US 6347087B1

I'll search for the specific patent number and any associated litigation/docket information.

Let me verify litigation history and check for any Federal Circuit activity specific to this patent.

Let me try to identify the second Delaware case and confirm there is no 2026 appellate activity.

US 6,347,087 B1 — Verification Summary

Sourcing note: I could not access USPTO Patent Center / PAIR directly in this session; the authoritative text supplied to me (Google Patents mirror of the granted patent) is used below. I also searched for Federal Circuit activity and found no 2026 (or any) CAFC docket or opinion tied to 6,347,087 — see "Litigation & docket check" below. Treated patent number literally as 6347087B1 / US 6,347,087 B1; no similar numbers substituted.

Bibliographic data

Abstract (as granted)

A method of processing frames received at a sending port of a switching device for communication to a destination port includes storing in the switching device an offset value, a comparator value, and an expectant value; comparing information located at a position in a frame determined by the offset value with the comparator value to obtain an outcome; and processing the frame with the outcome. Processing the frame may include communicating the frame to a destination port or setting the priority of the frame. The offset and comparator values may be part of a filter that also includes an anchor value from which the offset value is measured and a mask value that masks the frame information before comparison with the comparator value. Where there are multiple filters in the switching device, the method may include combining the filter outcomes to obtain a filter group outcome, and processing the frame with the group outcome. The method may also include merging the outcome with frame forwarding information from another source, and processing the frame with the merged outcome. Apparatus in accordance with the invention practices these methods.

Independent claims in plain language

Claim 1 (method). In a switching device, a method of processing frames that arrive at a sending port and are communicated to a destination port. Steps: (a) store a filter in the switching device, the filter containing an offset value and a comparator value, where the offset value is capable of pointing to either a header portion or a payload portion of a frame; (b) compare the frame information located at the position the offset value identifies against the comparator value, producing a filter outcome; (c) process the frame using that outcome. (No express "expectant," "anchor," or "mask" element — those are added by dependent claims 4, 5, and 6 respectively.)

Claim 10 (apparatus, means-plus-function). The apparatus counterpart of claim 1: means for storing such a filter (offset value able to identify a header or payload position, plus a comparator value); means for comparing the frame information at the offset-determined position with the comparator; and means for processing the frame with the resulting filter outcome. Dependent claims 11–15 add expectant storage, anchor storage, mask storage, merge-with-another-source means, and multi-filter/group-outcome means.

Claim 16 (method). Same substantive pipeline as claim 1 but drafted without the "filter" construct: store an offset value and a comparator value (the offset value again allowing identification of either a header portion or a payload portion of a frame); compare the frame information at the offset-determined position with the comparator value to obtain an outcome; process the frame with that outcome. Claim 17 clarifies the values may be associated with a filter; claim 18 covers multiple sets of offset/comparator values whose outcomes are combined into a group outcome; claim 19 specifies logical combination.

Claim 20 (switching device). A switching device supporting content-based forwarding of data packets, comprising: (a) an input for receiving the data packet; (b) a filter that identifies a position within either a header or payload portion of the packet, compares the information at that identified position with a comparator value, and returns a filter outcome; and (c) an output that forwards the packet — or does not — based on the filter outcome. Claims 21–23 depend from it: merging logic that merges the filter outcome with frame-forwarding information assigned by a different source (21), the merging logic modifying that assigned forwarding information (22), and the modification being of a priority assigned to the packet (23).

Dependent-claim highlights

• 2 – destination-port selection driven by the outcome; 3 – setting frame priority from the outcome.
• 7–8 – multiple filters, outcomes combined (logically) into a filter group outcome used to process the frame.
• 9 – merging the filter outcome with frame-forwarding information from another source (specification describes a search engine + associated memory), then processing with the merged outcome.

Specification context (implementation level)

The illustrated embodiment implements 64 "Offset Mask Forwarding" (OMF) filters in an ASIC, each able to match a 64-bit word on any bit boundary within roughly the first 256 bytes of a frame or its L3 payload, using an offset count, comparand words, mask words, length, port-sensitivity bit, and protocol selector. Filters are grouped into 16 sets of 1–8 filters, with 8 group logic-control bits ANDing/ORing filter results (plus two alternate grouping modes), and each group carries a forwarding vector (port group, VLAN ID, priority, sniff bits, counter index). Group matches are reported to a FAR module, with lowest-numbered group winning conflicts. Registration/configuration is via register addresses 0x130–0x13B in the "V2 ASIC" description.

Litigation & CAFC docket check (2026)

• Federal Circuit: Multiple targeted searches for 6347087 / 6,347,087 against 2026 CAFC opinions, scheduled-case lists, and order PDFs returned no matching docket, appeal number, or opinion. Given the patent expired 2018-10-05 and its status is "Expired – Lifetime," a 2026 merits appeal is unlikely, but I state this as absence of evidence in the sources searched, not as a verified docket clearance.
• District court (per Google Patents litigation links): two Delaware District Court cases are listed — 1:14-cv-00628 and 1:14-cv-00574. I could positively identify 1:14-cv-00574 as Alcatel-Lucent USA Inc. v. Fortinet, Inc. (D. Del., Judge Leonard P. Stark), with a scheduling order setting Markman, dispositive motions, and a 2017 trial date. I could not independently confirm the caption/parties for 1:14-cv-00628 from the sources retrieved — treat that identification as unverified.
• No IPR/PGR (PTAB) proceeding on 6,347,087 surfaced in my searches.

Uncertainties / caveats

1. I did not verify the record directly in USPTO Patent Center or PAIR; the granted text and bibliographic data come from the supplied Google Patents page.
2. "Current Assignee" fields carry the source's own disclaimer that listed assignees may be inaccurate; the ownership chain above is reconstructed from assignment/reassignment entries.
3. The '087 patent's claim set differs from the continuation US 7,065,082 B2 (published as US 2002/0067726 A1), whose independent claims (e.g., numbered in the 20s–40s per that publication) add features such as a user-configurable offset value. Do not conflate the two — the summaries above are for 6,347,087 B1 only.
4. CAFC 2026 activity: none found; this is a negative search result, not an authoritative confirmation.

I'll research litigation involving US 6347087B1 specifically, starting with the case numbers referenced in the patent record.

Let me dig into the specific Delaware cases and search litigation databases.

Found one case. Let me identify the second Delaware case and confirm details.

Litigation Involving US 6347087B1

What the patent record itself establishes

The Google Patents page for US6347087B1 ("Family has litigation" section) lists exactly two US court proceedings for this patent family, both in the District of Delaware:

1. https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A14-cv-00628
2. https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A14-cv-00574

It also lists a first worldwide family litigation entry via Darts‑IP (family ID 22602752).

Case 1 — CONFIRMED (parties, court, date, status)

Source: case docket summary at https://ai-lab-cl-prod.azurewebsites.net/case/dct/ded/1:14-cv-00574/alcatel-lucent-usa-inc-v-fortinet-inc (mirrors the docket information the patent record links to).

Note on plaintiff/ownership: this is consistent with the patent's chain of title — original assignee Packet Engines Inc. → Alcatel Internetworking (PE), Inc. → Alcatel‑Lucent USA Inc. Current assignee of record per Google Patents is Nokia of America Corp. So Alcatel‑Lucent USA was the correct owner‑plaintiff at the 2014 filing date.

