US 6891855

I'll search for the specific patent number and check litigation/docket sources.

Let me check for litigation involving the patent's current owner and any 2026 Federal Circuit activity.

US Patent 6,891,855 B2 — Analyst Summary

Sourcing note: The authoritative full text of the patent (Google Patents, https://patents.google.com/patent/[US6891855](/patent/US6891855)/en) was supplied and is preferred over search results. Search results below are used only for litigation/assignment status.

Bibliographic Data

Abstract (verbatim)

"A method and apparatus for transmitting data over a channel having a variable transmission rate. A transmitter determines the rate of transmission of the data over the channel and receives a datagram for transmission over the channel at the determined rate of transmission. The transmitter divides the datagram into fragments of a size no greater than a size limit that is set for the datagram responsive to the determined rate of transmission and transmits the fragments over the channel."

Technology Overview

The invention addresses the classic DSL delay/efficiency tension: to keep high-priority (e.g., voice) packet latency low, fragments must be short, but short fragments impose proportionally high per-fragment header overhead. The specification's exemplar formula is Temp_Frag_Length = Int{(Max_Wait × Rate)/8} − Overhead (col. 5), with Max_Wait being operator-programmed and Rate the instantaneous channel rate. Fragment size is therefore bounded in time (max transmit time per fragment) rather than fixed in bytes, and the actual fragment length is bounded below by the operator's MinL_Fr. High-priority packets can preempt remaining low-priority fragments. Table I shows dynamic fragmentation degrading gracefully to the static case at 128 kbps while cutting fractional overhead from 1/8 to 1/144 at 2300 kbps.

Independent Claims — Plain-Language Overview

Claim 1 (method). Determine the current transmission rate of the channel; receive a datagram to be sent at that rate; split it into fragments no larger than a size limit chosen based on the determined rate; transmit them. Two added limitations: (a) the size limit is set so that the time needed to transmit each fragment does not exceed a predetermined maximum time, and (b) the datagram is a low-priority datagram, and that predetermined maximum time is the maximum delay allowed for high-priority datagrams sharing the same channel. This is the narrowest claim because it requires the priority-differentiated delay budget.

Claim 8 (method). Same base steps, but the added limitations are: the size limit increases as the transmission rate increases, and the fragments have fixed overhead with variable payload, such that the payload-to-overhead ratio grows as the rate grows.

Claim 9 (method). Same base steps, with size limit increasing as rate increases, and the size limit is set according to fragmentation control parameters programmed by a human operator; the limit is then increased automatically as rate increases substantially without altering those parameters — i.e., no operator reprogramming is needed.

Claim 10 (apparatus). A transmitter that sets the channel rate, sets a fragment size limit responsive to that rate, and upon receiving a datagram divides it into fragments no larger than the limit and transmits them. Parallel to claim 1: the size limit ensures per-fragment transmit time does not exceed a predetermined maximum time, the datagram is low-priority, and the maximum time is the maximum delay applicable to high-priority datagrams on the same channel.

Claim 17 (apparatus). Transmitter counterpart to claim 8: size limit increases with rate; fragments have fixed overhead and variable payload with increasing payload-to-overhead ratio.

Claim 18 (apparatus). Transmitter counterpart to claim 9: size limit increases with rate and is set responsive to operator-programmed fragmentation control parameters, increasing automatically without operator alteration.

Dependent claims. Claims 2/11 recite the ~100 kbps–2300 kbps range; 3/12 the DSL access channel; 4/13 "data packet"; 5/14 preemption/interruption of low-priority fragments to serve a high-priority datagram within the maximum delay; 6/15 increase of size limit with rate; 7/16 receiver-side reassembly responsive to the variable rate (specification notes the reassembler need not know the rate, but benefit accrues in buffer sizing).

Litigation / Docket Findings

• E.D. Tex. 2:25-cv-01038 — Enhanced Data Streaming LLC v. Cisco Systems, Inc. Filed 2025-10-10 (Judge Rodney Gilstrap; 35 U.S.C. § 271). The '855 patent is one of the patents-in-suit (styled "the '855 Patent"). A Memorandum Opinion and Order (Doc. 48) addresses Cisco's Fed. R. Civ. P. 12(b)(6) motion to dismiss the indirect- and willful-infringement claims of the First Amended Complaint. The complaint's analysis of claim 1 maps the four base steps to Cisco ASR 1000 Multilink PPP over Ethernet / Link Fragmentation and Interleaving functionality, and maps the "predetermined maximum time" limitation to the ppp multilink fragment delay command. A March 20, 2017 notice letter is cited as pre-suit knowledge for willfulness. Sources: https://docs.justia.com/cases/federal/district-courts/texas/txedce/2:2025cv01038/[241096](/patent/241096)/48/ and https://www.courtlistener.com/opinion/[10893882](/patent/10893882)/enhanced-data-streaming-llc-v-cisco-systems-inc/
• CAFC 2026 dockets: I found no Federal Circuit appeal, mandamus, or petition docket specifically identifying US 6,891,855 as of this search. This is an absence of evidence, not proof of absence — CAFC docketing entries frequently reference only the lower-court case number rather than the asserted patent, so a 2026 appeal tied to 2:25-cv-01038 could exist without surfacing in a patent-number-keyed search. Treat this point as uncertain.

Uncertainty / Caveats

1. I could not confirm the current assignee/security-interest chain from a primary USPTO assignment record in this session; it is taken from the Google Patents legal-events listing (Enhanced Data Streaming LLC, 2025-08-15 recordation; Wye IP Capital LLC security agreement, 2025-09-29). Verify against USPTO Patent Assignment Search if the chain of title matters.
2. The patent's stated legal status is "Expired – Lifetime" with an adjusted expiration of 2023-04-21; maintenance-fee and reinstatement events are recorded (including a 2013 reinstatement after late payment). Any assertion of the patent post-2023 would require reconciling the expiration date, which is beyond what the pages state.
3. Primary examiner and art unit were not present in the retrieved record and are not supplied here rather than guessed.
4. My searches were run against general web sources indexed for the patent number; I did not directly query a USPTO-internal API or a PACER/CAFC docket system, so litigation details reflect secondary aggregators (Justia, CourtListener, Dockets.Justia) rather than certified docket copies.

I'll research litigation involving US patent 6891855 across patent litigation databases.

Let me dig deeper into this case and check for additional defendants/cases.

Litigation Involving U.S. Patent No. 6,891,855

Summary

Based on my searches of patent-litigation sources (Unified Patents litigation portal, CourtListener/RECAP, Justia federal dockets, Dockets.Justia, ExParte, and the Google Patents litigation links), one (1) active federal lawsuit asserting U.S. Patent No. 6,891,855 ("the '855 Patent") is identified. I found no other suit — past or present — in which the '855 Patent was asserted.

Patent(s) at issue in that case

The '855 Patent is one of five patents asserted in the complaint:

• U.S. Patent No. 9,185,151 (reissued April 22, 2025 as RE50,398)
• U.S. Patent No. 6,891,855 (the '855 Patent — the patent you asked about)
• U.S. Patent No. 7,336,605
• U.S. Patent No. 7,660,234
• U.S. Patent No. 7,032,135

Accused products and infringement theory (as pleaded)

According to the complaint-analysis record, EDS accuses the Cisco ASR 1000 Series router (and related switch/router features), specifically its Multilink PPP over Ethernet (MLPoE) and Link Fragmentation and Interleaving (LFI) functionality. The complaint maps claim 1 of the '855 Patent onto this functionality:

• "Determining the rate of transmission" ← Cisco's media-rate information received from a DSLAM via PPPoE vendor tags;
• "Dividing the datagram into fragments … responsive to the determined rate" ← multilink fragmentation of large packets using link-bandwidth parameters;
• "Setting the size limit such that a length of time required to transmit each fragment is no greater than a predetermined maximum time" ← LFI's fragment-delay budget (e.g., the ppp multilink fragment delay command);
• Low-priority/high-priority interleaving limitation ← LFI interleaving of voice/video between fragments of lower-priority data.

EDS also pleaded induced and willful infringement. A likely defense theme (per the case record) is that the accused functionality is simply the pre-existing Multilink PPP (MLP) / LFI industry standards, raising novelty/obviousness questions.

