US 6795399

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

Let me search for the patent's claims and any litigation involving it specifically.

Let me try to retrieve the actual claim text and check for any litigation database entries.

US Patent 6,795,399 B1 — Analyst Summary

Bibliographic Data (from the patent record; interpreted literally)

Abstract (verbatim)

"Methods and apparatus are provided for designing IP networks with substantially improved performance as compared to existing IP networks such as, for example, those networks designed under best-effort criteria. Particularly, the invention includes methods and apparatus for: computing worst-case and optimistic link capacity requirements; optimizing network topology; and determining router placement within a network."

Field / Technical Subject Matter

Methods and apparatus (software-implemented "design system," FIG. 1, reference 10) for capacity planning and topology design of IP networks that must support QoS/bandwidth guarantees — notably VPN bandwidth guarantees. The system models the performance of packet scheduling and buffer-management schemes — FIFO combined with RED ("FIFO/RED," legacy routers) versus weighted fair queuing combined with longest-queue drop ("WFQ/LQD," e.g., Lucent's PacketStar IP Switch) — and computes the link capacity each scheme requires to satisfy a given set of point-to-point IP demands.

Overview of the Independent Claims

Important caveat on uncertainty: The authoritative full text made available to me (the Google Patents rendering) includes the abstract, definitions, and detailed description but is truncated before the numbered claims, and my web searches returned description content rather than a verbatim claim set. I therefore cannot quote or precisely paraphrase the independent claims with confidence. The following is an inference from the specification's three stated "aspects," not verified claim language:

1. Link-capacity computation (first aspect; FIGS. 3–5). The specification's primary claimed subject matter appears to be a method (and corresponding apparatus/system) for computing required link capacity on a per-link basis given a network topology, a set of IP flow demands, and network delays. Distinct embodiments cover (a) FIFO/RED worst-case capacity using congestion scenarios designated H_worst, H_hop, H_one (e.g., eqs. 15, 17, 19, computed by processor 14); (b) FIFO/RED optimistic capacity using H_best and network-wide multiple-bottleneck effects (eqs. 16, 18, 20, computed by processor 16); and (c) WFQ/LQD capacity as the simple sum of per-link demands (eq. 14). A notable stated feature is that the designer need not know where bottlenecks are located; the user specifies only a congestion option H_O.

2. Network topology optimization (second aspect; FIGS. 6A–6D, 7). Iterative augmentation (classifying demands as "keepers" vs. "stragglers," with capacity-only or capacity-plus-connectivity augmentation, threshold-controlled or direct straggler-to-keeper conversion) and iterative deloading (removing lightly loaded links via a utilization threshold and re-computing routes/capacities/cost).

3. Router placement / heterogeneous networks (third aspect; FIG. 8). Determining placement of WFQ/LQD routers to replace FIFO/RED routers so as to maximize network-cost savings, formulated as a mixed-integer programming (MIP) problem.

Whether all three aspects are reflected in independent (as opposed to dependent) claims — and the exact number of independent claims — cannot be confirmed from the materials I retrieved. This should be verified against the USPTO PatentCenter full-text or the issued patent's claims column.

Litigation / Docket Search — Result: None Found

I searched for US 6,795,399 in connection with CAFC 2026 dockets and district-court/PTAB activity:

• No CAFC 2026 docket, appeal, or Federal Circuit decision referencing this patent was found.
• No district-court infringement action, ITC investigation, or PTAB/IPR proceeding referencing this patent was found in my search results.
• The record shows the patent expired (fee-related) effective 2021-02-06, which is consistent with it not being an active litigation vehicle in 2026.

I cannot rule out a filing that is not surfaced by general web search (e.g., a sealed or very recent docket), but I found no affirmative evidence of any 2026 CAFC activity for this number.

Explicit Uncertainty / Data-Quality Notes

• Claim text not verified. As noted above, I could not retrieve the verbatim independent claims; the claim overview is inferred from the specification and should not be relied upon as a quotation.
• Distinct ID caution. A search surfaced "JP 6795399 B2" (a Japanese patent, publication date 2020-12-02, appearing in a chemical/biotech family listing). Per the rule to interpret identifiers literally, this is a different patent document and is not US 6,795,399 B1. Do not conflate them.
• Related patent, flagged as unverified. A citation list shows US 6,909,700 (Benmohamed et al., June 21, 2005), which may correspond to the concurrently filed topology-optimization application (Ser. No. 09/198,728). I have not verified this linkage and state it only as a lead.
• Assignee note. The record lists the original assignee as Lucent Technologies Inc. and the current assignee as Nokia of America Corp (successor-in-interest via the Alcatel-Lucent/Nokia chain). The database itself cautions this is not a legal conclusion.

If you need the exact independent-claim language, the authoritative source is the issued patent's claims column via USPTO PatentCenter (or the PDF at patentimages.googleapis.com / USPTO full-text), which I was unable to fully extract here.

I'll search for litigation involving US patent 6795399 across patent litigation sources.

Let me search more targeted litigation databases and the specific patent.