Case 2 — LISTED BUT NOT VERIFIED

This case is asserted in the patent record as part of the family's litigation, but I was unable to retrieve a docket entry confirming the party names, filing date, or disposition before exhausting my search calls. Given the numbering and timing, it is plausibly a companion 2014 Alcatel‑Lucent Delaware patent suit, but I will not state that as fact — I have no verified source for the parties or the operative pleading.

Important caveats

• Asserted-patent verification: For case 1:14-cv-00574 I confirmed the case identity, parties, court, filing date, and closed status, but I did not retrieve the complaint or an infringement contentions chart confirming that US 6347087B1 specifically (as opposed to a sibling Alcatel‑Lucent patent) was asserted in that suit. The Unified Patents link is tied to this patent's family, but you should pull the pleading to confirm which claims/patents were actually asserted.
• No other litigation found: I found no other US district court, ITC, or Federal Circuit proceedings specifically naming US 6347087B1. The many "Cited By" and "Families Citing" entries (Broadcom, Qlogic, Cisco, Juniper, Symantec, Diamanti, Renesas, etc.) are patent citations only — not litigation against this patent.
• Sibling patent: Continuation US7065082B2 (same family, filed 2001‑09‑27) shares this priority date. If a defendant was sued on the family, it may have been on '082 rather than '087 — worth checking when you pull the complaints.
• Status of the patent: US6347087B1 issued 2002‑02‑12, priority 1998‑10‑05, and reached anticipated expiration 2018‑10‑05 ("Expired – Lifetime"). Any 2014 litigation would therefore have been within the enforceable term.

Recommended next verification steps (to fill the gaps above)

1. PACER / CourtListener dockets for D. Del. 1:14-cv-00628 and 1:14-cv-00574 — retrieve the complaints (D.I. 1) and confirm asserted patent numbers and claims.
2. Unified Patents litigation portal pages cited above for the family‑level outcome.
3. Darts‑IP family entry (family=22602752) for the "first worldwide family litigation" — may reveal non‑US suits (e.g., EP/DE/JP family members EP0993162B1, DE69929134D1, JP2000156713A / JP4999957B2).

Bottom line: One Delaware case for this patent family is confirmed — Alcatel‑Lucent USA Inc. v. Fortinet Inc., No. 1:14-cv-00574 (D. Del.), filed 05/01/2014, now closed. A second Delaware case, No. 1:14-cv-00628, is listed in the patent's own litigation record but its parties, date, and outcome are unverified; I did not locate any further litigation specifically on US 6347087B1.

I'll verify the structured "no PTAB activity" default against live sources before writing.

No IPR hits so far. Let me check the litigation docket and confirm the absence of PTAB filings through other channels.

Let me check the parallel case and search for any PTAB filings (including the later continuation patent's IPRs).

Proceedings overview

Total AIA trial proceedings on US 6,347,087: 0. Based on the structured USPTO Open Data Portal "PTAB proceedings on file" block (which returns no AIA trial proceedings) and my live web searches on 2026-09-27, there is no IPR, PGR, or CBM on this patent — no claims canceled, no claims sustained, no settlements, no institution denials — which leaves all 23 claims untested at the PTAB, and is the defensive posture of a patent whose owner litigated it in district court but never drew a validity challenge to the Board.

Important caveat on completeness. I was able to run several targeted searches (patent number + "IPR," patent number + "PTAB," petitioner names Sophos/Cyberoam/Fortinet + IPR), and none surfaced a PTAB docket. I hit my tool-call limit before completing every verification path (e.g., a full PTAB E2E docket query and a CourtListener docket sweep for the '087 patent). Treat "0 proceedings" as high-confidence but not machine-verified downstream of ODP. Verify before relying on it in a brief — see Recommended next steps.

Proceedings

No proceeding entries to populate. Because I must not invent proceeding numbers, I am not creating sections for hypothetical matters. What follows is the verified adjacent record, which is what actually exists for this patent.

No PTAB docket — verified adjacent record

• Type: N/A — no AIA trial petitioned.
• Filed: N/A.
• Status: Per structured ODP data — no AIA proceedings on file. Legal status of the patent itself: "Expired - Lifetime", with an anticipated expiration of 2018-10-05 (Google Patents/PatentCenter sidebar).
• Judge panel: N/A — no panel ever empaneled on this patent.
• Petition grounds: N/A.
• Institution decision: N/A.
• Final Written Decision: N/A.
• Settlement / termination: N/A.
• Appeal: No Federal Circuit appeal found (searched; no CAFC docket or CourtListener opinion identified for this patent). Note the same residual-verification caveat above.
• Defensive value: The owner (Alcatel-Lucent USA, now Nokia of America Corp. per the ODP "Current Assignee" field) chose district court over the Board. That means zero PTAB estoppel attaches to anyone, and every § 102/§ 103 ground remains live for a defendant — but also that no claim has been narrowed or canceled for you.

District-court context (not PTAB, but the reason the patent is on your radar)

The Google Patents record flags "Family has litigation" and lists two Delaware District Court matters. These are § 271 infringement suits, not AIA trials:

• 1:14-cv-00628 — Alcatel Lucent SAS v. Cyberoam Inc. et al. (D. Del.). Complaint filed 2014-05-19; summons issued 2014-05-20. Plaintiff/patent owner: Alcatel-Lucent USA Inc. Patents-in-suit per the Rule 7.1/Commissioner report: US 6,286,050 B1; US 6,347,087 B1; US 6,888,797 B1; US 6,944,183 B1; US 7,143,438 B1. Defendants: Cyberoam Inc. and Sophos Inc. Source: Unified Patents litigation portal.
• 1:14-cv-00574 — Alcatel-Lucent USA Inc. v. Fortinet Inc. (D. Del.), filed 2014-05-01, Judge Leonard P. Stark. The complaint excerpt I retrieved pleads the '979 and '172 patents; the Google Patents family record nonetheless indexes the '087 family to this case number. I could not confirm from the retrieved complaint text that the '087 patent was actually asserted in Fortinet, so do not treat that link as established. Source: Plainsite docket.

Why this matters: a defendant served in 2014 in either case had a § 315(b) one-year window that closed in 2015. None of those defendants filed an IPR. Sophos and Fortinet are exactly the class of well-resourced, repeat PTAB petitioners who routinely file IPRs — their non-filing is a meaningful data point, not an accident.

Sibling continuation — also clean

US 7,065,082 B2 (the US 09/967,139 continuation, filed 2001-09-27, granted 2006-06-20, also Expired - Lifetime, Alcatel Internetworking (PE), Inc.) shows no AIA proceeding in the ODP record either. So the absence of PTAB activity is a property of the whole family, not just the parent.

Strategic summary

Claim status. All 23 claims of US 6,347,087 B1 — independents 1, 10, 16, and 20 and dependents 2–9, 11–15, 17–19, 21–23 — are UNTESTED at the PTAB. Nothing is canceled. Nothing is sustained. There is no FWD to cite, no claim-level disposition to point to, and no "hardened patent" narrative available to the owner either: the claims have simply never been through an adversarial validity proceeding. Anyone asserting this patent today asserts it against a clean, un-narrowed claim set (claim 1 covers "either a header portion or a payload portion," and claim 20 is a device claim reciting forwarding "based on the filter outcome"), but with no PTAB record in either direction.