Outcome / procedural history

• Dec. 19, 2025 — Cisco filed a motion to dismiss EDS's indirect- and willful-infringement claims from the initial complaint.
• Jan. 16, 2026 — EDS filed a First Amended Complaint (FAC), re-asserting the same five patents.
• Jan. 30, 2026 — Cisco renewed its motion to dismiss (targeting the FAC's indirect/willful claims).
• June 2, 2026 — Judge Gilstrap issued a Memorandum Opinion and Order DENYING Cisco's motion to dismiss (Dkt. No. 26 terminated). (Note: this June 2026 order post-dates your stated "current date" of April 26, 2026; search results returned it, so I report it as the ground truth returned by the searches.)

No final judgment, settlement, dismissal, or appeal has been identified for this case. It remains an open, active infringement action.

Important caveats and identifiers

• Do not confuse with U.S. Patent No. 6,897,185. Several searches surfaced Lytone Enterprise, Inc. v. AgroFresh Solutions, Inc., IPR2021-00451 (Fed. Cir. 22-2269) — that case concerns 6,897,185 ("Formulation for Counteracting an Ethylene Response in Plants"), a completely different patent. Per your instruction to interpret identifiers literally, I exclude it; it is not litigation involving 6,891,855.
• No litigation was found in which an Orckit entity asserted the '855 Patent. Orckit-Corrigent/Orckit IP is the original patent family owner (the '855 Patent issued to Corrigent Systems Ltd., later Orckit-Corrigent), and Orckit entities have litigated many other patents against Cisco, Nokia, Arista, Extreme, etc. — e.g., Orckit Corp. v. Cisco Sys., No. 2:25-cv-00181 (E.D. Tex.); Orckit Corp. v. Arista Networks, No. 1:23-cv-00821 (D. Del.). None of those identified actions asserts the '855 Patent.
• The Google Patents "family has litigation" links for the '855 Patent point to the Texas Eastern District Court case — i.e., the EDS v. Cisco action above.

Legal-status note (relevant to enforceability)

Per the Google Patents record, the '855 Patent is "Expired – Lifetime," adjusted expiration 2023-04-21, and its ownership chain ran Orckit Communications → Corrigent Systems → Orckit-Corrigent → Orckit IP, LLC → Enhanced Data Streaming LLC (assigned Aug. 15, 2025, effective 2025-03-09) → with a patent security agreement to Wye IP Capital LLC (Sept. 29, 2025). The reissue application number 09/756,554 and priority date 2000-07-27 are unchanged. The fact that the patent's term expired in April 2023 materially limits the potential damages window in the EDS v. Cisco action, though I did not find a court ruling on that point.

Bottom line

Only one lawsuit involving U.S. Patent No. 6,891,855 is known: Enhanced Data Streaming LLC v. Cisco Systems, Inc., No. 2:25-cv-01038-JRG (E.D. Tex., filed Oct. 10, 2025), currently pending, with the district court having denied Cisco's motion to dismiss on June 2, 2026. If you need a more exhaustive negative clearance (e.g., pre-2016 litigation, settlements never docketed, or ITC Section 337 investigations not captured by these databases), a direct PACER/Docket Alarm query on the patent number would be the appropriate next step — I could not fully verify the absence of such older actions from the search results returned to me.

I'll verify the "no PTAB activity" default against public sources before drafting.

The first pass surfaced IPRs on other Orckit patents (notably the '111 patent). Let me confirm specifically that no AIA trial names the '855 patent.

Let me run two final confirmation searches — one for any 2026 Cisco petition, one for any reexamination/PTAB record keyed to this patent.

PTAB / AIA Trial Analysis — U.S. Patent No. 6,891,855 ("Dynamic packet fragmentation")

Sourcing note: Per my operating instructions, the canonical proceedings list is the structured "PTAB proceedings on file" block, which reports that the USPTO Open Data Portal API returns no AIA trial proceedings for this patent as of the most recent ingest. I used web search only to test that default and to look for recently-filed or non-indexed proceedings. Searches were terminated at the tool-step limit; the residual caveat is noted below.

Proceedings overview

There are zero AIA trial proceedings on U.S. Patent No. 6,891,855 — no IPR, no PGR, and no CBM, whether active, terminated, settled, instituted-but-unresolved, or denied institution. The defensive posture for a defendant is therefore not "the patent is hardened" and not "claims 1–5 are canceled," but a third and distinct thing: the patent is completely untested at the Board, and no petitioner is estopped from anything. All eighteen claims (1–18) stand exactly as issued, and every invalidity ground — including grounds a prior petitioner would have been barred from re-raising under 35 U.S.C. § 315(e)(2) — remains fully available to any defendant. That is a clean slate, not a hardened patent.

Proceeding inventory

No proceedings to enumerate

Because the canonical list is empty and my searches corroborated it, there is no {PROCEEDING_NUMBER} to populate the per-proceeding template with. I will not manufacture one. What follows is the null result, its evidentiary basis, and the closely-related activity I did find — clearly labeled as not involving the '855 patent.

Basis for the null result (three independent checks):

1. Canonical source. The USPTO ODP-derived "PTAB proceedings on file" block states no AIA trial proceedings exist for this patent. That block is authoritative for this exercise.
2. Patent-number-keyed web search. Searches pairing the '855 number with "IPR," "PTAB," "post-grant review," and "reexamination" returned (a) the patent's own Google Patents page, (b) the Enhanced Data Streaming LLC v. Cisco Systems, Inc. docket materials, and (c) unrelated patents that happen to be numbered '855 — specifically U.S. Patent No. 4,927,855 (Cephalon/armodafinil prior art) and references to a separate '855 patent in the Cephalon v. Mylan litigation. No PTAB trial record keyed to 6,891,855 appeared.
3. Litigation-context search. Cisco is a prolific IPR filer, including against this same patent family. A search for Cisco IPR petitions arising from Enhanced Data Streaming (the current plaintiff) and for 2025–2026 Cisco petitions surfaced petitions against Dynamic Mesh Networks, Lionra, IOENGINE, and ChriMar — but none against Enhanced Data Streaming and none against the '855 patent.

Related Orckit-family IPRs — expressly NOT proceedings on the '855 patent. For defensive-context purposes only, note that the '855 patent's former owners (Orckit / Corrigent) had other patents taken to the Board by Cisco, Juniper, and Arista, and that the Board has been receptive to such petitions:

None of these is an AIA trial on 6,891,855. I list them solely so the reader does not mistake portfolio-level activity for patent-level activity — a real risk here, because the assumed prior art in those cases ("Bruckman," U.S. Pub. 2004/0228278) is co-inventor-adjacent to the '855 patent's sole inventor, Leon Bruckman, and shares the Orckit corporate lineage. Different patent, different claims, no estoppel effect on the '855 patent.

Caveat on the null result. The absence is well-supported but not proven to an absolute. A petition filed after the ODP ingest and not yet surfaced by the aggregators I could reach would not appear above. The window for that is narrow — see the § 315(b) timing point below — but a direct PTAB E2E / Patent Center query on 6,891,855 remains the appropriate verification step before relying on "no IPRs" in a brief.

Strategic summary

Claim status. All 18 claims of the '855 patent are UNTESTED at the Board. Nothing is canceled, nothing is confirmed, and nothing has been narrowed by an IPR. The claim set is:

• Independent method claims: 1, 8, 9
• Dependent method claims: 2, 3, 4, 5, 6, 7 (1–7 depending from claim 1)
• Independent apparatus claims: 10, 17, 18
• Dependent apparatus claims: 11, 12, 13, 14, 15, 16

There is no FWD to quote, so I quote none. Any statement that a particular claim "survived" or "was canceled" in an AIA trial on this patent would be fabrication.

Estoppel landscape. § 315(e)(2) estops a petitioner, a real party in interest, or a privy from asserting in a civil action any ground raised in an IPR that resulted in an FWD, or that the petitioner reasonably could have raised. Because no IPR has ever been instituted or decided on the '855 patent, no estoppel attaches to anyone. Practical consequences for a defendant today:

• The full universe of § 102/§ 103 art remains available — including printed publications, patents, and system/device prior art, which is categorically outside IPR's reach under § 311(b) (see the well-known IOENGINE v. PayPal and S.D. Cal. formulations that "IPR estoppel does not apply to device art").
• Prior art already before the examiner during prosecution — notably FRF.12, Frame Relay Fragmentation Implementation Agreement (Dec. 1997), cited as the sole non-patent reference on the face of the '855 patent — is not shielded by § 325(d)-style "already considered" arguments the way it would be if it had been beaten at the Board. It was cited, which matters for § 325(d), but a defendant can still build a materially different § 103 combination around it.
• The eight U.S. patents cited by the examiner (e.g., US 5,557,608 IBM "Method and apparatus for transmission of high priority traffic on low speed communication links"; US 5,497,731 Northern Telecom; US 5,734,867 Motorola "instantaneous preemption of packet data"; US 6,594,278 Cisco; US 6,631,132 Veraz "Urgent packet transmission") are, if anything, more accessible now than they would be after an adverse FWD, because no panel has blessed or rejected them.

