US 11563655

I'll search for information on this specific patent number, including any litigation activity.

Let me run more targeted searches on litigation, PTAB, and the current assignee.

Let me try a few more targeted searches on the assignee entity and litigation databases.

Search scope and a candid caveat

I was able to locate and verify the bibliographic record for US 11,563,655 B2 (the authoritative full text you supplied, sourced from https://patents.google.com/patent/US11563655/en, matches what I can confirm independently, including via a secondary aggregator confirming inventor Minkoo Kang and the 2023‑01‑24 issue date: https://www.patentleaderboard.com/samsung/minkoo-kang/269076).

However, I could not find any CAFC 2026 docket, district court docket, ITC investigation, or PTAB (IPR/PGR) proceeding specifically naming US 11,563,655. My searches returned many Federal Circuit items from 2025–2026, but none referencing this patent number; searches combining the number with "CAFC," "IPR," and the current assignee entity produced no hits for this number. I do not have authoritative confirmation that litigation is absent — only the absence of search evidence. I did not directly query PACER, the CAFC docket system, or USPTO Patent Center/PTAB APIs, so I cannot rule out a recently filed or sealed action. Treat the litigation conclusion as unverified, not as a negative finding.

1. Bibliographic summary

2. Abstract (as issued)

"Disclosed is a computing apparatus implemented with a network hypervisor implementing software defined network (SDN)-based network virtualization. The computing apparatus include a statistics virtualization module configured to provide individual statistics to each of created virtual networks, a transmission disaggregation module configured to include a physical statistics cache that performs periodic monitoring of a plurality of physical switches and store statistics of the physical switches collected, and a physical statistics aggregation module configured to respond with statistics of the plurality of physical switches when a single monitoring request."

Technical field/context: SDN-based multi-tenant network virtualization. The specification frames three problems: (i) physical-network-based monitoring cannot produce per‑tenant statistics; (ii) virtualization adds monitoring delay (modeled as d_v + n·d_p + d_NH); and (iii) control-channel consumption scales with tenant count. The three disclosed modules map 1:1 onto those problems.

3. Plain-language overview of each independent claim

Claim 1 — Statistics virtualization module (per‑tenant isolation).
A computing apparatus running a network hypervisor that virtualizes an SDN. A statistics virtualization module receives a request from a first virtual network (of multiple separately managed virtual networks) for that network's own individual physical-resource-consumption statistics. The module then determines whether the underlying physical resource is shared with another virtual network, and isolatedly provides the requested statistics to the first virtual network based on that shared/not-shared determination. The module supplies (a) virtual flow-entry statistics and (b) virtual port statistics derived from the physical resource. In substance: if the physical object is dedicated to one tenant, its stats pass through unchanged; if shared, the module must attribute/derive the tenant's own share rather than returning the aggregate physical number (the spec's example: physical port traffic of 60 is split into 10 / 20 / 30 across three tenants instead of 60 to all).

Claim 8 — Transmission disaggregation module (caching to cut latency).
A computing apparatus with a network hypervisor including a transmission disaggregation module containing a physical statistics cache that periodically monitors multiple physical switches and stores the collected physical statistics. The module computes virtual statistics for each of multiple separately managed virtual networks sharing those physical switches from the cached physical statistics, and serves them in response to isolated per‑tenant requests. Plainly: decouple ("disaggregate") physical statistics collection from virtual statistics calculation so tenant requests are answered from cache instead of round-tripping to switches — reducing the n·d_p term.

Claim 10 — Physical statistics aggregation module (fewer control messages).
A computing apparatus with a network hypervisor including a physical statistics aggregation module that answers a single monitoring request with statistics of a plurality of switches. The module performs (i) a pCollector filter that determines an execution cycle, and (ii) a pCollector tuner that adds delay to the first execution of each pCollector. Plainly: merge many small per-flow polls ("tiny pCollectors") into one broad "return all" poll ("aggregated pCollector") per switch, and time that poll's first run so cached data is fresh when tenant requests arrive. Note the claim as issued reads "physical statistics filter (pCollector filter) to determine an execution cycle of each of the physical statistics filter and physical statistics tuner" — self-referential wording that appears to be a drafting artifact; a parallel reference appears in the specification as "determines a period of the pCollector."

