Machina docket pull — if a petition was filed in the last few weeks, a keyword search may not surface it, though the structured ODP block (the canonical source specified for this task) records none.

I'll try to pull the correspondent-of-record and any additional assignment details before writing up.

Let me dig for the Enhanced Data Streaming / WYE IP records and any recurring correspondent.

Let me try to pin down the 2025 recording correspondents and EDS's address.

Assignment & Ownership Reconstruction — U.S. Patent No. 7,336,605

Method note / limitation. The assignment data below is taken from the reel/frame entries in the patent's own Google Patents legal-events record (source: https://patents.google.com/patent/[US7336605](/patent/US7336605)/en). The Google Patents legal-events block does not expose the "correspondent of record" field, and my live queries to the USPTO Assignment Center (https://assignmentcenter.uspto.gov/, mirrored at https://assignment.uspto.gov/patent/index.html) returned before I could pull the per-reel correspondent detail. I therefore report the correspondent field as unverified for this patent, and flag a partial lead from a sibling Orckit recording in the NPE-signals section rather than presenting it as fact. Reel/frame numbers are quoted exactly as recorded.

Date note (contradiction flag): the earlier "Patent summary" section used April 26, 2026 as "today" yet reported docket activity from June 2, 2026. The current task sets today as 2026-09-28. I treat 2026-09-28 as controlling and note the prior internal inconsistency rather than reconciling it silently.

Inventors

• The filing-date assignment (Reel 014072/0524) names the assignors in the order BRUCKMAN, LEON; MOR, GAL; MENAHEMI, ZVI, and is stated with effective date 2003-04-05 — i.e., executed roughly five weeks before the 2003-05-13 application filing. Espacenet re-orders the inventors as MOR Gal; BRUCKMAN Leon; MENAHEMI Zvi; the assignment order is the controlling record here.
• Unusual-pattern check — departures within 12 months: I could not verify any of the three inventors leaving the assignee within 12 months of filing. I have no evidence of that pattern, and I am not inferring it. What is documented is the opposite of a fire-sale tell at the inventor level: the inventors assigned on filing, and the corporate owner (Orckit/Corrigent) retained the portfolio for ~13 years.
• Weak corroborating datum: Orckit-focused patent aggregators list **Bruckman Leon (6) ** and **Mor Gal (4) ** among Orckit IP's top inventors (GoodIP, Orckit IP LLC profile), consistent with these being career Orckit/Corrigent engineers rather than transients.

Original assignee

Corrigent Systems, Ltd. (Israel) — named on the May 13, 2003 assignment (Reel 014072/0524) and on the face of the issued patent.

• Relationship to Orckit: Corrigent Systems, Ltd. was the Orckit-family entity (Orckit Communications Ltd., founded in Israel in 1990 by Izhak Tamir; Nasdaq-listed 1996; ADSL/metro-Ethernet infrastructure vendor). A change of name from Corrigent Systems Ltd. to Orckit-Corrigent Ltd. is recorded (Reel 032579/0201) and is effective 2009-09-06 — see the timeline for the recording lag.
• Did they ship a product embodying the claims? Yes — Orckit-Corrigent shipped carrier-grade packet-transport / metro-Ethernet platforms; the '605 patent's own specification (FIG. 1–2) describes central-office equipment with line cards, concentrators and traces, and Classifications H04L47/00 / H04L47/10 (traffic control). This is an operating-company artifact, not a paper patent. I have not independently tied the claims to a specific SKU, so treat "product embodying" as high-confidence but not SKU-verified.
• Current status: The operating parent failed. Per the Orckit Corporation v. Arista Networks complaint (D. Del. 1:23-cv-00821), "with the economic downturn of 2007 and 2008 … the company was unable to weather the global recession and ultimately went into liquidation." The patent right, however, did not travel through a receivership sale — it went by corporate assignment to Orckit IP, LLC (2016), then to the current owner. So: original assignee = operating company; now defunct/liquidated.

Assignment timeline

Every recorded event on US 10/436,516 / US 7,336,605, chronologically:

2003-05-13 (executed 2003-04-05) / recorded 2003-05-13 — Reel 014072/0524

• Conveyance: Assignment (Assignment of Assignors' Interest)
• Assignor: Bruckman, Leon; Mor, Gal; Menahemi, Zvi (inventors)
• Assignee: Corrigent Systems, Ltd., Israel
• Correspondent: not exposed in the legal-events record — unverified
• Context: Standard inventor-to-company assignment at original filing.

2013-03-18 (executed 2013-03-18) / recorded 2013-03-18 — Reel 030033/0774

• Conveyance: Security Agreement
• Assignor: Orckit-Corrigent Ltd.
• Assignee: Hudson Bay IP Opportunities Master Fund, LP, New York
• Correspondent: not exposed — unverified
• Context: Securitization — patent collateralized under a financing/security grant (a lien, not a title transfer).

2013-07-26 (executed 2013-07-23) / recorded 2013-07-26 — Reel 030887/0983

• Conveyance: Release by Secured Party
• Assignor: Hudson Bay IP Opportunities Master Fund LP
• Assignee: Orckit-Corrigent Ltd. (release back to grantor)
• Correspondent: not exposed — unverified
• Context: Release of the security interest recorded ~4 months after grant — the 2013 Hudson Bay facility was extinguished; no change in beneficial ownership.