Pattern signals. Three observations, each of which cuts against the patent's defensive durability rather than for it:

1. No defensive aggregator has challenged it. I found no Unified Patents, RPX, or other aggregator IPR naming the '855 patent. Aggregators normally pick cheap, low-risk invalidity targets; the fact that this one was left alone likely reflects its expiration (2023-04-21) rather than its strength — there was no ongoing royalty stream to protect after the term ran.
2. The '855 patent was never asserted by an Orckit entity. Orckit litigated many patents against Cisco, Arista, Nokia, and Extreme; the '855 patent sat dormant through all of it and was only asserted after it expired, by a 2025 NPE purchaser (Enhanced Data Streaming LLC). That is a classic pattern of monetizing an old, unvetted asset rather than a patent the owner believed was a strong one.
3. Its claims are broad and the technology is standard-adjacent. The accused functionality in the live case is Multilink PPP over Ethernet / Link Fragmentation and Interleaving — pre-existing industry-standard functionality, with the ppp multilink fragment delay command mapped to the "predetermined maximum time" limitation. Standards-based functionality dated to the 1990s is fertile ground for both § 102 and § 103 challenges. An IPR on this record would be a plausible undertaking; the reason it hasn't happened is economic (expired term), not technical.

Recommended next steps

1. If you are the defendant and you need the FWD link — there isn't one. The operative fact for your invalidity contentions is that no AIA trial has ever been instituted on 6,891,855, so there is no Final Written Decision to cite and no claim has been canceled. Do not let an adversary conflate this patent with the Orckit-family IPRs in the table above (IPR2023-00554 on the '111 patent ended in an FWD indexed as "Patentable"; IPR2023-00401/IPR2024-00026/IPR2024-01237 on the '740 patent involved the "Bruckman" reference). Those are different patents with different claim language.
2. The § 315(b) window, if anyone wants it, closes almost immediately. Service of the Enhanced Data Streaming LLC v. Cisco Systems, Inc. complaint (E.D. Tex. No. 2:25-cv-01038-JRG, filed 2025-10-10) starts a one-year clock for an IPR petition by Cisco or its privies. On today's date (2026-09-28) that window closes on or about 2026-10-10 — roughly twelve days out. I found no petition in that window. If a petition is filed, institution would be due within six months of filing (§ 314(b)) and an FWD within twelve months of institution (§ 316(a)(11)); the trial-stage milestones would be institution decision, PO response, optional MTA, oral hearing, then FWD.
3. Weigh the IPR calculus against the expiration date before filing. The patent's recorded status is "Expired — Lifetime," adjusted expiration 2023-04-21. An expired patent can still be the subject of an IPR, but the Board construes expired claims under the Phillips framework rather than the SAS-era BRI approach, the patent owner cannot amend the claims, and there is no injunctive remedy to enjoin. Meanwhile the § 315(e)(2) estoppel that normally makes an IPR a strategic no-lose proposition for a defendant is only a downside if you file and lose. A district-court invalidity case may be the better venue, particularly because you can reach system/device prior art there — art that an IPR cannot touch under § 311(b).
4. Verify before relying. Run a direct PTAB E2E search on 6,891,855 (https://ptacts.uspto.gov/ptacts/) and a CourtListener/CAFC query for any appeal tied to 2:25-cv-01038. My searches did not surface a Federal Circuit appeal of anything involving this patent, but the earlier-generated litigation section correctly flags that CAFC docket entries often reference only the lower-court case number, so a patent-number-keyed search can miss them.
5. Flag for the record — one internal inconsistency in the prior sections. The previously generated litigation summary refers to "the reissue application number 09/756,554." Application 09/756,554 is the original application that issued as US 6,891,855; the record reflects reissue treatment only as to a different patent in the family (U.S. Patent No. 9,185,151, reissued 2025-04-22 as RE50,398). There is no reissue of the '855 patent in the material provided. Correct this before it propagates into a filing.

Key sources

• Canonical PTAB/OAI data: USPTO Open Data Portal "PTAB proceedings on file" block (empty for this patent).
• PTAB End-to-End: https://ptacts.uspto.gov/ptacts/
• CourtListener: https://www.courtlistener.com/opinion/[10893882](/patent/10893882)/enhanced-data-streaming-llc-v-cisco-systems-inc/
• Underlying district-court order: https://www.govinfo.gov/content/pkg/USCOURTS-txed-2_25-cv-01038/pdf/USCOURTS-txed-2_25-cv-01038-0.pdf
• Patent text: https://patents.google.com/patent/US6891855/en
• Related-family IPR docket (not this patent): https://www.docketalarm.com/cases/PTAB/IPR2023-00554/Cisco_Systems_Inc._v._Orckit_Corporation/

I'll research the assignment correspondents and NPE chain before writing. Let me run several searches in parallel.

Good context. Let me dig into the Orckit IP entity, its principal, and the recent NPE chain.

Assignment & Ownership Analysis — US 6,891,855 B2

Method note: The reel/frame numbers and conveyance types below are taken from the Google Patents legal-events listing for US 6,891,855 (https://patents.google.com/patent/US6891855/en), which reproduces USPTO assignment-record fields. I was not able to retrieve the Assignment Center "correspondent of record" field for each individual reel/frame (no direct API/PACER access in this session; the primary source to verify against is https://assignmentcenter.uspto.gov/ and https://assignment.uspto.gov/patent/index.html). Where I name a correspondent below, I say explicitly which other USPTO record it comes from. I have not invented any correspondent-to-reel pairing.

Inventors

Pattern note: No "inventor flight" signal. Bruckman assigned to his employer promptly, and the patent stayed with the Orckit corporate family for 15 years. The exit was corporate (insolvency), not individual.

Original assignee

Orckit Communications Ltd. (Israel) is the assignee on the issued patent's chain at issue; the record shows a 2002 intra-family transfer to Corrigent Systems Ltd. (Israel), later renamed Orckit-Corrigent Ltd.

• Line of business: DSL/telecom access equipment — Orckit was an HDSL/ADSL systems vendor, NASDAQ-listed, and its engineer Binder sat on ETSI-TM6 / ANSI T1E1.4 standards committees. The patent's own background (variable-rate DSL access, 100 kbps–2.3 Mbps) matches this product line.
• Product embodiment: Orckit/Corrigent shipped access and transport platforms (DSLAM-adjacent and packet-transport gear) in the relevant era; the specification is written around a DSL modem transmitter/receiver pair (FIG. 1, "transmitter 22 and receiver 24 belong to respective DSL modems"). So the original assignee was an operating company, not a holding vehicle.
• Current status: Defunct. Orckit Communications went through Israeli insolvency/liquidation proceedings (an Israeli court trustee issued a report on 2016-02-18, and the proceedings involved the Israeli Office of the Chief Scientist, which had funded the R&D). A litigation transcript records testimony that "I don't think Orckit-Corrigent exists." The US patents were purchased out of that liquidation by Orckit IP, LLC in 2016. (Sources: CourtListener RECAP transcripts for the E.D. Tex. Orckit action, https://storage.courtlistener.com/recap/gov.uscourts.txwd.[1169926](/patent/1169926)/gov.uscourts.txwd.1169926.304.0.pdf and https://storage.courtlistener.com/recap/gov.uscourts.txed.[216428](/patent/216428)/gov.uscourts.txed.216428.324.0.pdf.)

Assignment timeline

• 2001-02-26 (executed) / recorded 2001-03-15 — Reel 011593/0881

  • Conveyance: Assignment (inventor → employer)
  • Assignor: Leon Bruckman
  • Assignee: Orckit Communications, Ltd. (Israel)
  • Correspondent: Not verified for this reel/frame. (Orckit's 2001-era filings were handled by the company's own Israeli/US prosecution contacts; no correspondent field retrieved.)
  • Context: Ordinary employee invention assignment at filing.