Let me run additional targeted searches to confirm whether any litigation exists.

Litigation Search Results — US Patent 6,795,399

Summary Finding

No litigation involving U.S. Patent No. 6,795,399 was found in the sources searched.

What I searched

I ran multiple queries against patent-litigation-oriented sources (Google Patents, PACER-derived dockets surfaced via search, Unified Patents materials, and general web indices covering CAFC/ITC/district court activity), including literal searches on the identifiers 6795399 and 6,795,399 in combination with terms such as "litigation," "lawsuit," "infringement," "complaint," "defendant," and the patent's title and inventors.

Key observations from the results:

• Google Patents (authoritative page for this patent): The record shows no litigation section and no "asserted in litigation" data. Its status is "Expired – Fee Related," with an adjusted expiration of 2021-02-06. Current assignee is listed as Nokia of America Corp; original assignee Lucent Technologies Inc. Inventors: Lotfi Benmohamed, Subrahmanyam Dravida, Paramasiviah Harshavardhana, Wing Cheong Lau, Ajay Kumar Mittal. (Google Patents US6795399)
• No district court, CAFC, ITC, or PTAB case naming US 6,795,399 as an asserted or challenged patent appeared.
• Several results referencing "6795399"-like numbers were unrelated (e.g., a JP 6795399 B2 Japanese patent, a design patent referred to as the "'539 patent" in a coaxial-connector ITC matter, and misc. patent-family listings). Per the strict rule, I did not treat these as the same patent.

Answer to your request

If no litigation is known, say so: Based on the searches performed, no litigation is known involving US Patent 6,795,399.

Important caveats (so you can rely on this appropriately)

1. Negative-result limitation: I cannot prove a universal negative. A litigation search is only as good as the database coverage. I could not directly query paywalled dockets (e.g., Docket Navigator, Lex Machina) or fully traverse the PACER/Unified Patents portals within this session; my tool calls were capped before I could complete a direct portal query.
2. Plausibility of no litigation: This outcome is consistent with the patent's profile — it is a network-design/optimization method patent (link-capacity computation for IP networks) rather than a consumer-facing or standards-essential patent, it is expired, and it was held by Lucent → Nokia, entities not known for PAE-style assertion of this family.
3. Related patents exist: The specification notes two concurrently filed siblings — Ser. No. 09/198,728 ("Network Topology Optimization…") and Ser. No. 09/198,729 ("Router Placement…," issued as U.S. Pat. No. 6,240,463). Those are different patents; any litigation on them would not be litigation on 6,795,399. I did not find or attribute any cases to them either.

If you need a definitive clearance, I recommend running this specific number through Docket Navigator or Lex Machina, and directly querying the Unified Patents Portal litigation caselist (https://portal.unifiedpatents.com/litigation/caselist) and PACER — sources I flagged but could not fully traverse here. I can attempt those specific portal queries if you'd like me to continue.

I'll start by verifying the structured data against web sources, since the task explicitly asks me to flag anything the ODP ingest may have missed.

Let me run additional targeted searches to be thorough about any proceedings the ODP might not have indexed.

Let me check for any litigation or PTAB references via a couple more angles, including the related patents.

Proceedings overview

Total AIA trial proceedings on file for US 6,795,399: 0. The USPTO Open Data Portal (ODP) structured record returns no AIA trial proceedings, and my web searches surfaced no IPR, PGR, or CBM petition, no Final Written Decision, and no Federal Circuit appeal involving this patent. Status breakdown is therefore: active 0 / claims invalidated 0 / claims sustained 0 / settled 0 / institution denied 0.

Bottom line for a defendant: this is not a patent that has been hardened through PTAB — it is a patent that was never tested at the PTAB at all, and, more importantly, it is expired. The patent's recorded legal status is "Expired - Fee Related," with an adjusted expiration date of 2021-02-06. An expired patent cannot be infringed going forward (only past damages within the limitations period remain liveable, and § 286 caps that at six years). So the practical defensive posture is not "the claims are dead through IPR" (they weren't invalidated) but "the patent term is over — check the date of any demand letter before doing anything else."

Proceedings

No proceedings to list. I found no PTAB matter (IPR/PGR/CBM) and no CAFC appeal to report. Per the constraints of this task, I will not invent proceeding numbers: the absence of data is the finding.

If you are working from a demand letter or an internal docket that asserts a specific IPR number for this patent, that number would be the thing to verify — it does not appear in the ODP record or in any public source I could reach, which is a strong signal the number (or the patent it maps to) is being mis-cited.

Strategic summary

Claim status — CANCELED vs. SUSTAINED vs. UNTESTED. Because no PTAB proceeding ever issued, no claim of 6,795,399 has been canceled, confirmed, or construed by the Board. Every claim — the independent claims and their dependents as issued on 2004-09-21 — is UNTESTED at the PTAB. There is no FWD to quote and no claim-level disposition to rely on. Any statement that "claim X is dead" would be false here; there is no such record.