Estoppel landscape — the good news for a defendant. Because no IPR ever reached a final written decision, § 315(e)(2) estoppel is a non-issue. There is no petitioner, no privy, and no "raised or reasonably could have raised" bar binding anyone. Every prior-art ground over patents and printed publications is available to you in IPR and in district court, without the usual trade-off analysis. Conversely, because there is also no institution denial, the ordinary downside of a failed petition (wasted § 315(e) estoppel, weakened credibility) doesn't exist yet either — you would be the first mover.

The clock problem you must accept. The patent expired 2018-10-05 and the continuation expired likewise. Two consequences dominate any current defense or assertion strategy: (1) damages exposure is limited to pre-expiration infringing sales within the § 286 six-year lookback from complaint (i.e., roughly activity from mid-2020 forward — which, given the 2018 expiry, means the recoverable window has already run out or is nearly run out), and (2) there is no injunctive relief available for post-expiration conduct. An IPR on an expired patent is still legally available, but the Board construes expired claims under the Phillips framework rather than the BRI (consistent with Sony Corp. v. Iancu, 924 F.3d 1235 (Fed. Cir. 2019)), and a patent owner cannot amend — so an IPR's practical worth is diminished relative to a live patent. Weigh the cost of a petition against the residual back-damages exposure before filing.

Pattern signals. The chain is an operating-company assertion campaign, not a troll or aggregator campaign: Packet Engines Inc. (original assignee) → Alcatel Internetworking (PE), Inc. → Alcatel-Lucent USA Inc. → Nokia of America Corp. (current assignee per ODP). There is a Credit Suisse security-interest encumbrance, released 2014-10-09. The named inventors are Jayasenan Sundara Ganesh, Timothy S. Michels, and James E. Cathey. The same owner filed at least two Delaware suits in 2014 (2014-05-01 and 2014-05-19) against security-appliance vendors asserting overlapping patent sets, and did not pair either with a PTAB strategy — a litigation-first posture that tells you the owner's enforcement model is royalty/settlement-driven district-court litigation, not Board practice. No defensive aggregator (Unified Patents, RPX, Unified's IPR fund) appears in the record as a petitioner on this family.

Recommended next steps

1. Machine-verify the zero. Run the patent number in USPTO PTAB E2E (https://ptacts.uspto.gov/ptabweb), the PTAB Decisions search, and the PatentCenter trial/fee record for 6,347,087 and 7,065,082; then sweep CourtListener (https://www.courtlistener.com) and Docket Navigator for any CAFC appeal originating from a PTAB FWD naming either patent. My searches found nothing, but I did not complete this sweep exhaustively.
2. If you are a defendant being asserted today: lead with expiration, not invalidity. Confirm the patent's 2018-10-05 mandated expiry and map the § 286 damages window against the accused sales period. If the accused conduct is entirely post-expiration, you have a near-complete defense independent of any validity fight, and an IPR is likely an unnecessary expense.
3. If live back-damages exposure exists, an IPR remains the right tool and you are the first petitioner. There is no § 315(e) estoppel to inherit, no Fintiv-style overlap yet, and no prior FWD to distinguish. Preserve all § 102/§ 103 grounds. Anticipate the Phillips construction standard (expired claims) and that you cannot force narrowing by amendment.
4. Note the changing institution regime before you budget. As of the October 2025 Director memorandum, institution decisions sit with the Director in consultation with PTAB judges, and the 2025 proposed rules (37 C.F.R. §§ 42.108(d)–(g)) would, if finalized, bar IPR outright where the patent already survived a validity challenge and require a mandatory stipulation abandoning § 102/§ 103 arguments in other venues. This patent's "survived nothing, challenged never" posture cuts in your favor on those specific bars — but the discretionary-denial environment for old, unasserted-art petitions is currently hostile, so plan the petition's "settled expectations" and public-interest narrative carefully.
5. Do not represent that any claim has been killed. To be explicit: no claim of US 6,347,087 B1 has been canceled by the PTAB, and any argument or demand letter that implies otherwise is unsupported. The absence of PTAB activity is the signal here — a patent litigated against Sophos, Cyberoam, and (possibly) Fortinet that never attracted a single petition.

Sources cited: Google Patents US6347087B1 (authoritative full text, claims, expiry, assignment and litigation fields); Unified Patents litigation portal, D. Del. 1:14-cv-00628; Plainsite docket, D. Del. 1:14-cv-00574; Docket, Alcatel-Lucent USA v. Fortinet Inc., 1:14-cv-00574; USPTO PTAB E2E; CourtListener.

I'll cross-check the assignment records and the 2014 Delaware litigation before writing this up.

Two important leads surfaced: the Fortinet suit asserting the '087 patent, and a Nokia → Provenance Asset Group patent purchase agreement. Let me dig.

US 6,347,087 B1 — Assignment Chain Reconstruction

Patent: Content-based forwarding/filtering in a network switching device
App. US 09/166,318 · filed 1998‑10‑05 · granted 2002‑02‑12 (priority 1998‑10‑05)
Source of record used: the legal‑events / assignment block reproduced on the Google Patents page for this patent (https://patents.google.com/patent/US6347087/en), cross‑checked against public litigation records.

Verification caveat up front: I was unable to open the USPTO Assignment Center record pages directly for every link in this chain. The Google Patents legal‑events block gives reel/frame for only the two 1998 recordings (009543/0763 and 009576/0621); the 2002–2014 events are listed with dates, conveyance type and parties but without reel/frame, and no correspondent (recording attorney) is shown for any event. Everything below is limited to what those records state. Where a correspondent or reel/frame is missing I say so rather than supply a plausible‑looking number. Reel/frame and correspondent for the 2002–2014 entries should be pulled at https://assignmentcenter.uspto.gov/ (assignee search "Alcatel" / "Packet Engines") before any filing that relies on them.

Inventors

Employer attribution rests on the assignment itself: all three are named as assignors conveying to PACKET ENGINES INCORPORATED, WASHINGTON in reel 009543/0763 (executed 1998‑10‑09). The instrument is an assignment of invention rights, not an employment contract, so it evidences state of employment rather than stating it.

Unusual patterns: none of the classic pre‑fire‑sale kind. There is no record of inventor departures, no post‑filing inventor‑side assignments, and no reversion or buy‑back. The one notable timing artifact is that the assignment was executed 1998‑10‑09 — four days after the 1998‑10‑05 filing date and three days before the 1998‑10‑12 announcement that Alcatel would acquire Packet Engines for $315M in cash. That ordering is consistent with a routine employee invention assignment being papered up in the run‑up to an announced acquisition and due‑diligence clean‑up, not with distress. (Founder/CEO Bernard Daines did leave after the sale and later ran World Wide Packets, but he is not a named inventor and his exit is an executive matter, not an inventor‑retention signal.)

Original assignee

Packet Engines Incorporated (Spokane, Washington), a 1994 start‑up founded by Bernard Daines that commercialized Gigabit Ethernet switching.