• 2002-03-07 (executed) / recorded 2002-04-26 — Reel 012866/0880

  • Conveyance: Assignment
  • Assignor: Orckit Communications Ltd.
  • Assignee: Corrigent Systems Ltd. (Israel)
  • Correspondent: Not verified for this reel/frame.
  • Context: Internal corporate reorganization — the packet-transport business (and its patents) was moved into Corrigent, the Orckit/Cisco-systems spin-out vehicle.

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

  • Conveyance: Security Agreement
  • Assignor: Orckit-Corrigent Ltd.
  • Assignee (secured party): Hudson Bay IP Opportunities Master Fund, LP
  • Correspondent: Not verified. (The financing itself was advised by Schulte Roth & Zabel LLP for Hudson Bay Capital Management LP and by Goldfarb, Levy, Eran & Co. for Orckit, per the 2013 deal announcement on MarketScreener — but that is deal counsel, not the recorded-assignment correspondent.)
  • Context: Securitization of IP as collateral — recorded security interest against the patent to support Orckit Communications' $5.0M senior secured notes (closed 2013-03-18, part of a ~$5.5–8.0M 2013 financing package with Hudson Bay and Networks3). This is a lender lien, not a transfer of title. Source: https://in.marketscreener.com/quote/stock/BEEIO-HONEY-LTD-120961634/news/Orckit-Communications-Ltd-announced-that-it-has-received-5-50-million-in-funding-form-Hudson-Bay-C-38977497/

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

  • Conveyance: Release by Secured Party
  • Assignor: Hudson Bay IP Opportunities Master Fund LP
  • Assignee: Orckit-Corrigent Ltd. (Israel)
  • Correspondent: Not verified for this reel/frame.
  • Context: Lien release — collateral returned; IP title cleared ahead of the eventual sale.

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

  • Conveyance: Change of Name (no change in title)
  • Assignor: Corrigent Systems Ltd.
  • Assignee: Orckit-Corrigent Ltd
  • Correspondent: Not verified for this reel/frame.
  • Context: Change of name only.

• 2016-04-25 (executed) / recorded 2016-05-11 — Reel 038541/0984

  • Conveyance: Assignment
  • Assignors: Orckit Communication Ltd.; Orckit-Corrigent Ltd.; Corrigent Systems Ltd.
  • Assignee: Orckit IP, LLC (Delaware LLC)
  • Correspondent: Not verified for this specific reel/frame, but strongly indicated as Sheehan Phinney Bass & Green, PA, c/o Peter Nieves, 1000 Elm Street, Manchester, NH 03105-3701 — that firm/lawyer appears as correspondence address on the contemporaneous 2016 Statements Under 37 CFR 3.73(b) filed by Orckit IP, LLC for sibling Orckit patents (09/756,946 and 09/472,683) in the same month (June 2016 USPTO Power-of-Attorney notices, USPTO PTAB/PTActs document set). Flag: recurrence across the Orckit IP 2016 acquisition batch, but not confirmed on reel 038541/0984 itself.
  • Context: Fire-sale/insolvency disposition — this is the transfer of the Orckit patent family out of the Israeli insolvency proceeding into a newly formed single-purpose licensing LLC. Orckit IP, LLC's address of record is 831 Beacon Street, Suite 307, Newton, MA 02459, and its owner/CEO is Yehuda Binder (USPTO Reg. No. 73,612).

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

  • Conveyance: Assignment
  • Assignor: Orckit IP LLC
  • Assignee: Enhanced Data Streaming LLC (Texas)
  • Correspondent: Not verified for this reel/frame.
  • Context: Transfer to an NPE/asserter — a second-generation single-purpose LLC that asserted this patent against Cisco within about two months of the recording (see pre-litigation transfer signal below).

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

  • Conveyance: Patent Security Agreement
  • Assignor: Enhanced Data Streaming LLC
  • Assignee (secured party): Wye IP Capital LLC (New York)
  • Correspondent: Not verified — I found no public correspondent or registration detail for Wye IP Capital LLC in the searches returned.
  • Context: Litigation financing / monetization securitization — a security interest granted to a funder ~6 weeks before the Cisco complaint was filed, i.e. the new plaintiff encumbered the patent as collateral for the assertion campaign.

Recorded-assignment gaps worth flagging: the CourtListener transcripts reference Orckit IP transferring patents to a "Smart Path" entity around 2020 and a revenue-share arrangement with a "Mr. Pitcock" (Orckit IP retaining ~42%, Pitcock ~15%, with inventor compensation expressly disclaimed). No such assignment is recorded against US 6,891,855 — the '855 patent stayed with Orckit IP, LLC until the 2025 EDS transfer. I report this as a negative finding on this patent only.

Timeline diagram

timeline
    title Ownership of US 6891855
    2000 : Provisional filed Jul 27
    2001 : Filed by Orckit Communications
         : Inventor assigns to Orckit
    2002 : Assigned to Corrigent Systems
    2005 : Patent issued May 10
    2013 : Hudson Bay security interest
         : Lien released months later
    2014 : Corrigent renamed Orckit Corrigent
    2016 : Assigned to Orckit IP LLC
         : Sale out of Orckit insolvency
    2025 : Assigned to Enhanced Data Streaming
         : Wye IP Capital security interest
         : Suit filed against Cisco

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT. Reel 038541/0984 (recorded 2016-05-11) moves the patent from operating entities (Orckit Communications / Orckit-Corrigent / Corrigent Systems) to Orckit IP, LLC, a Delaware LLC registered to a three-line Newton, MA office suite (831 Beacon St., Suite 307) with no products; the PTAB record lists it as "DE LLC # 5896551." Reel 072030/0801 (recorded 2025-08-15) moves it again to Enhanced Data Streaming LLC (Texas), another productless single-purpose vehicle. Two consecutive operating→licensing-LLC hops.

2. Known asserter in the chain — PRESENT. The assignees are not on the enumerated classic lists (Acacia, Marathon, IV, Wi-LAN/Conversant, etc.), but both are recognized high-frequency asserters: Orckit IP, LLC filed a long campaign (e.g. Orckit IP, LLC v. Cisco, v. Extreme — see Extreme Networks' SEC-filed litigation disclosure at stockanalysis.com, v. Nokia, v. Arista, v. Juniper), and PTAB dockets show it as patent owner in IPR2023-00554, IPR2024-00037, IPR2024-00895, IPR2024-01450. Enhanced Data Streaming LLC is the plaintiff in 2:25-cv-01038 (E.D. Tex.), and the Google Patents "family has litigation" link for this patent points to Unified Patents' litigation record for that Texas case — i.e. the patent is tracked in Unified Patents' asserter dataset. I mark this present on the record evidence of assertion, while noting it is not an enumerated-name match.

3. Repeat correspondent across the chain — UNCLEAR / partially supported. The clearest recurrence I can document is not in the assignment reel fields but in adjacent USPTO records: Yehuda Binder (Reg. No. 73,612) is simultaneously (a) the CEO and owner of Orckit IP LLC, (b) the attorney/agent of record for Orckit IP LLC in PTAB matters (e.g. IPR2024-01450, address 831 Beacon Street Suite 307, email yehuda@maypatents.com — i.e. the affiliated May Patents Ltd. vehicle), and (c) signatory on the 2016 §3.73(b) statements. Separately, Sheehan Phinney Bass & Green (Peter Nieves) recurs across the 2016 Orckit IP acquisition batch. I cannot confirm either as correspondent on the specific reel entries requested, so I decline to call this present without the Assignment Center correspondence fields.

4. Cascading transfers — PARTIALLY PRESENT. Within 2025, three events land in ~7 months (assignment effective 2025-03-09, security interest effective 2025-08-13, recording 2025-09-29), i.e. the new plaintiff took title and immediately pledged the patent. The 2016→2025 gap is long, so the "chained LLCs in <24 months" formulation is not met across the whole chain, only within the 2025 cluster.

5. Pre-litigation transfer — PRESENT. Assignment to Enhanced Data Streaming LLC effective 2025-03-09, recorded 2025-08-15; complaint against Cisco filed 2025-10-10 (per the prior litigation section; Unified lists 2025-10-09). The recording precedes suit by under two months. This is the classic "clean the standing record, then file" sequencing.

6. Bankruptcy fire-sale — PRESENT. The original assignee did not file US Chapter 7/11, but Orckit Communications underwent Israeli insolvency/liquidation; a court trustee's report dated 2016-02-18 documents the proceeding, and the patents were bought out of it by Orckit IP, LLC in April 2016 (reel 038541/0984, executed 2016-04-25). The Israeli Office of the Chief Scientist had funded the underlying R&D, and the presence of government R&D repayment obligations is squarely alleged in the Cisco litigation transcripts. Mark present, with the jurisdiction qualification (Israeli insolvency, not US bankruptcy).