The operative fact is instead the term. The patent was filed 1998-11-24 (priority 1998-11-24) and carries a recorded adjusted expiration of 2021-02-06. It is listed as Expired - Fee Related. That is a far stronger defense than an IPR would have been: there is no ongoing infringement exposure, and past damages are time-barred outside the § 286 six-year lookback window measured from the filing of any suit. Confirm the expiration computation independently (terminal disclaimer, any PTA) rather than relying solely on the Google Patents status field.

Estoppel landscape. § 315(e)(2) estoppel is a non-issue: estoppel only attaches to a petitioner that obtained an institution decision, and there has been no petitioner. No party is estopped from anything, and equally, no prior-art ground has been "used up." The full universe of § 102/§ 103 art against the 1998 priority date remains available to a defendant in litigation. Practically, however, the expiration means an invalidity case is likely moot — the cheaper and cleaner motion is non-infringement-by-expiration rather than an IPR or a § 282 challenge.

Pattern signals. None. There is no repeat petitioner (there is no petitioner), no PTAB-appeal history, and no sign of a defensive aggregator such as Unified Patents taking an interest in this patent. That is consistent with the patent's age and expiration: well-asserted, still-live patents attract IPRs; a patent that lapsed in 2021 for failure to pay maintenance fees would not.

One adjacent point worth noting for context, drawn from the patent's own cross-references: 6,795,399 is one of three concurrently filed applications by the same Lucent inventors. Its siblings are Ser. No. 09/198,728 ("Network Topology Optimization Methods and Apparatus...") and Ser. No. 09/198,729 ("Router Placement Methods and Apparatus..."), the latter of which issued as U.S. Pat. No. 6,240,463. If a defendant is looking at this family, the sibling patents should be checked for their own status — but I did not confirm any PTAB or litigation activity on them within the available search steps, so do not assume it.

Recommended next steps

1. Confirm expiration before anything else. The single most valuable action is verifying that 6,795,399's term ended 2021-02-06 (check the maintenance-fee record in Patent Center / ODP and the PTA/terminal-disclaimer history). If confirmed, any forward-looking infringement demand citing this patent is unsupportable.
2. If a demand letter names an IPR number for this patent, treat it as unverified. No such proceeding exists in the ODP record or in public sources. Ask the sender to identify the proceeding; a real IPR citation should be resolvable at the USPTO PTAB E2E system (https://e2e.uspto.gov) and on the PTAB Decisions page.
3. Past-damages window only. If the patent owner is pursuing back royalties, the analysis is a § 286 six-year lookback from the complaint date, not an invalidity fight. By 2026, even that window has narrowed considerably relative to the 2021 expiration.
4. Sibling-patent check. Pull the prosecution and maintenance status of U.S. Pat. No. 6,240,463 (the Ser. No. 09/198,729 sibling) and Ser. No. 09/198,728, since a portfolio assertion may be built on those instead. I have not verified their current status or any PTAB activity on them.
5. No PTAB-based defense to preserve. Because the claims were never instituted on, there is no § 315(b) one-year bar clock running and no § 325(d) estoppel consideration to weigh. If, contrary to the record above, this patent is still being asserted and is somehow still in force, the full prior-art landscape remains open — but that is a contingency, not the expected baseline.

Caveat on sourcing: This conclusion rests on the ODP structured record (no proceedings) plus web searches that returned no relevant hits. The searches were limited to a small number of queries and were truncated by tool limits; the negative result is consistent across every angle I could check, but it is not the same as a definitive PTAB docket pull. If claim-level certainty matters, the authoritative check is a direct query at USPTO PTAB E2E and the Federal Circuit docket (or CourtListener) for any appeal referencing this patent number.

I'll research the assignment record for US 6795399 before reconstructing the chain.

Let me dig deeper into the specific assignment records and whether this patent moved to WSOU or remained with Nokia.

Ownership & Assignment Analysis — US 6795399

Before the findings: the tooling available to me surfaced the Google Patents legal‑events record and third‑party litigation/PTAB documents, but not the individual USPTO Assignment Center (Patent Assignment Search) abstract pages for this patent. Where a reel/frame is not visible in the retrieved sources, I say so rather than inventing one. The patent’s own front‑matter (authoritative text provided) supplies the inventor set, original assignee, and the single recorded assignment event.

Inventors

All five are named on the 1999‑02‑10 assignment of interest to Lucent Technologies, Inc., confirming they were Lucent employees/inventors at filing (application filed 1998‑11‑24, granted 2004‑09‑21 as US 6795399 B1).

Pattern note: No departure signal is determinable. I found no verifiable evidence of any inventor leaving Lucent within 12 months of the 1998‑11‑24 filing; the sources I could reach do not cover inventor personnel movements, so this must be scored unclear, not "present." The patent is a Bell Labs research output tied to Lucent’s PacketStar IP Switch product line (the specification expressly describes WFQ/LQD scheduling on PacketStar), which is consistent with the inventors being incumbent Lucent staff rather than a departing team.

Original assignee

Lucent Technologies, Inc. (Delaware; principal place of business at 600 Mountain Avenue, Murray Hill, NJ 07974 — per Lucent’s own litigation filings).