Claim 12 — Combination of all three modules.
A computing apparatus whose network hypervisor comprises, in combination: (1) a statistics virtualization module that isolatedly provides individual statistics to each of multiple separately managed virtual networks that simultaneously share a plurality of switches; (2) a transmission disaggregation module including a physical statistics cache performing periodic monitoring of those physical switches and storing the collected statistics; and (3) a physical statistics aggregation module that responds, with statistics based on the plurality of physical switches, to a single monitoring request for virtual statistics of at least one of those virtual networks. This is the broadest structural claim tying together isolation, caching, and aggregation.

Selected dependent-claim color: Claim 2 (flow-entry stats = packet count, byte count, duration; port stats = total TX/RX packets per tenant); Claim 3 (for shared physical flow entries, return the edge switch's flow entry statistics instead of the physical switch's); Claim 4 (per‑network individual edge switch); Claim 5 (virtual port stats consider the virtual flow-entry statistics); Claim 6 (port TX/RX derived by accumulating virtual flow entry stats); Claim 7 (virtual network = tenant network on a shared physical network); Claim 9 (transmission disaggregation performs request interval estimation using a mean and variance per physical flow); Claim 11 (the filter also checks whether collectors can be merged, and the tuner determines a starting delay).

4. Uncertainty and flags

• Litigation/CAFC status is unverified. No 2026 CAFC docket, district court case, or PTAB filing for US 11,563,655 surfaced in my searches. This is a negative search result, not an authoritative clearance. The current owner, "IP3 2023, Series 923 of Allied Security Trust I," is an entity name pattern commonly associated with patent monetization/assertion vehicles, which is a reason to expect litigation activity rather than evidence that it exists.
• Assignee verification caveat. The source itself states the assignee list "may be inaccurate" and that Google "has not performed a legal analysis."
• Internal inconsistency in the specification (not claim-affecting): the experimental section refers to "Example 2 is a non-virtualized native network" in one passage and "Comparative Example 2" elsewhere for the same baseline.
• Claim-number literalism: the four independent claims are 1, 8, 10, and 12 as issued (12 claims total). I did not auto-correct any identifier.

I'll search patent litigation databases for this specific patent number.

Those were mostly irrelevant. Let me target patent-specific litigation databases directly.

Let me try a few more targeted searches on the assignee entity and litigation trackers.

Litigation search — US Patent 11,563,655

Bottom line: No litigation found

I searched for litigation specifically involving US 11,563,655 (network monitoring in SDN network virtualization; current owner of record IP3 2023, Series 923 of Allied Security Trust I). I found no district court case, CAFC appeal, ITC investigation, or PTAB (IPR/PGR) proceeding naming this patent. I am reporting a negative search result, not a verified clearance that no case exists.

Because you asked for a specific case list, the honest answer is that the list is empty — I will not populate a table of plaintiff/defendant/jurisdiction/case number/filing date/status for cases I could not confirm, since fabricating those identifiers is exactly the failure mode to avoid.

What I actually searched, and what came back

I did not directly query PACER, the CAFC docket system (cafc.uscourts.gov docket), the ITC EDIS system, or USPTO Patent Center / PTAB APIs. Until those are checked, the absence of results above is search evidence, not an authoritative finding.

Relevant context on the owner (why litigation could exist even though I found none)

This is important framing rather than a case report:

1. The patent is now held by an AST "IP3" acquisition vehicle, not by the university. Google Patents records the assignment to IP3 2023, Series 923 of Allied Security Trust I, recorded 2024‑02‑06 (effective 2023‑11‑17; REEL/FRAME 066395/0826), and entity status changed from small to large entity on 2024‑03‑27 — consistent with the asset leaving the university and entering a monetization/acquisition structure. My earlier bibliographic summary flagged this same pattern.

2. AST's stated model is "catch and release" and AST itself does not litigate. AST buys patents, licenses members (currently ~35 companies, incl. Cisco, Google, Microsoft, IBM, Intel, Meta, Oracle, Sony, Uber, Verizon), then sells or donates the assets. A Czech-language AST overview states plainly that AST does not conduct litigation ("AST nevede soudní spory"): https://cs.wikipedia.org/wiki/Allied_Security_Trust. AST's own IP3 program announcements: https://www.ast.com/news/page/2/ ("AST Launches IP3 2023, a Collaborative Fixed-Price, Fixed-Term Patent Purchase Program").