2014-04-01 (executed / effective 2009-09-06) / recorded 2014-04-01 — Reel 032579/0201

• Conveyance: Change of Name
• Assignor: Corrigent Systems Ltd.
• Assignee: Orckit-Corrigent Ltd
• Correspondent: not exposed — unverified
• Context: Internal reorg / change of name only — no change in beneficial ownership; note the ≈4.5-year gap between the effective date (2009) and the recording date (2014), a documentation lag worth flagging in any chain-of-title clean-up.

2016-05-10 (executed 2016-04-25) / recorded 2016-05-10 — Reel 038529/0087

• Conveyance: Assignment (Assignment of Assignors' Interest)
• Assignors: Orckit Communication Ltd.; Orckit-Corrigent Ltd.; Corrigent Systems Ltd.
• Assignee: Orckit IP, LLC, Delaware
• Correspondent: not exposed for this reel; a sibling Orckit IP, LLC recording in the same reel series (Reel 038541/0984) lists correspondent "MAY PATENTS LTD." — see signal 3 below. Treat as a lead, not a verified entry for the '605 patent.
• Context: Transfer-to-asserter — the operating group parcels its patent estate into a dedicated licensing/holding LLC (Orckit IP, LLC).

2025-08-15 (executed / effective 2025-03-09) / recorded 2025-08-15 — Reel 072030/0801

• Conveyance: Assignment (Assignment of Assignors' Interest)
• Assignor: Orckit IP LLC
• Assignee: Enhanced Data Streaming LLC, Texas
• Correspondent: not exposed — unverified (this is the entry I most wanted; recommend pulling it directly from Assignment Center)
• Context: Transfer-to-asserter — outright sale of title to the entity that sued Cisco ~7 months later.

2025-09-29 (executed / effective 2025-08-13) / recorded 2025-09-29 — Reel 072978/0765

• Conveyance: Patent Security Agreement
• Assignor: Enhanced Data Streaming LLC
• Assignee: WYE IP CAPITAL LLC, New York
• Correspondent: not exposed — unverified
• Context: Securitization — the new NPE collateralizes the portfolio to a litigation/ monetization financier roughly 6 weeks before filing suit and days before the Oct. 10, 2025 complaint. A lien, not a title transfer.

Total recorded assignments: 7 (1 assignment-in, 2 financing/lien events + 1 release, 1 change-of-name, 1 assignment-out, 1 security grant). The Assignment Center does have records for this patent — this is not a "no-records" case.

Timeline diagram

timeline
    title Ownership of US 7336605
    2003 : Inventors assign to Corrigent Systems Ltd
         : Reel 014072/0524
    2008 : Patent issued Feb 26
    2009 : Name changed to Orckit Corrigent Ltd
    2013 : Security grant to Hudson Bay fund
         : Released four months later
    2014 : Name change recorded with USPTO
    2016 : Estate assigned to Orckit IP LLC
         : Reel 038529/0087
    2025 : Assigned to Enhanced Data Streaming LLC
         : WYE IP Capital security deal
         : Cisco suit filed in East Texas

(Per parseability rules: no parentheses, slashes, commas or symbols in event text; same-year events each prefixed with a colon.)

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT.
The chain crosses from an operating manufacturer (Corrigent Systems, Ltd. → Orckit-Corrigent Ltd.) into Orckit IP, LLC (Reel 038529/0087, executed 2016-04-25) and then to Enhanced Data Streaming LLC (Reel 072030/0801, effective 2025-03-09). Concrete non-name evidence for Orckit IP, LLC: it is repeatedly recorded at 874 Walker Road, Suite C, Dover, Delaware — a registered-agent service address (confirmed on a sibling Orckit IP, LLC filing, IPR2024-01237 Ex. 1002, and on the EPO/INPI record for EP1958364, which lists titular "ORCKIT IP, LLC – 874 WALKER ROAD SUITE C DOVER, DELAWARE"). That is a mass-registration address, not an operating site. EDS's tell is the licensing-only, no-products posture plus immediate assertion — it acquired title (eff. 2025-03-09) and sued the same year.

2. Known asserter in the chain — PRESENT (moderate-to-strong).
Neither "Orckit IP, LLC" nor "Enhanced Data Streaming LLC" appears on the classic named lists (Acacia, Marathon, IV, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation, Spangenberg entities). But the Unified Patents litigation portal classifies the plaintiff in 2:25-cv-01038 as "NPE (Patent Assertion Entity)", and the prior owner Orckit IP, LLC ran its own assertion campaign (e.g., the Feb. 1, 2018 Orckit IP, LLC suit against Extreme Networks, per Extreme's SEC filing) and sent the March 2017 portfolio-notice letter to Cisco that the E.D. Tex. court relied on for pre-suit knowledge. So the "repeat asserter family" signal is met via Unified Patents, not via the legacy name lists.

3. Repeat correspondent across the chain — UNCLEAR (partial lead only).
The correspondent field is not present in the legal-events record I could pull, so I cannot confirm recurrence. Lead worth verifying in Assignment Center: a sibling Orckit IP, LLC recording in the same reel series (Reel 038541/0984) lists correspondent "MAY PATENTS LTD.", and Orckit IP, LLC powers of attorney in the PTAB filings name Yehuda Binder (care of Dorit Shem-Tov, Ramat-Gan, Israel) — both are the kinds of recurring Israeli-prosecution contacts that often service an entire NPE family. Do not treat as a finding for the '605 patent until the Reel 038529/0087, 072030/0801 and 072978/0765 correspondent lines are pulled directly. Single-firm appearances are common to operating companies too; the signal only lands on recurrence.