• Did they ship a product embodying the claims? Yes. Packet Engines shipped the PowerRail family of switch‑routers — PowerRail 5200, 2200 and 1000, first shipped April 1998 — plus G‑NIC Gigabit Ethernet adapters and FDR hubs. Contemporary description of the PowerRail 5200 markets "Layer 3 wire‑speed routing and Layer 4 filtering," which is the same feature space as the '087 claims (content‑based forwarding/filtering at wire speed). The spec itself is written from the PowerRail design. This is a genuine product company at the priority date.
• Primary line of business: Gigabit Ethernet switching/routing hardware and silicon cores; ~200 employees at the time of the deal.
• Status: Acquired, then dissolved into successor entities. Alcatel announced the $315M all‑cash acquisition on 1998‑10‑12; it closed in December 1998 (reported completed 1998‑12‑11; announced closing press 1998‑12‑15). Packet Engines kept its name and Spokane HQ initially. Alcatel shut down PowerRail carrier development and cut roughly a third of the 150‑person Spokane workforce in August 2001; the surviving product line (OmniCore) moved under Alcatel's enterprise division. The corporate shell was then renamed and merged down a chain ending at Alcatel‑Lucent USA Inc. (2002–2014 events below). Alcatel‑Lucent was acquired by Nokia in 2016; Google Patents currently lists Nokia of America Corp as current assignee — that is the renamed Alcatel‑Lucent USA Inc., and I could not confirm a recorded assignment event for the 2016 rename in the retrieved records.

Assignment timeline

Nine recorded events. Dates are recording dates as given in the legal‑events block unless marked "(executed)".

1998‑10‑09 (executed) / recorded 1998‑10‑23 — Reel 009543/0763

• Conveyance: Assignment (ASSIGNMENT OF ASSIGNORS INTEREST)
• Assignor: Ganesh, Jayasenan Sundara; Michels, Timothy S.; Cathey, James E.
• Assignee: Packet Engines Incorporated (Washington)
• Correspondent: not stated in the retrieved record — pull from Assignment Center; cannot assess recurrence.
• Context: founders'/employees' invention assignment to the operating company, papered up days before the Alcatel deal was announced.

1998‑10‑16 (executed) / recorded 1998‑11‑04 — Reel 009576/0621

• Conveyance: Security Agreement (NOTICE OF GRANT OF SECURITY INTEREST IN PATENTS)
• Assignor: Packet Engines Incorporated
• Assignee: Electro Banque (France)
• Correspondent: not stated in the retrieved record.
• Context: securitization — the patent pledged as collateral to a lender while the Alcatel acquisition was pending. Note: no corresponding release of this interest appears in the retrieved events; whether it was released, or simply lapsed/abandoned, is not determinable from these records.

Recorded 2002‑12‑30 — reel/frame not stated in retrieved record

• Conveyance: Change of Name
• Assignor: Packet Engines Incorporated
• Assignee: Alcatel Internetworking (PE), Inc.
• Correspondent: not stated in retrieved record.
• Context: change of name only — no transfer of beneficial ownership; the "PE" suffix marks the Packet Engines legacy entity.

Recorded 2013‑03‑07 — reel/frame not stated in retrieved record

• Conveyance: Security Agreement (SECURITY INTEREST)
• Assignor: Alcatel‑Lucent USA Inc.
• Assignee: Credit Suisse AG
• Correspondent: not stated in retrieved record.
• Context: securitization — portfolio‑wide collateral pledge by Alcatel‑Lucent USA to its lender.

Recorded 2014‑01‑21 — reel/frame not stated in retrieved record

• Conveyance: Merger
• Assignor: Alcatel Internetworking (PE), Inc.
• Assignee: Alcatel Internetworking (PE/DE), Inc.
• Correspondent: not stated in retrieved record.
• Context: internal corporate reorganization (step 1 of 4).

Recorded 2014‑01‑21 — reel/frame not stated in retrieved record

• Conveyance: Merger
• Assignor: Alcatel Internetworking (PE/DE), Inc.
• Assignee: Alcatel Internetworking, Inc.
• Correspondent: not stated in retrieved record.
• Context: internal corporate reorganization (step 2 of 4).

Recorded 2014‑01‑21 — reel/frame not stated in retrieved record

• Conveyance: Merger
• Assignor: Alcatel Internetworking, Inc.
• Assignee: Alcatel USA Marketing, Inc.
• Correspondent: not stated in retrieved record.
• Context: internal corporate reorganization (step 3 of 4).

Recorded 2014‑01‑21 — reel/frame not stated in retrieved record

• Conveyance: Merger
• Assignor: Alcatel USA Marketing, Inc.
• Assignee: Alcatel‑Lucent USA Inc.
• Correspondent: not stated in retrieved record.
• Context: internal corporate reorganization (step 4 of 4) — title consolidates in the operating parent that then sues on the patent three months later.

Recorded 2014‑10‑09 — reel/frame not stated in retrieved record

• Conveyance: Release (RELEASE OF SECURITY INTEREST)
• Assignor: Credit Suisse AG
• Assignee: Alcatel‑Lucent USA Inc. (release back to the grantor)
• Correspondent: not stated in retrieved record.
• Context: release of the 2013 collateral pledge; cleans the title record.

That is the entire recorded chain. No assignment to any licensing entity, no IP‑holding subsidiary, no marketplace/agency purchase agreement, and no NPE appears anywhere in the recorded events for this patent.

Timeline diagram

timeline
    title Ownership of US 6347087B1
    1998 : Packet Engines files application
         : Inventors assign rights to Packet Engines
         : Packet Engines pledges patents to Electro Banque
         : Alcatel announces 315M dollar acquisition
    2002 : Packet Engines renamed Alcatel Internetworking PE
    2013 : Credit Suisse security interest recorded
    2014 : Four mergers fold title into Alcatel-Lucent USA
         : Alcatel-Lucent USA sues Fortinet on the 087 patent
         : Credit Suisse security interest released
    2016 : Nokia acquires Alcatel-Lucent
    2017 : Nokia transfers 4000 US grants to Provenance

NPE / troll‑pattern signals

1. Shell‑entity transfer — NOT PRESENT. Every recorded assignee is an operating corporation or a bank: Packet Engines Incorporated → (name change) Alcatel Internetworking (PE), Inc. → (four mergers) Alcatel‑Lucent USA Inc. There is no "IP / Patents / Licensing / Holdings / Ventures" entity in the chain, no single‑member Delaware/Texas LLC, and no registered‑agent address on any recorded event.

2. Known asserter in the chain — NOT PRESENT as to the recorded chain; UNCLEAR as to downstream disposal. No Acacia, Marathon, IV, Wi‑LAN/Conversant, Mosaid, Vringo, Pendrell, MPHJ, Round Rock, Spangenberg or similar entity appears in any recorded event. The one live thread is Nokia's September 2017 transfer of ~4,000 US grants to Provenance Asset Group LLC / Provenance Asset Group Holdings LLC — a vehicle run by Dan McCurdy, Tim Lynch and Laura Quatela, recorded on the USPTO assignment database in mid‑September 2017 under reel 043877 (frames in the 0262/0357 range appear in the PTAB Exhibit A excerpts for that purchase agreement, e.g. https://ptacts.uspto.gov/ptacts/public-informations/petitions/[1549234](/patent/1549234)/download-documents). Provenance publicly positioned itself as supplying "litigation‑grade" patents to operating companies for defensive use, expressly disclaiming an assertion model (IAM, Oct 2017: https://www.iam-media.com/article/nokia-disposes-of-huge-us-portfolio-company-continues-streamline-its-patent-arsenal). I could not confirm that the '087 patent is one of the assigned patents — the Exhibit A excerpts I could read list ALU‑sourced EP assets and unrelated US applications, not this grant. Treat any "this patent went to Provenance" claim as unverified.