7. Privateering — NOT PRESENT (on this record). No operating company appears to be funding an NPE to assert against competitors here. The original operating company is gone; the assertion is run by serial monetizers (Binder/Tamir lineage → Enhanced Data Streaming → a New York funder), not by a live competitor using a proxy.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. It terminates at an asserting plaintiff and its secured funder.

Verdict

NPE — high confidence. At least four independent strong signals are documented on the face of the assignment record: a 2016 operating-company → Orckit IP, LLC transfer out of insolvency (reel 038541/0984, executed 2016-04-25), a 2025 transfer to Enhanced Data Streaming LLC (reel 072030/0801, effective 2025-03-09, recorded 2025-08-15) followed by suit within ~2 months, a security interest to Wye IP Capital LLC (reel 072978/0765, effective 2025-08-13) evidencing assertion financing, and the institutional asserter identity of both LLCs across numerous district-court actions and PTAB IPRs. The 2013 Hudson Bay entry (reel 030033/0774) is a lender lien with a matching 2013 release (reel 030887/0983) and should not be miscounted as a title transfer.

Contradiction flag (carried from prior sections): the Google Patents record states "Expired – Lifetime," adjusted expiration 2023-04-21, yet the '855 patent was pleaded in a complaint filed 2025-10-10. An expired patent is not infringed, so either the "adjusted expiration" field is a PEDS artifact not controlling term (e.g. a PTA/terminal-disclaimer display quirk) or the assertion is defective. The prior sections already noted this unreconciled point; I did not find a court ruling resolving it, and it is directly relevant to whether the 2025 EDS transfer had any assertable value.

Verification links: USPTO Assignment Search — https://assignmentcenter.uspto.gov/ (search patent number 6891855) and https://assignment.uspto.gov/patent/index.html; Google Patents legal events — https://patents.google.com/patent/US6891855/en; Unified Patents litigation record for the E.D. Tex. case — https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A25-cv-01038.

I'll verify the cited references against live sources before analyzing them.

Prior Art Analysis — US 6,891,855 B2 ("Dynamic packet fragmentation")

Sourcing / method note. I ran live searches against the specific number 6891855 and against several individual cited references. I verified the actual disclosure content of five references directly (US 5,557,608; US 5,497,371; US 6,594,278; US 6,631,132; US 6,212,190) plus the EP 0582537 family. For the remaining references I relied on the citation titles, dates, assignees, and art-unit listing printed on the face of the '855 patent itself (authoritative as to identity of the reference, but not as to its disclosure). I did not reach a USPTO API for the "References Cited" bibliographic record before my tool budget was exhausted, so descriptions for the unverified references are title-derived and labeled as such.

Date flag. The task states "Current Date: April 26, 2026," while my system context is 2026-09-28. The prior Litigation section already noted a June 2, 2026 E.D. Tex. order (Doc. 26/48) that post-dates the task date. I flag the inconsistency rather than resolve it; it does not affect the prior-art analysis below.

Interpretation rule applied. I counted only references printed on the face of US 6,891,855 (per the Google Patents record at https://patents.google.com/patent/US6891855/en). I did not substitute any near-number (e.g., 6,897,185 Lytone/AgroFresh; 6,891,865).

1. The 18 cited references — citation, dates, description

The '855 patent's "Patent Citations (18)" table lists the following. "Pub." = grant date for patents / publication date for applications. §102(e)-type effective US filing date is given where it differs from the printed "priority" date.

Non-patent literature (1): Frame Relay Fragmentation Implementation Agreement, FRF.12, pp. 1–19, Dec. 1997 (Frame Relay Forum). Also incorporated by reference into the '855 specification at the Background section.

* Asterisk = flagged "Cited by examiner" in the Google Patents record; the unmarked entries appear as third-party/IDS-type citations (the dagger symbols did not render in the retrieved text). Verify against the USPTO "References Cited" page if the examiner/IDS distinction matters.

Two §102(e) dates worth flagging:

• US 2003/0103515 A1 (filed 1999-10-26) predates the '855 priority date (2000-07-27) and is available as §102(e) art.
• US 2002/0041595 A1 (Delvaux, filed 2000-10-05) is after the '855 provisional priority (2000-07-27) but before the '855 actual filing (2001-01-08). Its availability turns entirely on whether the challenged claim actually gets the 2000-07-27 provisional benefit. If yes → not §102(e) art; if a claim's subject matter isn't supported by provisional 60/221,039 → it becomes §102(e) art. This is a discrete validity lever, not a mere citation.

2. What the '855 claims actually require (for §102 element-mapping)

All six independent claims (1, 8, 9 method; 10, 17, 18 apparatus) share a common core:

• (A) determine the rate of transmission over the channel;
• (B) receive a datagram for transmission at that rate;
• (C) divide the datagram into fragments of size ≤ a size limit set responsive to the determined rate;
• (D) transmit the fragments.

Independent claim 1/10 adds: (E) the size limit is set such that per-fragment transmit time ≤ a predetermined maximum time, and (F) the datagram is low-priority and that maximum time is the maximum delay applicable to high-priority datagrams on the same channel. Claim 8/17 adds: (G) size limit increases with rate; (H) fixed overhead + variable payload so the payload/overhead ratio increases with rate. Claim 9/18 adds: (G) + (I) size limit set from operator-programmed fragmentation parameters and increased automatically as rate rises, substantially without altering those parameters. Dependent claims 2/11 (100–2300 kbps), 3/12 (DSL access channel), 4/13 (data packet), 5/14 (interrupt low-priority fragment stream to serve a high-priority datagram within the max delay), 6/15 (size limit ↑ with rate), 7/16 (receiver reassembly responsive to the variable rate).

The gating element is (C): dynamic, rate-responsive fragment sizing. Everything else is well-trodden. My analysis below is organized by how close each reference comes to (C).

3. Tiered §102 analysis

Tier 1 — Closest art; rate/delay-driven sizing logic, but not per-packet dynamic sizing

US 5,343,473 (IBM; EP 0 582 537 A2) — the single most relevant reference.
Its EP family member states the invention determines whether preempt/resume is needed "as a function of link speed, acceptable delay and maximum packet size," computing whether a fixed low-priority packet size satisfies the inequality max packet size ≤ link rate × T (worked example: 8 KB @ 18.432 Mbps fails a T = 0.5 ms budget → preemption required; 1 KB passes). This is precisely the rate × delay → size relationship underlying claim 1's element (E) and claim 6/15. However: the output of the computation is a binary protocol-selection decision made at link activation, not a fragment size limit continuously reset responsive to a variable transmission rate. That gap is exactly what the '855 patent touts over the art ("do not permit dynamic size adjustment by the transmitter").

• Potentially anticipates: nothing outright. Element (A) and (E) read on it strongly; (C) does not.
• Best role: primary §103 reference against claims 1, 6, 10, 15 (and, combined with FRF.12, claim 8/17).

US 5,557,608 (IBM Calvignac) — most relevant for the priority/delay architecture.
Verified content: the delay budget T for real-time packets is expressly the driver of preemption granularity ("the preemption granularity can be defined at a block level"); NRT packets are segmented into blocks (N0 = 128 bytes); scheduling policy is chosen by whether the service time of a maximum-length NRT packet exceeds 1.5 ms. It discloses the low-priority/high-priority split, a max delay T for high-priority packets, and preemption/resumption.

• Potentially anticipates claim 5/14 directly (interrupting low-priority transmission to serve a high-priority packet within the delay budget) — though 5/14 depend from 1/10, so they stand or fall with them.
• Against claim 1/10: discloses (F) and most of (E), but the 128-byte block is fixed, not set responsive to a variable channel rate → no anticipation. Strong §103 art.

US 5,497,371 (Northern Telecom Ellis) — fragmentation + priority + overhead, all fixed.
Verified: a "packet fragmentation protocol" lets mixed traffic classes share one link and "gives delay sensitive traffic priority over non-sensitive traffic"; fragments carry P/SEQ/C bits (fixed overhead) with variable payload; worst-case overhead 46 bits/fragment. Relevant to element (H) (fixed overhead + variable payload) and to claims 5/14, 7/16. The fragmentation parameters are not rate-adaptive.