4. Cascading transfers through chained LLCs in <24 months — NOT PRESENT (as to LLC chains).
Orckit IP, LLC → Enhanced Data Streaming LLC spans 2016→2025 (≈9 years). The only sub-24-month cluster is 2013-03-18 security → 2013-07-23 release → 2014-04-01 name-change recording, which is a financing + housekeeping cluster, not a chain of shell LLCs. So the chained-LLC tell does not fire.

5. Pre-litigation transfer — PRESENT.
Title moved to the asserting plaintiff via Reel 072030/0801 (effective 2025-03-09, recorded 2025-08-15), and the first complaint naming the '605 patent was filed 2025-10-10 (Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG, E.D. Tex.). Measured from the recording date that is ~8 weeks; measured from the execution date it is ~7 months (just outside a strict 6-month window). Either way the assignment timing and the suit are tightly coupled, and the WYE IP Capital security grant (eff. 2025-08-13) confirms the chain was arranged to fund/enable assertion.

6. Bankruptcy fire-sale — UNCLEAR.
The operating parent went into liquidation (Orckit Corp. v. Arista complaint, D. Del.), which is the economic condition that precedes fire-sales. But the '605 patent did not travel by a bankruptcy-court sale; it moved by corporate assignment (Reel 038529/0087) from the Orckit entities to a group-controlled LLC. I therefore cannot score this as a classic Chapter 7/11 sale, and I decline to over-read it.

7. Privateering — PRESENT (2016–2018 era), then clean NPE assertion.
The 2016 transfer (Reel 038529/0087) put the estate into Orckit IP, LLC, which then asserted against networking competitors (Orckit IP v. Extreme Networks, 2018) and noticed Cisco (March 2017 letter) — the classic operating-company-captive-NPE arc, made sharper by the operating company's liquidation. The current link (Reel 072030/0801) is no longer privateering on an operating company's behalf; it is a straight licensing-entity assertion.

8. Defensive aggregator (anti-NPE) — NOT PRESENT.
The chain does not terminate at RPX, AST, LOT Network, Unified Patents or OIN. It terminates at an asserting NPE with a litigation-finance lien. The patent is not neutralized.

Verdict

NPE — high confidence.

The title chain runs from an operating telecom-equipment maker (Corrigent Systems, Ltd., Reel 014072/0524) through a corporate reorganization (change-of-name, Reel 032579/0201), into a dedicated licensing LLC at a registered-agent address — Orckit IP, LLC, 874 Walker Road Suite C, Dover, Delaware (Reel 038529/0087, executed 2016-04-25) — and then by outright assignment to Enhanced Data Streaming LLC (Reel 072030/0801, effective 2025-03-09, recorded 2025-08-15), which collateralized the portfolio to WYE IP Capital LLC (Reel 072978/0765, effective 2025-08-13) and sued Cisco six weeks later (No. 2:25-cv-01038-JRG, filed 2025-10-10, with Unified Patents classifying the plaintiff as an NPE). That is three strong signals — shell-entity transfer, pre-litigation transfer, and post-transfer assertion by a Unified-Patents-listed NPE — on top of the prior Orckit IP assertion era; no operating-company or defensive-aggregator signal exists anywhere in the chain.

Verify against: USPTO Assignment Center search for patent 7,336,605 — https://assignmentcenter.uspto.gov/ (mirror: https://assignment.uspto.gov/patent/index.html). Pull the correspondent line for Reels 038529/0087, 072030/0801 and 072978/0765 specifically; that is the one field I could not confirm and the one most likely to tie EDS and Orckit IP to a shared filing agent.

I'll verify the cited prior-art references. Let me run targeted searches on the key references.

Let me gather details on several more cited references and confirm the USPTO record.

Prior-Art Analysis — U.S. Patent No. 7,336,605 B2

"Bandwidth allocation for link aggregation"

Scope and method (read first)

• I searched the literal identifier 7336605 / US7336605B2. No similar numbers were substituted. The bibliographic and claim text I rely on is the USPTO-format full text (mirrored at Google Patents and Espacenet). I was not able to complete a separate live USPTO PatentCenter/PAIR pull before the search tool limit was reached; the citation list below is taken from the authoritative full text you supplied, and I independently re-verified descriptions for the most relevant references (NTT, Lucent, Mosaid, Cisco, Nortel) via live search.
• Date caveat (carried forward from the earlier sections): you set today's date as April 26, 2026, but my live searches returned docket material dated June 2, 2026. I flag this rather than silently reconciling it.
• The '605 patent has expired (adjusted expiration Dec 1, 2025) and is in litigation (Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG). §102 is therefore most useful here as a validity lens (defenses/IPR), not an infringement lens.

Legal framework I applied

• All 26 cited references have priority/filing dates between 1995 and 2002, i.e., before the '605 patent's May 13, 2003 filing date. Every one is therefore available as prior art (pre‑AIA §102(a)/(b)/(e) for issued patents and §102(e) for published applications as of their filing dates).
• Anticipation standard: a reference anticipates a claim only if it discloses every element of that claim, arranged as in the claim. A dependent claim incorporates all limitations of its parent, so "anticipating" a dependent claim requires disclosing the parent's elements plus the dependent limitation.
• Important bottom line up front: Based on my review, none of the 26 cited references appears to disclose every element of independent claim 1 or independent claim 9. The granted independent claims are narrowly tied to (a) a logical link of parallel physical links, (b) a measured fluctuation caused by non-uniform load distribution, (c) per-link bandwidth allocation with a safety margin based on that fluctuation, such that the summed allocation exceeds the guaranteed bandwidth, and (d) distributing the data accordingly. The cited art maps onto individual elements, not the whole combination. These references are strongest as §103 (obviousness) combination art and as §102 art for discrete limitations. I say so explicitly rather than overstating anticipation.