3. Repeat correspondent across the chain — UNABLE TO ASSESS. The retrieved records expose no correspondent field, so I have no attorney names to test for recurrence. This is the single highest‑value data gap in this file; the signal cannot be marked either way. Retrieve reel/frame plus correspondent for the 2002‑12‑30, 2013‑03‑07, 2014‑01‑21 (×4) and 2014‑10‑09 recordings at Assignment Center.

4. Cascading transfers — WEAK / NOT PRESENT in the NPE sense. There is a same‑day four‑step merger cascade on 2014‑01‑21 (Alcatel Internetworking (PE) → (PE/DE) → Alcatel Internetworking, Inc. → Alcatel USA Marketing → Alcatel‑Lucent USA). That is a textbook internal upstream merger, not an NPE hop: all four entities are Alcatel‑/Alcatel‑Lucent‑named operating subsidiaries and the terminus is the group's principal US operating company, not an LLC. Not a signal.

5. Pre‑litigation transfer — NOT PRESENT. The only suit I could positively tie to this patent is Alcatel‑Lucent USA Inc. v. Fortinet, Inc., D. Del. No. 1:14‑cv‑00574, filed 2014‑05‑01, which pleads US 6,347,087 ('087) alongside US 6,046,979 ('979) and US 6,098,172 ('172), with Alcatel‑Lucent USA pleading it "is the owner of all right, title, and interest in the '087 patent" (complaint text circulated at https://www.scribd.com/document/222219131/Alcatel-Lucent-v-Fortinet). The gap is ~3.5 months from the 2014‑01‑21 consolidation, but the plaintiff had been the beneficial owner of the chain since at least 2002 and is the operating parent itself — not a pre‑suit assignment to a fresh assertion vehicle. A second Delaware case, 1:14‑cv‑00628, is listed against this patent by Unified Patents (https://portal.unifiedpatents.com/litigation/Delaware%20District%20Court/case/1%3A14-cv-00628); I was unable to confirm the parties in that case before running out of search budget — do not assume it is the same plaintiff/defendant pairing.

6. Bankruptcy fire‑sale — NOT PRESENT. No Chapter 7/11 of Packet Engines or Alcatel‑Lucent appears in the record. The 1998 Packet Engines outcome was a $315M cash acquisition of a going concern, and the 2016 Alcatel‑Lucent → Nokia transaction was a going‑concern acquisition, not a bankruptcy sale.

7. Privateering — NOT PRESENT. The 2014 assertion was made by the operating company in its own name against a competitor (Fortinet, a network security appliance vendor). There is no NPE plaintiff asserting on Alcatel‑Lucent's/Nokia's behalf anywhere in the retrieved record for this patent.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT, Unified Patents or OIN. (The 2017 Provenance structure is adjacent to defensive aggregation — McCurdy is a former AST CEO and RPX executive, and the stated model was licensing to operating companies for defence — but the assigned lands, not Provenance, remains the last confirmed owner of this patent.)

Verdict

Operating-company assertion.

Justification: the entire recorded chain runs from one shipping hardware company to the next — Packet Engines Incorporated (reel 009543/0763, executed 1998‑10‑09, assignee of the inventors; PowerRail switch‑router vendor) through the 2002‑12‑30 change of name and the four 2014‑01‑21 mergers into Alcatel‑Lucent USA Inc., with the only non‑operating parties being secured lenders Electro Banque (reel 009576/0621) and Credit Suisse AG (2013 pledge, released 2014‑10‑09). No licensing entity or LLC ever holds title, and the patent was asserted in the owner's own name against a competitor in Alcatel‑Lucent USA Inc. v. Fortinet, Inc., 1:14‑cv‑00574 (D. Del., filed 2014‑05‑01), which pleads the '087 patent directly. The two residual uncertainties worth resolving before anyone relies on "clean operating‑company history" are (a) the missing correspondents/reel‑frames on the 2002–2014 events, and (b) whether this specific grant was among the ~4,000 US assets Nokia moved to Provenance Asset Group in September 2017 under reel 043877 — which I could not confirm either way.

Verify at: https://assignmentcenter.uspto.gov/ (assignee search "Packet Engines" and "Alcatel Internetworking") and the mirrored index at https://assignment.uspto.gov/patent/index.html.

I'll research this systematically. Let me start by searching for the patent and its cited prior art references.

Let me research the remaining cited references.

Prior Art Analysis — US 6347087 B1

Analysis date: April 26, 2026
Patent under review: US 6347087 B1 — "Content-based forwarding/filtering in a network switching device"
Assignee (original): Packet Engines Incorporated | Current (listed): Nokia of America Corp
Application: US 09/166,318 | Filed: 1998-10-05 | Priority: 1998-10-05 | Granted: 2002-02-12
Status: Expired – Lifetime (anticipated expiration 2018-10-05)

0. Search notes and scope

• The authoritative full text used here is the USPTO/Google Patents rendition of US 6347087 B1 (https://patents.google.com/patent/US6347087/en), retrieved 2026-09-27. The USPTO PatentCenter record and assignment history are linked from that page.
• Important dating point: this application was filed 1998-10-05, i.e. pre-AIA. Therefore § 102(a), (b), and (e) govern — not the AIA's § 102(a)(1)/(a)(2). This matters for how the references below qualify as prior art.
• The patent's front page lists 7 patent citations (examiner-cited) and 4 non-patent citations. I address all 7 patent citations below.
• Caveat on completeness: For three of the seven references (US 5,959,976; US 5,999,530; US 6,061,368) I was able to confirm bibliographic data but not retrieve full claim/specification text within this session. For those, the § 102 mapping is explicitly flagged as provisional. A definitive anticipation opinion requires element-by-element full-text comparison, which was not possible for all references here. I have deliberately not fabricated content.
• Being listed as a citation does not mean the reference anticipates. The examiner cited these and still allowed the claims; citation usually indicates relevance/background, not anticipation.

1. US 5,473,607 A — Packet filtering for data networks

Description. A partial packet filter for data networks. A candidate field of a received packet is hashed into a plurality of independent bit-wise subsets, each compared against a previously prepared reference hash table to decide whether to forward or reject the packet. Target fields are configured in a preliminary (software-performed) operation when the network is reconfigured.

Relation to the claims. This is a genuine content-based packet filter (it examines a candidate field, not only a destination address) and it forward-or-rejects. However, its mechanism is hashing + hash-table lookup, not offset-then-compare.

• Potentially relevant to §§ 102 for: claim 1 / claim 16 / claim 10 / claim 20 only if the hashed "candidate field" mapping is read as an "offset value" — a stretch, since US 5,473,607 gives no programmable offset (compare claim 1's requirement that the offset "allows identification of either a header portion or a payload portion").
• Weak/no § 102 relevance to: claims 4–9, 11–15, 17–19, 21–23 (no expectant/anchor/mask, no grouping, no merge).