• Potentially anticipates: nothing outright; strong §103 art against claim 8/17 (fixed-overhead/variable-payload concept) and claims 5/14.

US 6,631,132 (Veraz Networks Sourani) — urgent-packet preemption + explicit overhead/length trade-off.
Verified: interrupts an in-progress packet to transmit an "urgent packet," resumes at the interruption point, using stop-point indication bits; the specification expressly discusses the ratio between data amount and header length growing worse as fragments shrink, citing Galand (US 5,956,341) segments of ≤60 bytes + 4-byte header. Relevant to claims 1/10, 5/14 and to the payload/overhead-ratio concept of claim 8/17. No rate-responsive sizing.

• Potentially anticipates: nothing outright; §103 art for claims 5/14 and (with FRF.12) 8/17.

Tier 2 — Fragment structure / overhead-vs-size and delay-class art

FRF.12 (NPL, Dec. 1997). Supplies the fragment header (Beginning/Ending flags, sequence number), the fixed-overhead/variable-payload structure, and the efficiency-vs-payload-size analysis, and states that max fragment size should be configured by the operator per channel for speed/application requirements. That is the exact baseline the '855 patent calls "static fragmentation."

• Potentially anticipates: the (C)/(D) mechanics and element (H) of claims 8/17 read directly on FRF.12; but FRF.12's operator-configured static size is the opposite of element (C)'s "responsive to the determined rate" → no anticipation of any independent claim. It is the primary art for the claim 8/17 overhead-ratio element and is the reference the '855 patent itself distinguishes.

US 6,594,278 B1 (Cisco Baroudi). Verified: stores delay-sensitive info in IP packets, fragments into FRF.12 sub-frames, and stores a payload type in each sub-frame header to distinguish delay-sensitive from non-delay-sensitive traffic, with reassembly. Maps to the low/high-priority distinction (F), sub-frame overhead (H) and receiver reassembly (7/16); fragment size is dictated by the FRF.12 sub-frame format, not by channel rate.

• Potentially anticipates: nothing outright; §103 art (with FRF.12) for claims 1/10, 7/16, 8/17.

US 6,212,190 B1 (Sun). Verified: generating datagrams sized to avoid MTU-driven fragmentation across heterogeneous networks, with explicit attention to header overhead and reassembly cost. Relevant to elements (C)/(D) and to the overhead-ratio motivation of claim 8/17, but the size driver is MTU, not a variable transmission rate.

• Potentially anticipates: nothing outright; §103 art.

Tier 3 — Variable-rate channel / DSL art (supplies "channel having a variable transmission rate" and claims 3/12)

US 5,999,565 A (Cisco). Modifiable number of XDSL modems → a DSL channel whose rate changes. Supplies the variable-rate-channel preamble and claim 3/12 context. (Content not independently verified in this session.)
US 6,366,589 B1 (3Com). Multiple parallel asymmetric xDSL interfaces with reversed asymmetric links — variable/asymmetric line rate. Same role. (Content not independently verified.)
Neither appears to teach rate-responsive fragment sizing.

• Potentially anticipates: nothing; §103 art for the preamble and claims 2/3/11/12.

Tier 4 — Interleaving / preemption / stream-insertion art (claims 5/14, 7/16)

• US 5,734,867 A (Motorola) — "instantaneous preemption of packet data." Directly on the preemption element (claims 5/14). (Title-derived.)
• US 5,742,592 A (Motorola) — communicating data in a wireless system. Low-to-moderate relevance. (Title-derived.)
• US 6,594,249 B1 (Motorola) — inserting data into a data stream in a wireless system. On the interrupt/insert element (5/14). (Title-derived.)
• US 6,633,564 B1 (Nortel) — inserting packets into a data stream. Same. (Title-derived.)
• US 2002/0041595 A1 (Delvaux) — suspend/resume transmission with no significant added overhead. On the interrupt/resume element (5/14). §102(e) date caveat noted above.
• US 2003/0103515 A1 (Brown) — efficient data-transmission control in wireless voice-over-data; on delay-sensitive-vs-data prioritization. (Title-derived.)
• US 6,553,003 B1 (Samsung) — RLP fragmentation/sequencing/retransmission; on the fragmentation + reassembly elements (claims 7/16). (Title-derived.)

Tier 5 — Peripheral

• US 5,173,898 A (IBM) — multiple-access control with order pad passing. Marginal; possibly cited for in-order fragment passing (claim 7/16 ordering). (Title-derived.)
• US 6,026,093 A (Sun) — dispatching data units via a telecommunications network. Marginal; possibly fragmentation/dispatch. (Title-derived.)
• US 6,654,811 B1 (Nokia) — backpressure in a client-server environment. Least on-point; plausibly cited for adaptive rate/flow control. (Title-derived.)

4. Bottom line on anticipation

No reference among the 18 (or FRF.12) is a clean §102 anticipation of any independent claim of US 6,891,855 — i.e., none discloses element (C), the fragment size limit set responsive to the determined (variable) transmission rate. Every reference that addresses size-at-all fixes it statically (FRF.12's operator-configured size; US 5,557,608's 128-byte N0 block), ties it to MTU (US 6,212,190), or uses rate/delay only to make a one-time protocol-enablement decision (US 5,343,473 / EP 0582537). Because dependent claims incorporate the independent claims' limitations, no dependent claim is separately anticipated either.

Ranked by §103 relevance (best combinations):

1. US 5,343,473 / EP 0 582 537 (IBM) — rate × delay → size logic. Primary reference for claims 1, 6, 10, 15.
2. US 5,557,608 (IBM) — max-delay-bounded preemption of low-priority traffic in fixed-size blocks; primary for claims 5/14 and corroborating for 1/10.
3. US 5,497,371 (Northern Telecom) and US 6,631,132 (Veraz) — fragmentation + priority + the fixed-overhead/variable-payload overhead-ratio idea; primary for claim 8/17 and claims 5/14.
4. FRF.12 (NPL) — fragment header structure, fixed overhead vs. variable payload, efficiency-vs-payload analysis, operator-configured size; the reference the '855 patent distinguishes, and the backbone of any claim 8/17 challenge.
5. US 5,999,565 / US 6,366,589 (Cisco / 3Com) — variable-rate DSL channel to close the preamble and claims 2/3, 11/12.
6. US 6,594,278 (Cisco Baroudi) — FRF.12 fragmentation with a delay-sensitive payload-type marker; reinforces (F) and (H) and claim 7/16.

A representative obviousness combination against claim 1 would be US 5,343,473 (rate × max-delay → packet/fragment size) + US 5,557,608 (low/high-priority preemption with a max-delay budget) + a variable-rate DSL reference (US 5,999,565 or US 6,366,589), with FRF.12 supplying the fragment structure. Against claim 9/18, the added hook is FRF.12's operator-configured fragment size, which the '855 patent claims to improve upon by changing size automatically without operator reprogramming — so the art's teaching-away posture (static operator configuration) cuts against, not for, obviousness, and is the patentee's strongest rebuttal theme.

5. Uncertainties and caveats

1. Verified vs. title-derived descriptions. Direct content was verified for US 5,557,608, US 5,497,371, US 6,594,278, US 6,631,132, US 6,212,190 and EP 0582537. All other descriptions are drawn from the titles/dates on the face of the '855 patent; treat them as identity-accurate but disclosure-unconfirmed.
2. Examiner vs. third-party citation flags. Based on the asterisk markers in the retrieved Google Patents text, eight citations appear examiner-cited and ten third-party; the dagger symbols did not render. Confirm via the USPTO "References Cited" page.
3. US 2002/0041595 A1 (Delvaux) §102(e) availability is claim-dependent (its 2000-10-05 filing falls between the '855 provisional date and its actual filing date). This should be resolved claim-by-claim against provisional 60/221,039's disclosure.
4. Live-search priority rule. Where a search result conflicted with my recollection (e.g., US 5,557,608's actual filing 1995-05-22 with EP priority 1994-05-25), I used the search result.
5. Prosecution history not examined. I have no office-action record showing which references the examiner actually relied on for the §102/§103 rejections or how the claim terms "size limit … responsive to the determined rate of transmission" were construed. Any anticipation opinion that matters should be run against the file wrapper and any IPR/validity record.

I'll ground the analysis by pulling the cited prior-art references from the record before building the §103 combinations.