The list below reproduces the examiner/applicant citation table exactly as printed (the * marks the references the record shows as cited by the examiner).

A. The 26 patent citations

Group 1 — Closest art (link aggregation / distributed bandwidth with headroom)

1. US 2005/0030948 A1 — "Link aggregation" — Mosaid Technologies, Inc. — priority 2000‑02‑29; published 2005‑02‑10 (issued as US 6,765,866 B1 on 2004‑07‑20; family also US 7,352,760 B2). Examiner‑cited (*).

• Description (verified): A switch builds a logical link / trunk group comprising multiple physical links. Flow‑hash logic indexes a flow hash on the data flow (source/destination addresses in the header); trunk‑port‑selector logic uses the hash to pick the physical link, and all packets of a flow stay on one link to preserve ordering.
• §102 relevance: This is the closest art on the "logical link of parallel physical links + hash‑based distribution" elements (claim 1 element 1 & 4; claim 5; claim 9; claim 13). It does not disclose a measured fluctuation, a per‑link bandwidth allocation with a safety margin, or a summed allocation exceeding the guaranteed bandwidth. No clean §102 anticipation of any granted claim (because independent claim 1 requires the fluctuation/safety‑margin elements). Its value is as §103 art and as §102 support for the distribution/hashing limitations.

2. US 6,560,231 B1 — "Multiplex transmission system and bandwidth control method" — NTT Mobile Communications Network, Inc. — filed 1997‑07‑23; granted 2003‑05‑06 (inventors via the JP priority of 1996). Examiner‑cited (*).

• Description (verified): A bandwidth managing controller assigns bandwidth to quality classes/channels; the record states the assigned bandwidths (X1–Xm) actually allotted "usually have a bandwidth equal to or greater than the required one … to fulfill the quality conditions," with dynamic reassignment based on monitored discard rates.
• §102 relevance: Relevant to the "allocate more than the strict requirement / deliberate headroom" concept underlying claim 1's "sum … substantially greater than the guaranteed bandwidth." But it allocates across service classes multiplexed onto a single transmission path — not across parallel physical links forming a logical link — and says nothing about a measured per‑link fluctuation or hash‑based frame distribution. No §102 anticipation of claim 1/9; meaningful §103 art on the safety‑margin/over‑allocation idea.

3. US 6,625,155 B1 — "Model‑based admission control adjustment in data networks" — Lucent Technologies Inc. — filed 1999‑09‑28; granted 2003‑09‑23 (inventor Z. Dziong). Examiner‑cited (*).

• Description (verified): Connection Admission Control is tuned using measured mean and variance/standard‑deviation of the aggregate cell process, yielding an allowed over‑booking gain and aggregate effective bandwidth.
• §102 relevance: This is the best cited art on the "measure of fluctuation" element — specifically standard deviation / variance (claim 7; claim 15) and the general proposition of setting margins from measured statistics. It does not disclose parallel physical links, per‑link link‑bandwidth allocation, or summing allocations above a connection's guaranteed bandwidth. No §102 anticipation of claim 1/9; strong §103 art for the "measure of fluctuation" limitation.

4. US 7,184,402 B1 — "Method for multi‑link load balancing to improve sequenced delivery of frames at peer end" — Cisco Technology, Inc. — filed 2001‑08‑30; granted 2007‑02‑27 (Sharma et al.; app. 09/944,782). Examiner‑cited (*).

• Description (verified): Establishes a distribution pattern to distribute multilink frame‑relay (MFR) fragments across the links of a link bundle, with link entries ordered by fastest transmit time; the pattern repeats.
• §102 relevance: Relevant to distributing data across parallel links of an aggregation/bundle (claim 1 element 4; claim 9 distributor). It is not a bandwidth‑allocation/safety‑margin reference and does not address fluctuation‑based guaranteed-bandwidth allocation. No §102 anticipation of claim 1/9; useful §103 art for the distribution element, and it is closer to the PPP/MLPPP environment than to 802.3ag.

5. US 6,826,147 B1 — "Method and apparatus for aggregate flow control in a differentiated services network" — Nortel Networks Limited — priority 2000‑07‑25; granted 2004‑11‑30.

• Description (verified): Manages aggregate flows between network edges using a credit system; SLAs define CIR and PIR per aggregate, and excess bandwidth is handled per the DS agreement.
• §102 relevance: Relevant background on guaranteed vs. excess bandwidth (CIR/PIR) — the concept of "guaranteed bandwidth … and excess bandwidth … available for other connections" mirrored in the '605 specification. Not directed to link aggregation and discloses no per‑physical‑link allocation. No §102 anticipation; §103/background art on the guaranteed‑vs‑excess concept.

6. US 6,929,394 B2 — "Communication apparatus with selective route optimization capabilities" — Fujitsu Limited — filed 2000‑08‑02; granted 2005‑07‑26.

• Description (inferred from title/assignee): Routing/route‑optimization in a communication apparatus. Discusses rearranging/optimizing routes.
• §102 relevance: At most tangential to "conveying data" and route/load decisions. No §102 anticipation of any claim.

Group 2 — Failure/recovery, trunking, spanning tree (background; relevant mainly to the un‑granted "protection policy" embodiment)

7. US 5,933,422 A — "Communication network recoverable from link failure using prioritized recovery classes" — NEC Corporation — filed 1996‑08‑20; granted 1999‑08‑03. Examiner‑cited (*).