• Line of business: Telecommunications/network equipment (spun out of AT&T in 1996; Bell Labs R&D arm). The patent’s design methodology explicitly targets Lucent’s PacketStar™ IP Switch router product, i.e., the assignee commercialized hardware embodying the WFQ/LQD scheduling concepts the claims/embodiments address.
• Product embodying the technology: Yes — PacketStar IP Switch (Lucent), cited in the specification as supporting per‑flow WFQ and longest‑queue drop.
• Current status: No longer an independent operating entity. Lucent became a wholly owned (now indirect) subsidiary of Alcatel SA on 2006‑11‑30, when Alcatel renamed itself Alcatel‑Lucent; Alcatel‑Lucent was then acquired by Nokia (completed 2016), and the U.S. operating entity carries the name Nokia of America Corporation. Google Patents lists Nokia of America Corp as the current assignee. The registrant is not dissolved or in bankruptcy; its assets sit inside Nokia. Caveat: I could not retrieve a fee‑related or terminal‑disclaimer document explaining the 2021‑02‑06 adjusted expiration; the patent is recorded as Expired – Fee Related.

Assignment timeline

Only one post‑filing assignment is visible in the records I could reach (Google Patents legal events). The name changes from Lucent → Alcatel‑Lucent → Nokia of America appear in corporate history and on Google Patents’ "current assignee" field, but I found no reel/frame for those — they may be recorded by change‑of‑name/merger entries I could not open. I do not fabricate them.

• 1998‑11‑24 (executed, presumed ~filing date) / recorded 1999‑02‑10 — Reel not surfaced in retrievable sources (Google Patents shows the event without reel/frame; Assignment Center abstract page not accessible to my tools)

  • Conveyance: Assignment ("ASSIGNMENT OF ASSIGNORS INTEREST")
  • Assignor: Dravida, Subrahmanyam; Benmohamed, Lotfi; Harshavardhana, Paramasiviah; Lau, Wing Cheong; Mittal, Ajay Kumar (all five inventors)
  • Assignee: Lucent Technologies, Inc.
  • Correspondent: not available from the retrieved record — cannot be scored
  • Context: Standard inventor‑to‑employer assignment at time of filing (not an acquisition, fire‑sale, or NPE transfer).

• Post‑issuance (date/execution not verifiable here) — No confirmed second assignment.

  • Google Patents lists Nokia of America Corp as current assignee, implying ownership stayed inside the Lucent → Alcatel‑Lucent → Nokia corporate family through successive mergers/reorganizations (2006 Alcatel–Lucent merger; 2016 Nokia acquisition). These are internal reorg / change‑of‑control events, not arms‑length sales, but I have no reel/frame to cite and therefore mark them unverified.

Important negative finding on the WSOU/Provenance channel: Alcatel‑Lucent, Nokia Solutions and Networks BV, and Nokia Technologies Oy did sell a large patent portfolio in 2017 to Provenance Asset Group / Wade and Company → WSOU Investments, LLC (USPTO Reel 044000/0053 and follow‑on Reel 052372/0675, /0540; WSOU Investments is a well‑known high‑volume litigant, e.g. WSOU Investments, LLC v. Cisco Systems, Inc., W.D. Tex. 6:21‑cv‑00128). However, I found no evidence that US 6795399 was among the transferred assets — Google Patents’ current‑assignee field still shows Nokia of America Corp, not WSOU. I therefore do not place this patent in the WSOU chain. Treat that as an unconfirmed lead, not a finding.

If the Assignment Center truly has no further records beyond the 1999‑02‑10 entry for this patent number, that itself is the finding: the patent never left the original assignee’s corporate family.

Timeline diagram

timeline
    title Ownership of US 6795399
    1998 : Inventors assign to Lucent Technologies
    1999 : Assignment recorded Feb 10 1999
    2004 : Patent issued Sep 21 2004
    2006 : Lucent folded into Alcatel-Lucent
    2016 : Alcatel-Lucent acquired by Nokia
    2021 : Patent expired Feb 6 2021

(The 2006 and 2016 rows reflect corporate succession of the assignee, not separate recorded patent assignments; the 2021 row reflects the recorded Expired – Fee Related status.)

NPE / troll‑pattern signals

1. Shell‑entity transfer — not present. The only recorded link is inventors → Lucent Technologies, Inc., an operating telecom‑equipment manufacturer. No licensing‑only LLC appears anywhere in the confirmed chain. No "IP/Holdings/Ventures" assignee is evidenced for this patent number.

2. Known asserter in the chain — not present (for this patent). No Acacia, Marathon, IV, Wi‑LAN, Conversant/Mosaid, Vringo, Pendrell, Round Rock, MPHJ, etc. appears as assignee here. Contextual flag: the Alcatel‑Lucent/Nokia family did route a portfolio to WSOU Investments, LLC (a high‑frequency plaintiff surfaced in RPX/Unified‑type litigation trackers) in 2017 (Reel 044000/0053; Reel 052372/0675 & /0540), but I could not tie US 6795399 to that transaction. Do not score it as present without the reel/frame showing this patent in the assigned‑patents exhibit.