3. AST-acquired assets do get asserted — but by third parties after resale, not by AST. RPX reported that GeoSymm Ventures LLC launched its first suits over a portfolio "received from Allied Security Trust (AST)" in June 2023, suing IBM (2:23‑cv‑00325) and Samsung (2:23‑cv‑00324) in E.D. Tex. and PTC (6:23‑cv‑00493) in W.D. Tex. (https://insight.rpxcorp.com/news/details?searchq=ents%3A%2813138195%29). Separately, a Chinese practitioner report noted that Secure Matrix LLC acquired two patent groups from "ALLIED SECURITY TRUST I" in May 2024 and has filed 39 patent suits, mostly in E.D. Tex. and W.D. Tex., all closed by settlement. Neither of those reports identifies US 11,563,655 as an asserted patent.

This matters: if '655 has been asserted, the plaintiff would most likely be a downstream acquirer of the IP3 2023 Series 923 asset, not AST or the university. I found no such suit naming the number.

Caveats you should carry forward

• Unverified negative. My searches were web-index–based and returned no hit for the number; I could not reach PACER, the CAFC docket, ITC EDIS, or PTAB direct APIs before my search budget ran out. A recently filed, sealed, or low-profile case could exist.
• Do not rely on the "no litigation" statement as a legal conclusion. For a defensible answer, verify directly in: PACER (all districts), the CAFC docket, USPTO Patent Center's "Litigation" tab / PTAB API for '655, and commercial trackers (RPX Insight, Docket Alarm, Lex Machina, Unified Patents Portal) keyed to patent number 11,563,655.
• Flag on prior analysis: my earlier litigation status was also unverified; this search does not change that — it confirms the same null result from additional angles, still without a primary-source docket check. No contradiction with the prior section was found.
• Identifier literalism honored: I did not auto-correct or normalize the number, the IP3 series designation ("IP3 2023, Series 923 of Allied Security Trust I"), or any case numbers cited (those case numbers belong to the GeoSymm/Secure Matrix matters, not to '655).

If you want, I can next attempt targeted docket-level lookups (e.g., CourtListener/Docket Alarm RECAP queries on the patent number and on "IP3 2023, Series 923 of Allied Security Trust I" as plaintiff) to try to convert this negative web-search result into a primary-source negative.

I'll verify the canonical "no proceedings" entry against live sources before writing up.

The first pass found no PTAB hits. Let me check the enforcement/monetization side and PTAB docket sources directly.

Let me make a final targeted pass on the enforcement entity and any related filings.

Proceedings overview

Zero AIA trial proceedings are on file against US 11,563,655. The structured "PTAB proceedings on file" block — sourced from the USPTO Open Data Portal, which is the canonical list — returns no IPRs, no PGRs, and no CBM petitions. My independent web searches (patent number + "IPR," "PTAB," "petition," "inter partes review") surfaced no petition, institution decision, Final Written Decision, or appeal referencing this number, and no PTAB docket entry naming the patent owner or its current assignee. Defensive posture: this patent is completely untested at the PTAB. There is no canceled claim to attack from, but there is also no adverse institution ruling to overcome, no § 315(e)(2) estoppel on any petitioner, and no Board claim-construction ruling to build a § 112 or § 101 theory around. A defendant today has a blank slate — maximum freedom on prior art, minimum roadmap.

No proceedings to report — what was checked

Because the required per-proceeding detail (petitioner, panel, grounds, FWD claim-level dispositions, settlements, CAFC dockets) does not exist for this patent, I am not manufacturing the sections. Fabricating a proceeding number or a claim-level cancellation here would be the single most damaging error possible in this deliverable.

Why the count is plausibly zero, rather than an indexing gap. This is a post-AIA patent (application 17/128,353, filed 2020-12-21; priority 2020-04-28), which forecloses one vehicle entirely and closes another:

• CBM is unavailable — permanently. The CBM transitional program sunset on 2020-09-16 and was in any event limited to claims directed to financial products or services. This patent claims SDN network monitoring; it is doubly outside the door.
• PGR is now time-barred. The PGR window is nine months from grant. Grant was 2023-01-24, so the window closed on or about 2023-10-24 with no filing.
• IPR remains the only AIA vehicle, and it is open-ended in time (subject only to the § 315(b) one-year bar running from service of a district court complaint, which has not occurred as far as I can determine). An IPR could therefore be filed tomorrow with no statutory bar.