• Relevant to link‑failure recovery / protection priorities. Potentially relevant, as background or §103, to the disclosed but not‑claimed "protection policy" concept (spec pp. describing P = max failed links). The granted claims are fluctuation‑based, not protection‑based, so this is not an anticipation of the granted claims.

8. US 6,370,121 B1 — "Method and system for shortcut trunking of LAN bridges" — Cisco Technology, Inc. — filed 1998‑06‑29; granted 2002‑04‑09.

• Relates to trunking/aggregating LAN links. Background on aggregated links; no bandwidth‑safety‑margin teaching. No §102 anticipation.

9. US 6,032,194 A — "Method and apparatus for rapidly reconfiguring computer networks" — Cisco Technology, Inc. — filed 1997‑12‑24; granted 2000‑02‑29.

• Spanning‑tree/rapid reconfiguration background. No §102 anticipation of the bandwidth‑allocation claims.

10. US 6,801,506 B1 — "Method and apparatus for providing fast spanning tree re‑starts" — Cisco Technology, Inc. — filed 1999‑03‑31; granted 2004‑10‑05.

• Spanning‑tree restart background. No §102 anticipation.

11. US 6,678,241 B1 — "Fast convergence with topology switching" — Cisco Technology, Inc. — filed 1999‑11‑30; granted 2004‑01‑13.

• Network convergence/topology‑switch background. No §102 anticipation.

12. EP 1 052 808 A1 — "Communications network using priority classes" — British Telecommunications plc — published 2000‑11‑15.

• Priority‑class QoS background. No §102 anticipation.

13. WO 2000/074318 A1 — "Transmitter‑based path protection switching in a ring network" — Marconi Communications, Inc. — filed 1999‑06‑02; published 2000‑12‑07.

• Ring protection switching. Background for "protection"; no §102 anticipation of the granted fluctuation claims.

Group 3 — Admission control, QoS policy, load estimation (element‑level relevance)

14. US 2002/0186661 A1 — "System and method for hierarchical policing of flows and subflows of a data stream" — Terago Communications, Inc. — filed 2001‑05‑04; published 2002‑12‑12.

• Hierarchical policing of aggregated flows; relevant background to bandwidth guarantees/policing. No §102 anticipation.

15. US 2003/0002443 A1 — "System and method for enhancing the availability of routing systems through equal cost multipath" — International Business Machines Corp. — filed 2001‑05‑24; published 2003‑01‑02.

• Equal‑cost multipath distribution across parallel paths for availability/load sharing. Relevant to distributing traffic across parallel links and to the availability rationale. Does not teach fluctuation‑measured per‑link bandwidth allocation. No §102 anticipation; §103 on the "distribute across parallel links" element.

16. US 2003/0055920 A1 — "Method and apparatus for automatic quality of service configuration based on traffic flow and other network parameters" — Deepak Kakadia — filed 2001‑09‑17; published 2003‑03‑20.

• Automatic QoS/bandwidth configuration from traffic parameters. Background; no §102 anticipation.

17. US 2003/0158930 A1 — "Executable application network impact and load characteristic estimation system" — McBride — filed 2002‑01‑22; published 2003‑08‑21. Examiner‑cited (*).

• Estimating network load characteristics. Tangentially relevant as §103 art on measuring/estimating load, but not to link aggregation or per‑link safety margins. No §102 anticipation.

18. US 2003/0223428 A1 — "Method and apparatus for scheduling aggregated resources" — Blanquer Gonzalez — filed 2002‑05‑28; published 2003‑12‑04. Examiner‑cited (*).

• Scheduling of aggregated resources. Tangential to "aggregating bandwidth/resources"; no parallel‑link logical link or fluctuation‑margin teaching. No §102 anticipation.

Group 4 — Network design, TDM/optical, routing, cost (minimal relevance)

19. US 6,754,286 B1 — "Self‑configuring communication network" — Alcatel — filed 1997‑03‑24; granted 2004‑06‑29. Background; no §102 anticipation.

20. US 6,934,259 B2 — "Apparatus and method for designing a network" — AT&T Corp. — filed 2000‑01‑18; granted 2005‑08‑23. Network‑design/planning; no §102 anticipation.

21. US 6,826,158 B2 — "Broadband tree‑configured ring for metropolitan area networks" — Onfiber Communications, Inc. — filed 2000‑03‑02; granted 2004‑11‑30. Metro ring architecture; no §102 anticipation.

22. US 2004/0071089 A1 — "Costs in data networks" — Daniel Bauer — filed 2000‑11‑21; published 2004‑04‑15. Cost‑based routing; no §102 anticipation.

23. US 6,763,025 B2 — "Time division multiplexing over broadband modulation method and apparatus" — Advent Networks, Inc. — filed 2001‑03‑12; granted 2004‑07‑13. Examiner‑cited (*). TDM over broadband; no §102 anticipation.

24. US 7,158,486 B2 — "Method and system for fast computation of routes under multiple network states with communication continuation" — Opcoast LLC — filed 2001‑03‑12; granted 2007‑01‑02. Examiner‑cited (*). Routing under failure states; no §102 anticipation.

25. US 6,624,917 B1 — "Optical power adjustment circuits for parallel optical transmitters" — International Business Machines Corp. — filed 1999‑10‑28; granted 2003‑09‑23. Examiner‑cited (*). Optical/physical‑layer hardware; unrelated to the claims. No §102 anticipation.

26. US 6,169,783 B1 — "Line testing in a telecommunications network" — British Telecommunications plc — filed 1995‑09‑12; granted 2001‑01‑02. Examiner‑cited (*). Line testing; unrelated. No §102 anticipation.