3. Repeat correspondent across the chain — not present / unclear. I was unable to retrieve the correspondent‑of‑record for the 1999‑02‑10 entry, and there is only one confirmed link, so recurrence cannot be established. Not scorable.

4. Cascading transfers (<24 months through chained LLCs) — not present. Only one assignment is recorded; no chain of LLCs exists for this patent.

5. Pre‑litigation transfer (assignment within 6 months of a first suit) — not present. No infringement suit naming US 6795399 was surfaced, and the patent is recorded Expired – Fee Related (adjusted expiration 2021‑02‑06). No assertion activity is evidenced.

6. Bankruptcy fire‑sale — not present. No Chapter 7/11 proceeding involving the assignee is evidenced for this asset; Lucent’s patents moved via merger (Alcatel, then Nokia), not bankruptcy.

7. Privateering — unclear / not evidenced for this patent. The broader Lucent/Alcatel‑Lucent patent‑monetization pattern (e.g., Lucent’s 2006 assignment of a patent trust to MPT, and the later Alcatel‑Lucent→WSOU sale) exists as corporate context, but there is no evidence that US 6795399 was funneled to an NPE to assert against competitors. Score unclear, no citation supports "present."

8. Defensive aggregator (anti‑NPE) — not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN.

Verdict

Insufficient data.

The only recorded assignment I can confirm is the 1998‑11‑24/1999‑02‑10 inventor‑to‑Lucent Technologies, Inc. assignment (Google Patents legal events); the current assignee field shows Nokia of America Corp, i.e., the asset stayed inside the Lucent → Alcatel‑Lucent → Nokia operating‑company family, and the patent is recorded Expired – Fee Related (adjusted expiration 2021‑02‑06). No shell‑entity, known‑asserter, cascade, pre‑litigation, or privateering link is evidenced for US 6795399 — the WSOU Investments/Provenance sale is a plausible contextual lead but is unconfirmed for this number, and I could not retrieve a reel/frame for the Nokia‑family name change or the original assignment. With only the original assignment verified on the record, the safe call is Insufficient data (no NPE pattern), pending the reel/frame entries from Assignment Center.

Verify at: USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search patent number 6795399); mirrored index at https://assignment.uspto.gov/patent/index.html. Google Patents legal events for this number: https://patents.google.com/patent/US6795399/en.

Confidence caveat: I could not open the Assignment Search abstract page for US 6795399 with the tools available, so reel/frame numbers and correspondent‑of‑record are not verified. If Assignment Center returns no entries beyond the 1999‑02‑10 filing‑time assignment, that confirms the "stayed with original assignee family" reading above.

I'll search for the USPTO record and prior-art citations for US 6,795,399.

Let me pull the front-page citations directly from the patent record.

Let me try the USPTO-derived databases and full-text patent sites for the cited references.

Prior Art Analysis — U.S. Patent No. 6,795,399

1. Patent identification (USPTO record)

Source: Google Patents US6795399

Critical cutoff for §102 analysis: any reference must qualify as prior art as of 1998-11-24 (or the applicable invention date). Anything published after that date cannot anticipate.

2. What "each patent citation" comprises — and an important limitation

Under 35 U.S.C. §102/§103 analysis, the citation set for a U.S. patent is drawn from three distinct buckets, and they carry very different legal weight:

⚠️ Honest limitation — read this before relying on the tables below

I was unable to retrieve the front-page (56) U.S. Patent Documents list for 6,795,399. My tool budget was exhausted before I could pull the actual "References Cited" block from the patent image/PDF (U.S. format (56), or the EPO/Espacenet equivalent). The Google Patents text supplied to me contains the abstract, description text, classification, and machine-generated "prior art keywords" — but not the numbered U.S. patent citations.

I will not fabricate that list. Per the operating rules, where I do not know something with high confidence I say so explicitly. So for bucket (A) the correct answer here is: not retrieved — verify directly (see §6 for the exact retrieval path).

What I can document with high confidence is bucket (B), because those references are cited within the body of the specification text itself (the authoritative source provided).

3. Category B — Non-patent literature cited in the specification (verifiable)

These appear in the specification as inline citations and constitute the "Other Publications" portion of the (56) references. All are dated before 1998-11-24 and therefore qualify as potential prior art.

Flagged discrepancies / notes (per instructions):

• B5 date discrepancy: The specification gives the Kumar/Lakshman/Stiliadis reference as "(March 1998)" in the §1.2 opening paragraph and "(May 1998)" later in the same section. Both pre-date the 1998-11-24 filing, so the inconsistency does not change its prior-art status, but it should be resolved against the printed source if relied upon.
• Several section cross-references inside the specification are internally inconsistent (e.g., equation numbering, and "Section 5.0" invoked where §4.0 is the case-study section). These are drafting artifacts, not prior-art issues.