Obviousness Analysis — U.S. Patent No. 6,891,855 B2 ("Dynamic packet fragmentation")

Sourcing rule applied: the full text at https://patents.google.com/patent/[US6891855](/patent/US6891855)/en is authoritative. The prior-art universe analyzed below is the "Prior Art" section of that page, i.e., Patent Citations (18) and Non-Patent Citations (1). I have full-text-level excerpts for only four of the eighteen cited patents plus the NPL; for the remainder I have the record's bibliographic data only. Every characterization is flagged accordingly rather than asserted.

1. Contradictions / flags to carry forward (must read first)

2. Legal framework and effective date

• Pre-AIA 35 U.S.C. §103(a) governs: application 09/756,554 was filed 2001-01-08, before the AIA's 2013-03-16 change, with a §119(e) benefit claim to provisional 60/221,039 (2000-07-27). Pre-AIA §102/§103 applies in court and in any IPR.
• Framework: Graham v. John Deere, 383 U.S. 1 (1966); KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) (motivation may come from "the nature of the problem," "design incentives," "market forces," and "common sense"; "a person of ordinary skill is also a person of ordinary creativity"). MPEP §§2141–2144 supply the enumerated rationales (A)–(G).
• Presumption of validity / burden: §282, clear-and-convincing in court; preponderance in an IPR (§316(e)).
• Analogous art: all cited references are in the same field of endeavor (packet/frame transmission over a bandwidth-limited link), so all qualify as analogous (In re Bigio; MPEP §2141.01(a)).
• Advantage of the record: FRF.12, US 5,497,371 and US 5,557,608 were cited by the examiner (asterisked in the "Citations (18)" table) and US 6,594,278, US 6,633,564, US 6,631,132 are also of record. This means the §103 argument is not new-art — it is an argument that the examiner's implicit motivation finding was wrong. That is a materially harder posture, and it should be acknowledged up front.

3. The prior art of record

3.1 The eighteen cited patents + FRF.12 (dates as given in the record)

The record's algorithmic "Prior Art keywords" are transmission, channel, rate, datagram, fragments — note that "rate" is surfaced as a keyword even though, as discussed in §4, no cited reference was read by the examiner as teaching a rate-setting of fragment size.

3.2 Qualification of the record's art (date discipline)

• All eighteen citations except one have filing/priority dates in 1990–2000-04-13, i.e., before the 2000-07-27 priority date → available under pre-AIA §102(a)/(b)/(e) as applicable. US 5,497,371, US 5,557,608, FRF.12, US 5,173,898, US 5,343,473, US 5,734,867, US 5,742,592 are pre-1997 → unambiguous §102(b) art.
• US 2002/0041595 A1 (Delvaux) has a priority date of 2000-10-05 — after the '855's 2000-07-27 date. Its appearance in the citation list is anomalous and suggests that during prosecution someone applied a later effective date (or that the provisional 60/221,039 did not support the then-pending claims). This is worth verifying in the file wrapper, because it cuts in both directions: if the '855 is not entitled to 2000-07-27 for some claims, then more art (including several "Cited By" items) becomes available.
• The "Cited By" lists are not prior art. They are later publications (e.g., US 7,275,093's priority is 2000-04-26, but US 7,154,854 and US 7,339,892 are 2000-08-18; the rest are 2001–2020). Anything filed after 2000-07-27 cannot be §102(e) art against a claim supported by the provisional. Do not build the obviousness case on the Cited By set.
• No cited reference is commonly owned with the '855 (IBM, Northern Telecom/Nortel, Motorola, Sun, Cisco, 3Com, Samsung, Nokia, Veraz, Frame Relay Forum), so pre-AIA §103(c) disqualification of §102(e)/(f)/(g) art is unavailable to the patent owner.

4. Construction of the dispositive limitations and identification of the point of novelty

Point of novelty. The specification is candid: "methods of fragmentation known in the art require that the fragment size be configured in advance, and do not permit dynamic size adjustment by the transmitter." Stripped of the arithmetic, the '855's asserted contribution is therefore a single concept: let the fragment size follow the current channel rate rather than an operator-fixed value. Everything else in every independent claim is either admitted prior art (FRF.12 fragmentation; DSL environment and its 100 kbps–2.3 Mbps range; ATM 53-byte cells as the delay-reduction alternative) or a mathematical consequence of that one concept. Under KSR, that is the correct level at which to frame the §103 inquiry.

5. Combinations that render the claims obvious

Combination A — the core attack on claims 1 and 10 (and dependents 2–7, 11–16)

FRF.12 + US 5,557,608 + US 5,497,371, optionally reinforced by US 5,995,565 or US 6,366,589 for the variable-rate element.

Why a POSITA would combine them (not hindsight):

1. Same problem, same field, expressly stated. US 5,497,371's abstract names the goal the '855 asserts as its object ("satisfy network delay constraints … while achieving enhanced transmission efficiency"). FRF.12 supplies the fragmentation mechanism and the efficiency curve; '371 supplies the priority/preemption mechanism; '608 supplies the block-size/delay-requirement relationship. Combining references that each solve a different half of the identical problem is the paradigm KSR rationale (A) (known elements combined by known methods → predictable result).
2. Express suggestion in '608. "[T]he selection of the appropriate service policy is dependent on the characteristics of the communication link, the delay requirements of the high-priority packets, and the size of the low-priority packets" is a literal, written directive to size low-priority units as a function of the link and the high-priority delay budget — i.e., exactly claim 1's linkage. That is the strongest available anti-hindsight quotation in the record.
3. Design incentive from FRF.12's own efficiency analysis. FRF.12 states efficiency increases with payload size. FRF.12 then instructs sizing for the slowest case, which the '855's background identifies as the source of the efficiency loss. Once the artisan recognizes (as FRF.12 itself does) that the only reason for the small size is the delay bound, taking the largest fragment the bound permits at the current rate is the obvious efficiency move — KSR rationales (C) and (D).
4. The variable rate is not incidental. DSL links retrain and re-negotiate; US 5,995,565 and US 6,366,589 are of-record examples of systems whose rate is a tracked, changing quantity. KSR rationale (F) (design incentives / market forces): a rate-adaptive link creates pressure to stop hand-tuning parameters.
5. The sizing step is arithmetic, not invention. The specification's own formula, Int{(Max_Wait × Rate)/8} − Overhead, is the definition of a time budget converted to bytes. In re Aller, 220 F.2d 454, 456 (CCPA 1955) ("where the general conditions of a claim are disclosed in the prior art, it is not inventive to discover the optimum or workable ranges by routine experimentation"); In re Boesch; MPEP §2144.05.

Combination B — claims 2, 3, 4, 11, 12, 13 (the "environment" claims)

• Claim 3/12 (DSL access channel): DSL is the '855's own admitted working environment. The specification's Background describes DSL as prior art with rates "from about 100 kbps up to 2.3 Mbps." That is an admission that the claimed environment is prior art; US 5,997,565 is a §102(e) Cisco xDSL reference.
• Claim 2/11 (100 kbps–2300 kbps): a claimed range that (a) covers the admitted ADSL range and (b) is not shown to be critical. In re Woodruff, 919 F.2d 1578 (Fed. Cir. 1990); In re Aller; MPEP §2144.05. No criticality data appear in the specification; Table I shows only smooth, monotonic improvement with no inflection.
• Claim 4/13 ("data packet"): the specification defines "datagram" to include packets. The dependent claim adds nothing.

Combination C — claim 5 / claim 14 (interruption of low-priority fragments)

Directly met, multiple times over by US 5,497,371 (priority change → "low priority packet is interrupted, the complete (C) bit is set to 0"; interrupted low-priority fragments discarded), US 5,557,608 (preempt/resume of the non-real-time packet mid-block with later resumption), US 6,633,564 (interrupting packet inserted between physical-layer blocks; the stated purpose is "so they [packets] can be transmitted anytime without delaying the transmission of higher priority packets"), and US 6,631,132 (urgent packet transmitted before the remainder of the interrupted packet). Motivation: each reference states the same jitter/latency problem the '855 states, and each is expressly directed to bounding the wait of the priority packet. This is the least defensible claim family for the patent owner.