B. Summary matrix

C. Non‑patent literature cited (relevant context)

The record cites 15 non‑patent items; the two most probative for validity are worth noting even though your question focused on patent citations:

• IEEE Std 802.3, Clause 43 (2002) / Annex 43A — defines link aggregation, the distributor/collector, and the conversation‑to‑port hashing requirement. This is the foundational art for the "logical link of parallel physical links" and hash‑distribution elements (claims 1, 2, 5, 9, 10, 13).
• Dziong, "A Framework for Bandwidth Management in ATM Networks — Aggregate Equivalent Bandwidth Estimation Approach," IEEE/ACM ToN, Feb. 1997 — the academic counterpart to US 6,625,155, supporting the "measure of fluctuation → bandwidth margin" concept.
• Also cited: RFC 1990 (PPP Multilink), IETF PWE3/MPLS drafts, RFC 2892 (IMA), and VPLS/L2TPv3 drafts — relevant to the ATM/IMA and PPP claim families (claims 3, 4, 11, 12).

D. Analyst's bottom line

1. No single one of the 26 cited patent references anticipates independent claim 1 or claim 9, because each lacks at least one of: (a) parallel physical links forming a logical link, (b) a measured fluctuation due to non‑uniform load distribution, or (c) per‑link bandwidth allocation with a safety margin whose sum exceeds the connection's guaranteed bandwidth.
2. The most probative cited references are, in order: US 2005/0030948 A1 (Mosaid) for the logical-link/hash-distribution elements; US 6,560,231 (NTT Mobile) for deliberate over‑allocation above the requirement; US 6,625,155 (Lucent) for the measured mean/variance fluctuation element; and US 7,184,402 (Cisco) for distributing frames across a bundle of parallel links. A §103 case — if one is asserted — would most naturally combine these with the 802.3ad link‑aggregation frame (non‑patent).
3. Caveats/limits on this analysis: (i) This is a preliminary analyst assessment, not a legal opinion on validity or infringement. (ii) I was unable to complete independent live USPTO PatentCenter/PAIR verification before hitting the tool limit; descriptions for the non‑searched references are drawn from the authoritative citation table plus the reference titles. (iii) The date discrepancy (April 26, 2026 vs. June 2026 docket material) persists and should be resolved for your timeline. (iv) All identifiers are reproduced literally (through US 7,336,605 B2 and publication US 2004/0228278 A1).

Sources: https://patents.google.com/patent/US7336605/en (full text/citation table) · https://patents.google.com/patent/US6560231 · https://patents.google.com/patent/US6625155 · https://patents.google.com/patent/US20050030948 · https://patents.google.com/patent/US7184402 · https://patents.google.com/patent/US6826147B1

I'll analyze obviousness using the prior-art references listed on this patent's page. Let me pull details on the most probative references first.

Let me examine the remaining cited prior art references relevant to the claimed elements.

Obviousness Analysis — U.S. Patent No. 7,336,605 under 35 U.S.C. § 103

0. Framing and scope (building on prior sections)

I build on, and do not repeat, the earlier Patent summary and Litigation summary. Three cross-references matter here:

• The granted claims (18; independents 1 and 9) are directed to the fluctuation-measure embodiment — not the "protection policy" embodiment that dominated the published application (US 2004/0228278 A1, 26 claims, independents 1/11/19). The obviousness grounds below are therefore aimed at the fluctuation limitations that were added during prosecution (the "wherein the fluctuation is due to a non-uniform distribution of load" clause and the safety margin "based on the measure of fluctuation").
• The '605 is pre-AIA (filed May 13, 2003; priority May 13, 2003), so § 103 is governed by pre-AIA law and pre-AIA § 102(e) art is available as 103 art.
• The '605 is expired (adjusted expiration Dec 1, 2025 — the page labels it "Expired – Lifetime"). This does not bar an IPR or an invalidity defense in Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG, where the '605 is one of five asserted patents. These grounds are the kind of § 103 theory that would be run as an IPR or at trial.

Date caveat (carried forward): my earlier searches returned docket items dated June 2, 2026 while the task states today is April 26, 2026. That inconsistency is unresolved and does not affect the prior-art analysis below, which only requires art predating May 13, 2003.

1. Prior-art inventory from the "Prior Art" section, mapped to claim 1 limitations

Claim 1 (method) requires: (a) a logical link of parallel physical links between endpoints; (b) determining a measure of fluctuation in transmission rate over those links when data are distributed among them, the fluctuation being due to non-uniform load distribution; (c) allocating per-link bandwidth including a predefined safety margin based on that fluctuation measure, such that the sum of allocated link bandwidths exceeds the connection's guaranteed bandwidth; and (d) conveying the data by distributing it among the links "in accordance with the allocated link bandwidth." Claim 9 is the apparatus mirror (controller + distributor + data transmission circuitry).

Note on § 102(e) dates. Three cited references published/issued after May 13, 2003 but have earlier U.S. filing dates and therefore qualify as pre-AIA § 102(e) art usable in § 103:

• US 2005/0030948 A1 / US 6,765,866 — parent application filed Feb. 29, 2000 (Wyatt).
• US 2003/0223428 A1 — filed May 28, 2002 (Blanquer Gonzalez).
• US 7,184,402 B1 — filed Aug. 30, 2001 (Sharma).