2. US 5,608,662 A — Packet filter engine ← closest § 102 reference

Description. A state machine / interface processor sits between a data processor and a communication line and compares packets against programmable selection criteria. Key disclosure, quoted from the specification:

"The selection criteria are programmed into the state machine as a decision tree of any length, configuration or data requirements … and can include examination of arbitrary sections of the packet for equality/inequality, greater-than/less-than, signed and unsigned comparisons and bit mask comparisons. Thus content is variably examined, as opposed to checking for an address or key code at a given byte position. The state machine operates on recognition instructions including byte offset and content specifics."

Claims 10–16 recite a "pipeline register" through which "the bytes being shifted … during execution of successive instructions," and "instructions specify at least a relative position of bytes being compared to the selection criteria," with comparisons for equality/inequality, greater/less-than, and bitwise mask comparison.

Relation to the claims. This is the most substantive citation and maps to the core of the independent claims:

• Claim 1 / claim 16 (storing an offset and comparator; comparing frame information at an offset-determined position): US 5,608,662's "recognition instructions including byte offset and content specifics" with a comparand is a strong § 102 candidate for the offset + compare concept.
• Claim 4 / claim 11 (expectant value) and claim 6 / claim 13 (mask value): its equality/inequality outcomes and "bit mask comparisons" are on-point.
• Claim 10 / claim 20 (apparatus): covered in device form.
• Distinguishing features to argue: US 5,608,662 is a filter that accepts/blocks packets toward an attached processor — it is not clearly a multiport switching device forwarding between ports ("communicated to a destination port," claim 1), and does not disclose grouping of filter outcomes (claims 7–8/18–19) or merging with separate forwarding information (claims 9/21–23).
• Potentially anticipates: claims 1, 4, 6, 16, 17, 20 (and dependent claims 10, 12, 13 as apparatus equivalents), subject to the "switching device" limitation.

3. US 5,951,651 A — Packet filter system using BITMAP vector of filter rules…

Description. A hardware-implemented filter in a router/packet-forwarding engine. Off-line, filter rules are partitioned into window arrays per packet parameter (source address, destination address, port, protocol type); each partition is bit-mapped into a bitmap vector. On-line, packet header parameters are compared with window intervals, the corresponding bitmap vectors are logically ANDed (intersected) to find common candidate rules, and the highest-priority common rule is applied to the packet.

Relation to the claims.

• Claim 7 / claim 8 / claim 15 / claim 18 / claim 19 (multiple filters; combining filter outcomes; logically combining outcomes): US 5,951,651's bitmap-vector intersection is a direct § 102 candidate for multi-filter combination by logical operation.
• Claims 21–23 (merging logic that modifies assigned forwarding info; priority modification): its rule-priority selection is relevant, though it selects among filter rules rather than merging a filter outcome with a separately obtained forwarding decision — a distinction worth pressing.
• Claims 1–6, 10–13, 16, 17, 20: US 5,951,651 compares header parameter fields against window ranges; it does not disclose a programmable offset value locating an arbitrary header or payload position, nor an explicit mask-and-compare. Weaker § 102 relevance to the offset/compare claims.

4. US 5,959,976 A — Method and device for filtering transmission

Description (provisional). Title and bibliographic record confirm a method and device for filtering transmissions. I was not able to retrieve the full text in this session.

§ 102 mapping (provisional). Based on the title alone, this reference is likely relevant at a general level to claim 1 (a filtering method) and claim 16, but the title gives no indication of offset-based comparison, mask, anchor, expectant, grouping, or merge. I do not have enough information to assert anticipation of any specific claim, and I explicitly decline to speculate. A full-text review is required.

5. US 5,999,530 A — Method and apparatus for transparent intermediate system based filtering on a LAN of multicast packets

Description (provisional). A transparent intermediate system (bridge/switch) that filters multicast packets on a LAN. The filtering is directed to multicast/address-based decisions at an intermediate system.

§ 102 mapping. This reference is relevant as background prior art for the switching-device context of claim 20 (an intermediate system that forwards or not based on filtering), and arguably to claim 1/16 as a "filtering in a switching device" disclosure. However, it is address/multicast-oriented and gives no apparent teaching of a programmable offset + comparator + mask content filter, of combining filter outcomes into a group, or of merging with a separate forwarding lookup. I was unable to retrieve full text; the mapping above is provisional.

6. US 6,061,368 A — Custom circuitry for adaptive hardware routing engine

Description (provisional). Custom (ASIC) circuitry implementing an adaptive hardware routing engine. Full text was not retrievable in this session.

§ 102 mapping (provisional). Titled/positioned as a hardware routing engine, this is the most likely of the three provisional references to overlap with the hardware-implementation aspects of claims 10–15 and 20–23 (ASIC-based content forwarding/filtering at wire speed). I cannot assert anticipation of any specific claim without the full text. Flag for follow-up retrieval.

7. US 6,147,976 A — Fast network layer packet filter

Description. A packet filter for a screening router that avoids an exhaustive rule search. A source domain identifier, destination domain identifier, and a protocol index (from a protocol vector data structure) jointly index a filtering matrix; the entry contains a forwarding flag that determines whether the received packet is forwarded or dropped. It thereby combines multiple filter dimensions into a single decision and supports filtering on TCP/UDP ports.

Relation to the claims.

• Claim 1 / claim 16 (compare frame information to obtain an outcome; process the frame with the outcome): the filtering matrix returns a forwarding flag that decides forward/drop — relevant as § 102 prior art to the "outcome drives processing" concept, though the mechanism is matrix indexing rather than offset+compare.
• Claim 2 (communicating the frame to destination ports using the outcome): the forward/drop flag is directly on point for the forward-or-not decision.
• Claim 20 (switching device: input, filter comparing identified position with a comparator returning an outcome, output forwarding or not based on the outcome): strong textual overlap with the "forwarding flag determines whether the packet should be forwarded or dropped."
• Weak/no relevance to: claims 3–9, 11–19, 21–23 (no priority-based merge, no grouping, no anchor/mask/expectant architecture).

8. Non-patent literature (also cited on the face of the patent)

Four non-patent references were cited; they are general background rather than § 102 anticipatory art, but two are notable:

• Bransky & Passmore, "Layer 4 Switching," White Paper, Decisys (Sep. 1997) — cited expressly at http://www.decisys.com/Documents/WhitePapers/L4_switching. Directly relevant to the general problem of content/L4-based forwarding, but listed as background.
• "Fast Ethernet," Quinn & Russell, ch. 7, pp. 109–147 (1997).
• "Switched and Fast Ethernet: How It Works and How to Use It," Breyer & Riley, ch. 3, pp. 41–78 (1995).
• "High-Speed Networking with LAN Switches," Held, ch. 2, pp. 33–61 (1997).

Separately, the EPO search report for EP 0 993 162 B1 (the European sibling) considered US 5,790,554 A (Pitcher et al.) and US 5,515,376 A (Murthy et al.) as the closest [X/A] art, plus the Srinivasan et al. "Fast and Scalable Layer Four Switching" paper (SIGCOMM, Oct. 1998). These are not on the US face of the patent but are worth noting for a full invalidity landscape, particularly since the US 09/967,139 continuation issued as US 7,065,082 B2.