Combination D — claims 6, 8, 17 (increase of the limit with rate; fixed overhead / variable payload with increasing ratio)

• Claim 6/15: follows from Combination A; if the size is set by rate × time budget, the limit increases monotonically with rate. In re Fulton; "discovery of an optimum value of a variable in a known process is normally obvious."
• Claim 8/17 (the ratio limitation) adds nothing patentable. For any fragmentation scheme with a fixed per-fragment overhead (FRF.12's flags + sequence number; '371's 46 bits; '132's described 4-byte segment header), payload/overhead increases monotonically and mathematically as fragment length grows. The limitation is therefore the definition of fixed-overhead fragmentation plus the claim 6 limitation. The specification treats "fixed overhead, variable payload" as the given structure of the fragment (FRF.12) and then claims the arithmetic consequence. This is the single strongest §103 point in the analysis, and it is an inherent property, not a new physical step.
• Reinforcement for "increase with rate": FRF.12's efficiency analysis (longer payload → higher efficiency) is the incentive; the delay bound is the constraint; the current rate is the exchange rate between the two.

Combination E — claims 9 and 18 (operator parameters, automatic increase without alteration)

FRF.12 + US 5,557,608 (or US 5,497,371) + any rate-tracking disclosure of record (US 5,997,565 / US 6,366,589).

• FRF.12 expressly discloses the first half: the maximum fragment size "should be configured by the operator for each channel depending on the speed." So an operator-programmed fragmentation control parameter is admitted.
• The added element is automation of a response to a changing link condition. Automating a manual, condition-responsive parameter adjustment — with a predictable result and no change to the underlying control parameter — is the classic "use of a known technique to improve similar devices in the same way" and "obvious to try" case. KSR, 550 U.S. at 417, 421; MPEP §2144.04(IV).
• Note also that claims 9/18 are framed as a negative limitation ("substantially without alteration of the parameters"). Absence of a step cannot supply novelty where the parameters themselves are conventional; a negative limitation is only limiting when the reference teaches (or the claim would otherwise read on) the excluded positive step. Here FRF.12 teaches the operator does configure initially — the claim does not exclude initial configuration, only re-configuration on rate change. A reference that teaches the operator re-configures on rate change would not teach away; it would merely show the known manual practice that the claim automates.
• Residual strength: the word "substantially" and the recitation of "operator-programmed parameters" give the owner room to argue that no reference teaches a persistent, unaltered parameter set governing an automatically varying limit. This claim is the patent's best remaining life raft, though the claim is still a dependent-style add-on to the same core concept.

Combination F — claims 7 / 16 (receiver-side reassembly "responsive to the variable rate")

Weak limbs, and the patent's own specification cuts against the owner:

• FRF.12 (per the specification's admission) supplies the header/sequence information enabling reassembly and discard-on-loss; US 5,497,371 (receiver extracts priority/sequence/complete fields, discards fragments after an interruption, reassembles), US 5,557,608 (receiver detects the trailer byte, stores non-final blocks, reassembles), US 6,631,132 (receiver parses the stream into packets from the indication bits), US 6,633,564 (reassembly across interrupted blocks) all disclose reassembly that must tolerate variable-size units.
• The '855's own Detailed Description states: "Reassembler 34 does not need to know the data rate of channel 25 in order to reassemble the received packets, since the reassembly process remains essentially the same regardless of changes in the fragment sizes. It is advantageous to provide this rate information to the reassembler, however, since the reassembler can then determine the sizes of fragments that it should expect and can adjust the sizes of its buffers accordingly." A limitation whose disclosed benefit is buffer sizing belongs in the "design choice / predictable result" bucket (KSR rationale (A); In re Harza; "adjusting buffer size to expected data" is routine). Because 7/16 depend from 1/10, they inherit claim 1's narrowness but add no independent obstacle.

6. Consolidated motivation-to-combine showing (ranked by persuasiveness)

1. Express written suggestion — US 5,557,608's statement that the scheme/size depends on link characteristics, high-priority delay requirements, and low-priority packet size.
2. Express efficiency incentive — FRF.12's own efficiency-vs-payload curve, combined with the '855's admission that worst-case configuration caps efficiency (an admission about the prior art's known disadvantage, not a teaching away).
3. Same-problem/goal-directed combination — US 5,497,371 and US 6,594,278 each state the '855's object almost verbatim (delay-sensitive traffic coexisting with bulk data on one link, with priority). US 6,594,278 even performs FRF.12 fragmentation of delay-sensitive IP traffic into Frame Relay sub-frames — the literal bridge between the primary reference and the priority element.
4. Bounded-wait sizing is known — US 6,631,132's description of Galand's ≤60-byte segments "so that segments of high priority need wait only for transmission of a single packet," and '132's own adjustable stop points tied to header/external information. This shows the artisan already sized low-priority transmission units by reference to the priority packet's wait, and already made such locations adaptive.
5. Predictable arithmetic / routine optimization — In re Aller; the specification's own formula is rate × time ÷ 8 − overhead.
6. Automation of a manual step with predictable results — KSR 550 U.S. at 417.
7. Design incentives from variable-rate links — US 5,997,565, US 6,366,589; plus the admitted DSL rate variability.

7. Anticipated patent-owner rebuttals and responses

8. Claim-by-claim vulnerability summary

Bottom line: the claims are, in my assessment, more likely than not obvious over the combination already of record (FRF.12 + US 5,557,608 + US 5,497,371, with US 5,995,565 / US 6,366,589 for the rate-varying link and US 6,594,278 / US 6,633,564 / US 6,631,132 for the priority handling), with claims 8 and 17 the weakest to defend and claims 1, 9, 10, and 18 the strongest. The outcome turns almost entirely on two questions: (i) whether FRF.12's "configure depending on the speed of the channel" is read as teaching a size limit set responsive to the determined rate, and (ii) whether US 5,557,608's express statement about link characteristics + delay requirements + low-priority packet size supplies the priority-differentiated time budget. If both are answered yes, essentially every claim falls.

9. Procedural / strategic notes

• IPR is available for a pre-AIA patent and pre-AIA §102/§103 applies. Because the '855 is expired (adjusted expiration 2023-04-21), the Board would apply Phillips claim construction rather than BRI, and the practical payoff is limited to historical damages and estoppel — but a final written decision of unpatentability remains a complete defense in the pending E.D. Tex. action (2:25-cv-01038-JRG, Enhanced Data Streaming LLC v. Cisco Systems, Inc.).
• All the key references are of record in the '855 (FRF.12, '371, '608 asterisked as examiner citations). A petition therefore cannot argue "art not considered"; it must argue that the examiner's implicit motivation finding was legally erroneous — a heavier but not impossible burden, and one that pairs well with a §112/claim-construction challenge to "set … responsive to the determined rate" as unbounded.
• File-wrapper retrieval is the highest-value next step: the examiner's reasons for allowance, any applicant remarks distinguishing FRF.12, and any evidence of the '855's effective priority date (see the Delvaux anomaly, §3.2) would materially sharpen or blunt the motivation showing.
• Litigation linkage: the EDS complaint's mapping of claim 1 to the ppp multilink fragment delay parameter is, for invalidity purposes, an admission that the asserted claim reads on a standardized time-budget fragmentation control — reinforcing rather than rebutting the §103 case. (Rule 12 practice does not reach invalidity; the June 2026 order reported in the prior section is unverified here — see §1.)

10. What I could not verify (stated so it is not mistaken for analysis)

1. Full text of fourteen of the eighteen cited patents. For US 5,343,473, 5,734,867, 5,742,592, 5,997,565, 6,366,589, 6,553,003, 6,594,249, 6,212,190, 6,026,093, 5,173,898, 6,654,811, and US 2003/0103515 A1, I relied on the record's titles/dates/assignees and on the descriptions given in the references I did retrieve. Their specific disclosures (e.g., whether US 5,997,565 expressly computes a current aggregate DSL rate) are unverified and should be confirmed against full text before being relied on.
2. RFC 1990 Multilink PPP / LFI standards text. These are not in the page's Prior Art section, but the accused functionality and the claim mapping strongly suggest them. I did not retrieve them in this session and therefore do not assert their content; they are the most important supplementary art to search next.
3. The '855 provisional (60/221,039) contents, beyond the record's statement that FRF.12 was appended to it. This governs whether the 2000-07-27 date holds for each claim — and therefore whether Delvaux and some Cited By items are art.
4. Any secondary-considerations evidence. None appears in the retrieved record; absence of evidence is not proof of absence, but nothing supplied supports a nexus-based rebuttal.
5. The June 2, 2026 district-court order and the Doc. 48 / Dkt. 26 discrepancy (see §1) — unverified from an April 26, 2026 vantage point.
6. US 6,891,855's own examination record (art unit, primary examiner, reasons for allowance) — not present in the retrieved record; not guessed.