4. Category — Related co-pending applications (cross-reference section)

The specification's CROSS REFERENCE section names two concurrently filed applications:

• Ser. No. 09/198,728 — "Network Topology Optimization Methods and Apparatus For Designing IP Networks With Performance Guarantees"
• Ser. No. 09/198,729 — "Router Placement Methods and Apparatus For Designing IP Networks With Performance Guarantees," issued as U.S. Pat. No. 6,240,463

§102 status: These share the same 1998-11-24 filing date and are not "by others before the invention," so they are not anticipating prior art to 6,795,399. They are relevant only as members of the same patent family (subject-matter overlap / potential obviousness-type double considerations), not as §102 references. I found no litigation attributing any case to them either.

5. Category C — Forward citations (NOT prior art)

Searches surfaced later patents that cite 6,795,399. These post-date the patent and therefore cannot anticipate it; they are useful only to show the patent's influence (or as potential obviousness combinations in later patents — the reverse direction).

⚠️ Do not treat these as prior art. They are evidence of forward citation only.

6. Provisional §102 anticipation analysis (with explicit caveat)

Caveat that governs this section: I was not able to retrieve the verbatim claim text of 6,795,399 (the provided full text includes the abstract and description but not the numbered claims), nor the front-page (56) U.S. patent citations. Claim numbers quoted below are therefore not asserted — the mapping is by claim subject-matter group, derived from the disclosed embodiments and the summary of the invention. This is a provisional, examiner-style mapping, not a final anticipation opinion.

Net conclusion on §102: Based on the references I was able to verify, no single cited reference anticipates the full scope of the patent, because the patent's distinguishing subject matter — the worst-case/optimistic link-capacity bounding framework across congestion scenarios and the multi-bottleneck share computation — is not disclosed by B1–B12. The references B1, B2, B5, and B6 are the most substantive and would be the citations an examiner would build a §103 rejection around, not a §102 rejection.

7. What still must be retrieved for a complete answer

To complete the task as posed you need the numbered (56) citations, which I could not fetch. Retrieve them here:

1. USPTO Patent Public Search / PatentCenter — full document for 6,795,399, front page "(56) References Cited," both "U.S. Patent Documents" and "Other Publications." (PatentCenter)
2. USPTO Patent Full-Text (PatFT/PatentsView) for the OCR'd front page.
3. Espacenet — for the equivalent "Cited documents" list and the search report (Espacenet).
4. Global Dossier — for the file-wrapper citations, including any examiner-cited art not printed on the face of the patent.

Once that list is in hand, run each U.S. patent document against the 1998-11-24 cutoff: only those with a publication or effective filing date before 1998-11-24 can support §102(a)/(b); U.S. patents/applications qualifying under §102(e) must have been filed before that date.

Bottom line: I identified and characterized the verifiable non-patent citations (bucket B) and the forward citations (bucket C), but I could not retrieve the front-page U.S. patent citations (bucket A) for 6,795,399, and I will not invent them. Based on the art I could verify, none anticipates the patent outright.

I'll pull the prior-art/citation data for this specific patent number before doing the § 103 analysis.

Let me try to retrieve the front-page "References Cited" list and the actual claims for US 6,795,399 itself.

Obviousness Analysis — U.S. Patent 6,795,399 B1 under 35 U.S.C. § 103

0. Threshold caveats (please read before relying on this)

Two data problems materially constrain this analysis, and I am stating them up front rather than papering over them:

1. There is no "Prior Art" section in the page text supplied to me. The Google Patents rendering attached to this task contains classifications, definitions, the description, and the abstract — but no "References Cited" / "Cited By" tables. So I could not literally "use the Prior Art section of this page." Instead I have reconstructed the prior art from two sources: (a) the references cited inside the specification itself (which are reproduced in the authoritative text), and (b) the "References Cited" list of the sibling patent US 6,240,463, which is the closest available proxy. ⚠️ These are not necessarily the references on US 6,795,399's own front page. (uspto.report/patent/grant/6240463; freepatentsonline.com/6240463.html)

2. The claim text of US 6,795,399 is still not verified. This is the same gap flagged in the previously generated summary, and it persists. My earlier searches returned the description and the claims of the sibling patents (US 6,240,463 and US 6,909,700) at patents.justia.com/patent/6795399, but not the numbered claims of '399. A § 103 analysis without claim language is necessarily an analysis of the likely claim scope, not of the claims as issued. Everything below should be re-run against the actual claims column before being used.

Governing law note: application 09/198,727 was filed 1998-11-24, so pre-AIA § 102/§ 103 apply. This matters, because it makes the one-year § 102(b) critical date 1997-11-24.

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

A practitioner with an M.S. or Ph.D. in electrical engineering or computer science and roughly 2–5 years of experience in (i) IP/OSPF network design and capacity planning, (ii) analytical modeling of TCP congestion-avoidance throughput, and (iii) router architecture (packet scheduling and buffer management). This is a modeling-and-optimization art, which is important: in such arts the Federal Circuit and the Board routinely treat the application of known mathematical models to a newbut analogous sizing problem as an obvious use of a known technique.