2. Ground 1 (primary): 802.3 Clause 43 (+ Annex 43A) in view of Mosaid (US 6,765,866) and Dziong

Why each reference qualifies. Clause 43 is cited on the face of the '605 and is incorporated by reference into the '605 specification itself. Mosaid is a § 102(e) reference (Wyatt, filed 2000) and is listed in the '605's own "Citations" table. Dziong is a printed publication listed in the '605's "Non-Patent Citations" table (a 1997 IEEE/ACM journal article).

Claim 1 mapping.

• (a) Clause 43 defines a link aggregation group of parallel physical links joined into one logical link with a distributor and collector — verbatim the claim's "logical link comprising a plurality of parallel physical links between the endpoints." Mosaid supplies the corresponding switch structure (logical link/trunk group 134 of physical links 132c–e; logical link 140; egress port queues).
• (b) Clause 43/Annex 43A acknowledges that hash-based distribution cannot be made uniformly balanced across arbitrary traffic ("it is difficult to select a hash function that will generate a uniform distribution of load across the set of ports for all traffic models" — quoted in the '605 Background, which is an admission that this was known). Mosaid states the goal of even distribution and explains why it is hard: "To achieve maximum bandwidth utilization on the logical link, data packets to the destination must be evenly distributed amongst the physical links"; and flows cannot be mis-ordered, so they are pinned to one link. Blanquer supplies the quantitative treatment of exactly this phenomenon for link aggregation: the multi-server system's "rate … may vary over time and thus differ from the rate of the reference system," and partitioning flows "can easily occur when the sum of the flow weights is not balanced across all partitions" (bandwidth fragmentation). A POSITA reading these three items has an explicit teaching that per-link transmission rate fluctuates due to non-uniform loading of the parallel links — the claim's "measure of fluctuation … due to a non-uniform distribution of load."
• (c) This is the element the primary pair alone does not fully supply, and it is precisely where Dziong fits. Dziong teaches an adaptive connection admission framework built on "estimation of the aggregate equivalent bandwidth required by connections carried in each output port," where "the estimation process takes into account both the traffic source declarations and the connection superposition process measurements," and, critically, "To provide a required quality of service guarantee, bandwidth is reserved for possible estimation error." That reservation-for-estimation-error is a predefined safety margin driven by a measured fluctuation of aggregate traffic; applied per output port it yields an allocation above the nominal aggregate demand — i.e., a sum of per-link allocations greater than the guaranteed bandwidth.
• (d) Mosaid and Clause 43/Annex 43A provide the distribution step (flow hash → trunk port entry → physical link), and Blanquer provides the corresponding multi-server scheduling discipline that allocates bandwidth per flow across N servers under QoS guarantees. Distributing "in accordance with the allocated link bandwidth" is met by, e.g., Mosaid's per-link distribution plus Blanquer/Sharma's rate-proportional allocation.

Motivation to combine. (i) Same field and same problem — Mosaid and Blanquer are expressly about Ethernet link aggregation bandwidth scaling, and 802.3 Clause 43 is the very standard the '605 invokes; Dziong addresses "bandwidth management" and QoS in a multiplexed-link context. (ii) Blanquer expressly identifies the gap Dziong fills: existing aggregation implementations "already provide different techniques for load balancing the traffic among the interfaces but none of them address the provision of QoS over these aggregated links," and "provisioning of such systems is naturally described as a function of the total link capacity rather than for each of the links." A POSITA seeking to provide a guaranteed rate over an aggregated link is thus directly led to a bandwidth-management/admission-control teaching such as Dziong. (iii) Dziong's own motivation ("to provide a required QoS guarantee … bandwidth is reserved for possible estimation error") is the motivation to add the safety margin — the classic KSR "known technique to a known device ready for improvement." (iv) Predictable result: allocating a statistical margin above the mean is the routine engineering response to the non-uniform-load problem both 802.3 and Blanquer identify.

3. Ground 2 (alternative primary): Blanquer (US 2003/0223428) in view of Dziong, optionally with 802.3 Clause 43

Blanquer alone teaches (a), (b) and (d) at a high level for link aggregation specifically (the published application even labels aggregated multi-link systems as a "multi-server" abstraction whose reference service is a single server "operating at a rate equal to the sum of the underlying servers' rates"). Its admitted weakness — "none of them address the provision of QoS over these aggregated links" — is the express motivation to bring in Dziong's reserved-bandwidth-for-estimation-error teaching to reach element (c). This ground is cleaner on the "link aggregation" framing than Ground 1 because Blanquer names link aggregation by name.

4. Ground 3: Cisco Sharma (US 7,184,402) in view of NTT (US 6,560,231), with 802.3

This pairing is attractive because it supplies element (c)'s over-allocation with a very short logical step.

• Sharma (US 7,184,402 B1) — verified — establishes (a) and (d): a "link bundle" of parallel links; a distribution pattern in which "each link entry is placed … when the associated link is capable of transmitting a current fragment in a fastest transmit time," and "a faster link in the link bundle may appear in the distribution pattern more than once." That is literally per-link allocation of distribution weight (bandwidth) across parallel links and distribution "in accordance with the allocated link bandwidth."
• NTT (US 6,560,231 B1) — verified — supplies (c) almost verbatim: a "bandwidth managing controller" monitors discard rates and determines the required bandwidths per class, and the specification states that "the assigned bandwidths (X1–Xm) actually allotted to the transmission path usually have a bandwidth equal to or greater than the required one … to fulfill the quality conditions," converting those assigned bandwidth ratios into extraction information that the extracting block enforces. That is a per-resource allocation that deliberately exceeds the required/guaranteed amount to satisfy QoS — the "sum … substantially greater than the guaranteed bandwidth" element — and it is measurement-driven (monitored discard rates), which reads on the "measure of fluctuation" element under the broadest reasonable construction.
• 802.3 Clause 43 supplies the aggregation framing and the known hash-distribution non-uniformity to connect the NTT margin to the load-skew cause recited in the claim.