9. Summary matrix — citation vs. likely-challenged claims

Key take-aways

1. US 5,608,662 is the only cited reference that expressly teaches byte-offset + content comparison including bit-mask comparison, and is therefore the leading § 102 candidate for the independent offset/compare claims (1, 16) and their apparatus counterparts (10, 20).
2. US 5,951,651 is the leading § 102 candidate for the multi-filter logical combination claims (7, 8, 15, 18, 19).
3. US 6,147,976 is the leading § 102 candidate for the "forward or not based on filter outcome" limitation (claims 2, 20).
4. None of the seven references, on the material I could retrieve, clearly discloses the combination of (a) a programmable anchor + offset + mask + comparator/expectant filter plus (b) per-group Boolean combination with per-filter enable/invert bits plus (c) merge logic that selectively overwrites a separate search-engine forwarding decision (claims 9, 21–23). These appear to be the strongest distinguishing features of US 6347087 B1.
5. US 5,959,976 and US 6,061,368 could not be fully evaluated; they should be retrieved and read before any final § 102 opinion is issued. I explicitly refrain from asserting anticipation for them.

10. Litigation / family context (contextual note)

The patent's record shows US district court litigation in Delaware (cases 1:14-cv-00628 and 1:14-cv-00574), and a continuation US 7,065,082 B2 (from US 09/967,139) plus published application US 2002/0067726 A1. Also of note: the EP sibling EP 0 993 162 B1 was opposed/granted 2005-12-28, with corresponding JP 2000-156713 A and JP 4,999,957 B2. Any invalidity study should treat the continuation's claims separately, since claim scope and prior-art exposure may differ.

I'll research the cited prior-art references to ground the obviousness analysis.

Let me research the remaining cited references and non-patent literature.

Obviousness Analysis — US 6347087 B1

Patent: US 6347087 B1, "Content-based forwarding/filtering in a network switching device"
Inventors: Jayasenan Sundara Ganesh; Timothy S. Michels; James E. Cathey
Original assignee: Packet Engines Incorporated (later Alcatel Internetworking (PE), Inc.; current listing Nokia of America Corp)
Filing/priority date: 1998-10-05 · Granted: 2002-02-12 · Status: Expired – Lifetime (anticipated expiration 2018-10-05)
Statutory framework: Pre-AIA 35 U.S.C. § 103(a), applied with KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) and the Graham factors.

Note on sourcing. The claim charts below rely on the "Patent Citations" (7 references), "Family Cites Families" (5 references) and "Non-Patent Citations" (4 references) listed on the patent page. I verified the disclosures of US 5,473,607; US 5,608,662; US 5,951,651; US 6,141,976; and US 6,061,368 with live lookups. I could not verify the specific disclosures of US 5,999,530; US 5,959,976; US 5,515,376; US 5,790,554; JP H07-200436; or JP 3670372 in this session, and I flag them as secondary/corroborating art rather than primary grounds.

1. The claims at issue

The point of novelty is essentially claim 1's "offset value allowing identification of either a header portion or a payload portion" combined with a hardware filter whose result drives forwarding/filtering decisions — i.e., generalized content-based matching at an arbitrary (header or payload) position, rather than address-field lookup. Claim 16 is the same subject matter without the "filter" label.

2. Prior-art inventory and prior-art status

Level of ordinary skill. A B.S. in EE/CS (or equivalent) with ~2–4 years of network-switch/ASIC design experience, familiar with Ethernet frame formats, lookup-based switching, and hardware packet-classification engines.

Admitted prior art in the specification. The '087 specification itself concedes that (a) switches forward based on destination MAC/IP addresses via lookup tables, (b) forwarding decision logic conventionally "examine[s] the level of priority of the network frame[,] [h]igher priority frames are typically forwarded… before lower priority frames," and (c) prior content-based routing approaches "have been too slow and limited in scope." These admissions narrow the distance between the claims and the art.

3. Grounds of rejection

Ground 1 — Claims 1, 2, 4, 6, 16, 17, 20 obvious over US 5,608,662 (Large et al.) in view of US 5,473,607 (Grand Junction)

Mapping. US 5,608,662 discloses a programmable packet-filter engine in which "the state machine operates on recognition instructions including byte offset and content specifics," and where the selection criteria "can include examination of arbitrary sections of the packet for equality/inequality, greater-than/less-than, signed and unsigned comparisons and bit mask comparisons." Its abstract states that "content is variably examined, as opposed to checking for an address or key code at a given byte position." (https://patents.google.com/patent/[US5608662A](/patent/US5608662A); https://patentimages.storage.googleapis.com/89/7e/6e/cccab66c1bc02f/US5608662.pdf)

• Claim 1 "offset value … determining a position in the frame" → the byte offset of the recognition instructions; "comparator value" → the content specifics; "either a header portion or a payload portion" → the express "arbitrary sections" and the deliberate distinction from fixed header/key-code matching.
• Claim 4 "expectant value" → claim 12–14 comparisons for equality and inequality and greater-than/less-than, i.e., programmable expectants.
• Claim 6 "mask value" → claim 13's "bitwise mask comparison."
• Claim 16/17 → the same subject matter recited generically as offset/comparator.
• Claim 20's "output for forwarding the data packet or not based on the filter outcome" → the engine "passing accepted packets and blocking rejected ones" and coupling accepted packets to the data processor.

US 5,473,607 supplies the network-switching context missing from US 5,608,662: filtering of Ethernet data packets at a controller that "examines packets in real-time," with "filtering criteria … expressed as simple Boolean functions of data fields within the packet," and criteria stored in a target memory. (https://patents.google.com/patent/[US5473607A](/patent/US5473607A))

Motivation to combine. Both references are in the same field (real-time packet filtering in data networks) and address the same problem the '087 patent identifies — filtering/routing on packet content rather than only addresses. US 5,473,607 explicitly calls out the burden on the host processor and the need for "filtering mechanisms that can be implemented with a minimum amount of logic and memory" at the controller; US 5,608,662 offers precisely such a programmable hardware comparator. A PHOSITA would substitute US 5,608,662's engine for (or add it to) the controller/target-memory filtering of US 5,473,607 to obtain programmable, offset-based content filtering in a LAN switch with a predictable aggregation of functions (KSR, 550 U.S. at 417). Reasonable expectation of success: both are hardware comparators against stored criteria; no new physics or untested principle is required.

Ground 2 — Claim 2 and claim 9/14/21 ancestor features obvious over Ground 1 further in view of US 6,141,976 (Cabletron)

Mapping. US 6,141,976 discloses a screening router whose "filtering matrix entry includes a forwarding flag," where "[t]he value of the forwarding flag determines whether the received packet should be forwarded or dropped," and which "permits filtering based on both source and destination TCP/UDP ports." (https://patents.google.com/patent/[US6147976A](/patent/US6147976A))

• Claim 2 — "communicating the frame to destination ports using the outcome to determine which destination ports are to receive the frame" — the filtering-matrix outcome deciding forward/drop and the screening-router topology deciding where the packet goes.
• Claim 20's forwarding-or-not output is met nearly verbatim by the forwarding flag.