2. The available prior art and its § 102 status

Three critical consequences:

• Kumar (May 1998) and Suter (Mar. 1998) are not § 102(b) art. They postdate the 1997-11-24 critical date. They are available only under § 102(a), i.e., only if they predate the applicants' actual date of invention — and they are Bell Labs/Lucent-adjacent publications by different inventors, so "by others" is satisfied but the date burden remains. An examiner must either establish an invention date after March/May 1998 or secure an affidavit; otherwise these two — the best references for the WFQ/LQD "sum of demands" result — fall out.
• I cannot verify the technical content of US 5,822,301, 5,884,046, 6,069,895, or 6,147,971. I am not going to attribute specific disclosures to them. Treat them as leads only.
• The inventors' own Bell Labs Technical Journal paper ("Designing IP Networks with Performance Guarantees," Oct.–Dec. 1998, pp. 273–296) is cited on the sibling patent's face but is contemporaneous with the filing and by the same inventors. It should not be treated as prior art against '399 on this record — a point frequently mishandled when this family is analyzed.

3. Inferred claim elements

Building on the earlier summary (which correctly inferred the three "aspects") and on the verified claim language of the sibling US 6,240,463, the likely claim structure of '399 is:

• E1 — receive a network topology G=(V,E), a set of point-to-point IP flow demands f_i=(s_i,t_i,a_i,n_i,d_i,r̂_i), and delay τ_i.
• E2 — route demands (shortest path / OSPF) and determine the traffic set F_l per link.
• E3 — assign connection weights w_i = d_i/(τ_i·√h_i).
• E4 — compute per-link FIFO/RED capacity as a maximum over demands of d_i·(Σ_j w_j / w_i), i.e., the "worst-off demand determines the capacity" rearrangement.
• E5 — select a congestion scenario H_O from a stored set (H_worst, H_hop, H_one) as upper bounds and H_best / multiple-bottleneck as a lower bound, without knowing where the bottlenecks physically are.
• E6 — add the UDP demand term and apply a ceiling to trunk granularity.
• E7 — output/display the capacity vector, routes, and network cost (processor + memory; article of manufacture).
• E8/E9 (if claimed) — WFQ/LQD capacity = Σ d_i; topology augmentation/deloading; and/or the MIP router-placement formulation.

4. Combination 1 — FIFO/RED link-capacity computation (E1–E6)

Primary references: Floyd (1991) + Mathis et al. (1997), optionally + Lakshman & Madhow (1994).

What each teaches. Mathis et al. supply a validated closed-form expression for the steady-state throughput of a TCP connection operating in congestion avoidance as a function of round-trip time and loss probability — the identical model the patent reproduces in its § 5.0 and FIG. 11 sawtooth derivation. Floyd (1991) supplies the multi-gateway extension: when a connection traverses h congested hops, the effective path loss probability compounds, yielding the per-connection share proportional to 1/(τ·√h). The patent's equation (1) is the direct composition of these two teachings.

Why a PHOSITA would combine them. They are in the same narrow sub-field (TCP congestion-avoidance throughput modeling), Mathis explicitly builds on and validates against Floyd, and Floyd's own model is a special-case linear topology expressly acknowledged in the patent's § 5.0 assumption (iii). Under KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), combining two references addressing the same problem with a finite number of identified, predictable solutions is the paradigm of obviousness. MPEP 2143 rationales (A) combination of prior art elements and (C) use of a known technique to improve a similar device in the same way (a) are both satisfied.

The inverse/dimensioning step (E4). A network designer asked "how much capacity does this link need?" would invert the throughput relation as a matter of routine algebra: if each demand's share is w_i/Σw_j · c, then setting the worst-served demand's share equal to its requirement d_i and solving for c yields c = max_i { d_i · Σw_j / w_i }. This is elementary algebraic rearrangement of Floyd/Mathis, not a new technique; MPEP 2143 rationale (B) ("simple substitution of a known technique in a known way"). The additive UDP term (E6) and the ceiling to integer trunks are ordinary engineering accommodations of a discrete resource.

Where the § 103 case is weaker (E5). The patent's genuinely non-trivial contribution is recognizing that h_i — the number of congested hops — is unknown at design time, and then bounding over the space of feasible congestion vectors {H}: consistency, the "common vector," the mapping g: I→H, and the extremal assignments H_worst (h_i = h̄_i for the bottleneck demand, h_j = 1 for all others) versus H_best (h_i = 1, h_j = h̄_j). Floyd and Mathis both assume the number of congested hops is known or given; neither formulates a bound over an unknown congestion pattern, and neither discloses the finding that c̄_l^FIFO(H_best) ≤ c_l^FIFO({H}) for the H_hop and H_one cases. I did not find, in any of the references surfaced, a motivation to enumerate congestion vectors as a design-time device. This is the crux of any nonobviousness position.

5. Combination 2 — WFQ/LQD capacity = Σ d_i

References: Kumar et al. (1998) + Suter et al. (1998).