Motivation: NTT's dynamic bandwidth assignment and Sharma's per-link distribution are complementary halves of one system (how much per link vs. which link gets the next fragment); combining them is the application of a known technique (dynamic, measurement-driven over-allocation) to a known device (a multi-link bundle) to yield a predictable result.

5. Dependent claims — where each ground lands

6. Motivation-to-combine synthesis (KSR / MPEP § 2143)

For every ground above, the teaching, suggestion, or motivation is documentary rather than hindsight:

1. Same field of endeavor and same problem. The '605, Clause 43, Mosaid, Sharma and Blanquer all address aggregating parallel links into one logical link and distributing frames across them. The '605's own Background admits the known problem ("it is difficult to select a hash function that will generate a uniform distribution of load across the set of ports for all traffic models").
2. Express gap-filling suggestion. Blanquer states that existing aggregation load-balancing techniques "none of them address the provision of QoS over these aggregated links." That is a textbook express suggestion to combine an aggregation/load-balancing reference with a QoS/bandwidth-management reference (Dziong, NTT, Lucent).
3. Known technique applied to a known device ready for improvement. Reserving statistical headroom above a mean demand is a decades-old queuing/teletraffic technique (Dziong's "bandwidth is reserved for possible estimation error"; NTT's "equal to or greater than the required one"). Applying it to the newly popular link-aggregation logical link is the KSR situation.
4. Predictable results / no unexpected mechanism. The claimed benefit (guaranteeing B despite hash-induced skew by over-allocating F·B/N per link) is the arithmetic consequence of the combined teachings.
5. Design incentive / market pressure. All references sit in the 2000–2003 carrier-Ethernet push (802.3 Clause 43 finalized 2002; Sharma's MFR fragmentation; Blanquer's SIGCOMM'01 work; Dziong's ATM CAC) — a POSITA had strong incentive to make aggregated Ethernet links carry SLA-guaranteed services, which requires exactly the allocation discipline claimed.

7. Strengths, weaknesses, and where a § 103 case is vulnerable

Strongest elements. Element (c) — a measurement-driven safety margin producing over-allocation — is well covered by Dziong (verified) and, independently, by NTT US 6,560,231 (verified). Element (b)'s "non-uniform load distribution" cause is essentially admitted by 802.3/Annex 43A and by Mosaid/Blanquer.

Weakest points a patent owner would press:

• "Substantially greater than the guaranteed bandwidth." Dziong's reservation is for estimation error of aggregate equivalent bandwidth at an ATM switch output port, not a per-link margin in a link-aggregation group. A patent owner will argue the sum "substantially greater" is a deliberate LAG allocation choice, not a byproduct. This is answerable but needs careful claim-charting of Dziong's equations.
• Claim 8 (max-vs-average utilization). This is the most specific measurement limitation, and the cited art set does not cleanly teach that metric. Expect this to be the surviving claim in an IPR. If a true-utilization-difference reference is needed, it may lie outside the face-of-page set — I did not identify one in this session.
• Claims 17–18 (line-card/trace architecture, links on different line cards). The cited art is thin. These claims are the '605's hardware-specific residue and the likeliest to survive; invalidity on 17–18 with only the page-listed art is doubtful.
• Analogy/§ 102(e) attacks. Mosaid, Blanquer and Sharma are § 102(e)/published-after art. Expect argument over (i) whether each is "by another," (ii) the § 102(e) date, and (iii) whether a 1997 ATM CAC paper (Dziong) is analogous art to a 2003 Ethernet LAG claim. On (iii), the patent's own specification cites Dziong-adjacent ATM/IMA art and invokes IMA/ATM as alternatives, which undercuts a non-analogous-art defense.
• Secondary considerations. I found no evidence in the available record of unexpected results, licensing-due-to-merits, or industry praise tied to the '605's fluctuation-margin feature. Absent such evidence, the Graham factor 4 does not cut against obviousness.
• Claim-scope trap. Any analysis (or IPR petition) that charts the published application's "protection policy" claims against these references will mis-target the granted claims, which recite the fluctuation measure and omit "protection policy" from the independent claims. This distinction, flagged in my earlier Patent summary, is outcome-determinative.

8. Confidence and limitations

• High confidence (verified from primary text this session): Dziong abstract and reservation-for-estimation-error teaching; NTT US 6,560,231 over-allocation quote; Sharma US 7,184,402 distribution-pattern teaching; Mosaid US 6,765,866 / US 2005/0030948 hash-based link selection; Blanquer's aggregated-links/QoS-gap statements; the facial § 102(e) dates of Mosaid, Blanquer and Sharma.
• Lower confidence (title/bibliographic only): US 6,625,155 (Lucent), US 6,826,147 (Nortel), US 2002/0186661 (Terago), US 6,934,259 (AT&T), US 2003/0002443 (IBM), Cisco ONS 15454, and the NEC/Marconi/BT failure-protection references. My verifying searches for these four were cut off; I am characterizing them by title and general field knowledge and would re-verify before relying on them in a written opinion or petition.
• I am reporting all identifiers literally as they appear on the '605 page (including US 2005/0030948 A1, US 6,765,866, US 7,184,402 B1, US 2003/0223428 A1, US 6,560,231 B1) with no auto-correction.
• Nothing here is a legal conclusion of invalidity; it is a technical § 103 mapping for the EDS v. Cisco context.