Motivation. US 5,473,607's controller only "classifies" (reject / exact / partial) and defers the forwarding decision; US 6,141,976 completes the loop by making the classification result directly control forwarding. Combining them is the routine use of a known technique (outcome-driven forwarding) in a known way to improve a similar device — an enumerated KSR rationale.

Ground 3 — Claims 7 and 8 (and 18, 19) obvious over US 5,608,662 in view of US 5,951,651 (Lakshman et al., Lucent)

Mapping. US 5,608,662 already teaches combining filters: claim 16 recites "means for combining a plurality of independent selection criteria into a single recognition program," and claim 27 recites "combining a plurality of independent recognition programs into a single recognition program." That alone reads on claim 7 (multiple filters → group outcome) and claim 18/19.

US 5,951,651 strengthens the ground by disclosing a hardware-implemented filter for a router in which "all the potential filter rules for each partition are bit mapped into a vector," and the bitmap vectors are intersected with a logical AND operation to produce a resultant bit vector "having bits set that correspond to the intersection of the individual bitmap vectors," the applicable rule being the "rule of greatest priority." (https://patents.google.com/patent/[US5951651A](/patent/US5951651A))

• Claim 7 → plural filters feeding a combined group result (US 5,951,651's vector intersection; US 5,608,662's claim 16).
• Claim 8 → combine by logical operation (the express AND gate 79 / bitwise AND).
• Claims 18–19 → the same, generically recited.

Motivation. US 5,951,651's stated problem is exactly the '087 patent's stated problem: "to maintain a great level of throughput the filtering function must be performed at very high speeds," and it offers an architecture "suitable to be manufactured in integrated circuit form … in a packet forwarding engine or router." Combining plural filter results by Boolean/logical gating to produce a single policy outcome was a known design choice; doing so in an ASIC to hit wire speed is an obvious implementation.

Ground 4 — Claim 3, and claims 9, 14, 21, 22, 23 obvious over Ground 1/3 further in view of US 6,061,368 (Xylan) and the "Layer 4 Switching" white paper (Bransky & Passmore, Sept. 1997)

Mapping.

• Claim 3 (setting priority) and claims 9/14/21–23 (merging filter outcome with forwarding information from another source, and modifying that information, including priority): US 6,061,368 discloses a routing engine whose "routing decisions generally result in inbound packets either being transmitted in unmodified form, transmitted in modified form, or being filtered," with packet parsing that extracts L3/L4 identifier sets and a header cache that generates an outbound packet header. (https://patents.google.com/patent/[US6061368A](/patent/US6061368A))
• The "another source" of claim 9/21 — a lookup/search engine distinct from the content filter — is taught by the combination of US 6,141,976 (matrix/database-driven decision) and US 5,951,651 (rule database lookup), which coexist with content matching in the same forwarding engine.
• The "modification … of a priority assigned to the data packet" (claim 23) is supported by US 5,951,651's priority-ordered rule application plus the "Layer 4 Switching" white paper cited on the face of the patent, which describes classifying traffic by L4 content for QoS/forwarding treatment. The '087 specification additionally admits that priority-based forwarding was conventional.

Motivation. The 1997 Decisys white paper ("Layer 4 Switching," cited in the '087 patent's own Non-Patent Citations as published at http://www.decisys.com/Documents/WhitePapers/L4_switching) evidences the then-current market push to switch on content beyond L2/L3 addresses — the same "need or problem known in the field of endeavor at the time of invention" that KSR accepts as a reason to combine. Once a content filter produces an outcome, using that outcome to override or modify a port/priority decision produced by a conventional address lookup is no more than the predictable combination of two known decision sources, and US 6,061,368 confirms that hardware routing engines routinely modified headers before transmission.

Caveat. Claim 9/14/21 distinguish the filter from "another source," and the specific merge-logic architecture of claims 21–23 (a merging logic that modifies rather than merely overrides) is the weakest link in this record. US 6,061,368 teaches modification and US 5,951,651 teaches priority, but a strict reading may require an additional reference expressly showing a merge of two independent decision pipelines. The patent's own continuation, US 7,065,082 B2 (filed 2001-09-27), and the family EP 0 993 162 B1 / JP 4999957 B2 may contain clarifying claim language worth checking for that limitation.

Ground 5 — Claims 5 (anchor value) and the "header vs. payload" limitation, over Grounds 1–4 in view of US 6,061,368

Mapping. The '087 specification defines the anchor as, e.g., "Beginning of a frame; Beginning of a Layer-2/Layer-3 or any other layer header; End of a Layer-2/Layer-3 … header; A fixed byte-offset from the beginning of the frame." US 6,061,368's "Gateway parsing unit … parses inbound packets based on protocol and extracts identifier sets," including "Layer 3 (Network) and/or Layer 4 (Transport) identifiers," establishes the packet-structure reference points that make an offset measured from a header/payload anchor routine. US 5,608,662's "relative position" instruction field supplies the relative-offset concept.

Motivation. Choosing whether to measure an offset from the frame start versus the L3 payload start is a straightforward design choice dictated by variable-length L2/L3 headers; the '087 patent itself treats anchors, word lengths, filter counts and group counts as "application choices and not limitations of the invention."

4. Overall assessment of nonobviousness / rebuttal considerations

1. No teaching away. None of the references disparages content-based matching at arbitrary offsets; US 5,608,662 and US 5,951,651 actively pursue it, and US 5,473,607's target-memory/target-field architecture invites generalization.
2. Predictable results. Each claimed element maps to a discrete, previously known mechanism (byte-offset comparator + mask + expectant; logical combination of filter results; forwarding flag; header modification). The combination is the "predictable use of prior art elements according to their established functions" (KSR, 550 U.S. at 417).
3. Weakest limitations for the patentee. (a) the "merge with another source" / merging-logic limitations of claims 9, 14, 21–23; (b) the "setting/modifying priority" limitations of claims 3 and 23; and (c) the express anchor value of claim 5. A secondary-considerations case would presumably rest on wire-speed ASIC implementation, but the specification attributes that goal to the problem being solved, not to an unexpected result, and US 6,061,368 and US 5,951,651 both already target high-speed custom/semiconductor implementation.
4. Litigation context. The page records two Delaware District Court cases (1:14-cv-00628 and 1:14-cv-00574) and a Darts-IP first-worldwide-family litigation entry for family ID 22602752. Those records do not change the prior-art analysis but indicate the claims have been tested; any invalidity positions actually litigated should be checked against the record rather than assumed.

Bottom line. Claims 1, 2, 4, 6, 7, 8, 10–20, and the generic method/apparatus claims 16–19 are, on this record, strongly vulnerable under § 103 as disclosed by US 5,608,662 combined with US 5,473,607, and further combined with US 5,951,651 (logical combination of filter results) and US 6,141,976 (outcome-controlled forwarding). Claims 3, 9, 14, 21, 22, 23 are also vulnerable via US 6,061,368 and the Layer 4 Switching white paper, but the "merge with another source" and merge-modifies-priority limitations are the most defensible and would benefit from a primary reference showing two independent forwarding-decision pipelines whose results are merged.