Kumar et al. disclose a hierarchical WFQ scheduler with per-flow queues; Suter et al. expressly analyze "design considerations for supporting TCP with per-flow queuing" and conclude that per-flow fair queueing with longest-queue-drop buffer management delivers sufficient isolation and fairness for bandwidth guarantees. If a scheduler guarantees each flow its weighted share, then the capacity needed on a link to satisfy demands {d_i} is simply their sum — a one-step inference a PHOSITA would make without exertion. MPEP 2143 rationale (F) — "the known technique has been recognized as available to solve the problem" — applies squarely, since Suter's own title frames the work as a design investigation.

Caveat: as noted, Kumar (May 1998) and Suter (Mar. 1998) are § 102(a)-only art. This combination is therefore the most vulnerable to a sworn-behind or § 1.131 affidavit, and an examiner relying on it must affirmatively establish the invention date.

6. Combination 3 — topology optimization (augmentation / deloading)

References: Floyd/Mathis (capacity) + Hwang (1993, demand scaling) + the ordinary skill of a network planner, with Garey & Johnson for the complexity backdrop.

The deloading criterion is nothing more than a tunable utilization threshold against required capacity, followed by trial removal and cost recomputation — classic greedy network-design heuristics (Kruskal-like "remove the least useful link first"). Ordering candidates by Σ (demand × hop-count) is a straightforward perturbation-minimization heuristic. MPEP 2143 rationale (C) (known expedient) and rationale (G) (design incentives/market forces — reduce network cost) both apply; KSR expressly sanctions "the desire to enhance commercial opportunities" as a motivation.

The "keeper/straggler" partitioning is the weak point for a rejection. I found no reference teaching classification of end-to-end demands into a trunk-sized "core" set and a residual "straggler" set with threshold-controlled conversion. Absent such art, this limitation likely supports patentability — and note the patent's own FIG. 10D data show the capacity-only variant performing >30% worse, i.e., the specific conversion heuristics had measured, non-trivial consequences.

7. Combination 4 — router placement as an MIP

References: Garey & Johnson (1979) + Papadimitriou & Steiglitz (1982) + Fourer/Gay/Kernighan AMPL (1993) + the capacity computations of Combinations 1 and 2.

This is, on the face of it, the strongest § 103 case in the patent, because the specification itself supplies the motivation and characterizes the problem as known: § 6.0 concedes the problem "can also be transformed into a generalized knapsack problem," and the savings term S_i,j is realized only when both endpoints of a link are upgraded — exactly the pairwise-utility structure of a maximum-clique/graph-selection problem that Garey & Johnson and Papadimitriou & Steiglitz cover. Given (i) a known cost function M(C,L)+T(C), (ii) known per-link capacity requirements for two router types, and (iii) an express budget constraint N_max, formulating max Σ y_ij·S_ij − Σ x_i·Q_i subject to 0 ≤ y_ij ≤ x_i, x_j and Σ x_i ≤ N_max is the routine application of a standard MIP package. Rationale (C) — "obvious to try" with a finite number of identified solutions and a reasonable expectation of success — plus the patent's own admission that a 333-MHz Pentium II solves 100-node/300-link instances "within seconds" (evidence of predictable, routine optimization) make this combination difficult to defend.

8. Secondary considerations and countervailing arguments

A well-advised applicant would respond with:

Important framing: the § 103 case is materially stronger against the router-placement/MIP aspect and the WFQ/LQD sum-of-demands aspect than against the H-vector bounding aspect. If '399's independent claims are directed to the bounding methodology (as the "first aspect" and FIGS. 3–5 suggest), a § 103 rejection built on Floyd+Mathis alone is unlikely to succeed without additional art directed to worst-case congestion-scenario bounding.

9. Bottom line

• Likely obvious (Combinations 1–2): claims reciting the use of a published TCP throughput model to derive a per-link required capacity, the additive UDP term, and trunk-rounding — Floyd (1991) + Mathis (1997), with Kumar (1998) + Suter (1998) for the WFQ/LQD Σd_i case.
• Likely obvious (Combination 4): any claim substantively covering the MIP router-placement formulation — Garey & Johnson + Papadimitriou & Steiglitz + AMPL + the capacity references, reinforced by the specification's own admission that the problem reduces to generalized knapsack.
• Likely NOT obvious on this record: claims reciting the design-time bounding over feasible congestion vectors {H}, the consistency/common-vector formalism, and the specific H_worst/H_hop/H_one/H_best extremal assignments without prior knowledge of bottleneck locations; and the threshold-controlled/direct keeper-to-straggler conversion heuristics.
• Evidentiary weak points an examiner must cure: (i) Kumar (May 1998) and Suter (Mar. 1998) are § 102(a)-only art given the 1997-11-24 critical date; (ii) the four cited U.S. patents could not be content-verified here and two postdate the filing; (iii) the inventors' own Bell Labs Technical Journal paper is not prior art.

What must be verified before this analysis is final: the issued claims of US 6,795,399 from USPTO PatentCenter, and the actual "References Cited" list on its face — neither of which the supplied page text or my searches delivered. I am flagging this as an unresolved gap rather than an inference, consistent with the earlier sections of this analysis.