Adjacent signals a defendant should weigh

These are not PTAB proceedings and I am not presenting them as such, but they materially change the defensive picture and sharpen the risk assessment.

1. The current assignee is a defensive aggregator's holding vehicle — a correction to the earlier section

The previously generated bibliography characterizes "IP3 2023, Series 923 of Allied Security Trust I" as "an entity name pattern commonly associated with patent monetization/assertion vehicles." That inference is wrong, and I want to flag the contradiction explicitly. IP3 2023 is Allied Security Trust's eighth annual Industry Patent Purchase Program — a collaborative, fixed-price patent buying program run by AST, a member-driven defensive cooperative whose members include Cisco, Google, IBM, Intel, Meta, Microsoft, Oracle, Salesforce, Snap, TikTok, Uber, and Verizon (AST announcement; results release). AST does not sue; its mandate is "catch and release."

That reframes the meaning of the 2024-02-06 assignment from Korea University Research and Business Foundation. This patent was acquired defensively, by consortium, to remove it from the assertion market — or at minimum to give AST's members a license position. It is not evidence of an impending PAE campaign.

2. But the patent has moved on again — and that is the exposure signal

AST's catch-and-release mandate means these assets are re-sold. Tangible IP announced on 2025-01-20 that it brokered the sale of AST's IP3 2023 patent portfolio (36 unique lots), and Tangible IP states it is "neither a law firm nor a patent assertion entity" but is a broker (Tangible IP press page). Whether US 11,563,655 was in a lot that sold, to whom, and on what terms is not determinable from public sources I could reach. Two consequences follow:

• The recorded assignee may already be stale, and the "IP3 2023 Series 923" entity may no longer hold title. Verify current ownership in USPTO Assignment records before relying on it.
• If a lot containing this patent passed to a litigating entity in 2025, an assertion wave and a corresponding IPR wave could be imminent. That is the scenario that would produce the first PTAB proceeding. The absence of PTAB activity today is consistent with the patent having sat in a defensive aggregator's inventory for its first three years.

3. The Korean family member has lapsed

The Google Patents legal-status table in the authoritative text records the Korean counterpart KR102322454B1 as "not_active" and "Expired - Fee Related." That has no effect on the U.S. patent's validity or enforceability, but it is a signal about the owner's willingness to pay maintenance costs on the family. Combined with the entity-status change from small to large on 2024-03-27 (the AST assignment), it tells you maintenance fees are being paid on the U.S. asset by a well-funded holder. Do not build a "co-owner abandoned the family" theory.

4. Prosecution history is short and gives a defendant useful ammunition

From the authoritative record: non-final rejection 2021-11-23 → response 2022-03-30 → final rejection 2022-06-27 → advisory action 2022-09-13 → allowance 2022-10-24. A final rejection followed by an advisory action and then allowance means the applicant amended the claims into their current form to get out of a § 103 rejection. Whatever was said there could carry prosecution-history estoppel or be leveraged in a § 112 / indefiniteness attack. Separately, the earlier section already flagged a drafting artifact in claim 10: the pCollector filter is recited as determining "an execution cycle of each of the physical statistics filter and physical statistics tuner" — self-referential language the parallel specification describes as determining "a period of the pCollector." That is a live § 112(b) exposure in an independent claim, and it is unaddressed by any PTAB ruling because none exists.

Strategic summary

Claim status: uniform and untested. All 12 claims as issued — independents 1, 8, 10, and 12, plus dependents 2–7, 9, and 11 — are UNTESTED. None are CANCELED. None are SUSTAINED. There is no narrowing amendment, no certificate of correction, and no reexamination certificate of any kind. Any assertion of this patent today asserts the original, as-granted claim set. The practical corollary: the four independent claims each stand on their own, so invalidating one gets a defendant nothing unless it takes out all four.

Estoppel landscape: completely clear. Because no IPR, PGR, or CBM was ever filed, no § 315(e)(2) estoppel attaches to anyone. There is no petitioner, no real party in interest, and no privy carrying a raised-or-reasonably-could-have-raised bar. Every prior-art ground remains fully available in district court — including:

• § 102/§ 103 combinations of patents and printed publications, usable both in an IPR and in litigation;
• System prior art, public use, on-sale art, and § 112 grounds, none of which are even available in an IPR — these are untouched and are a defendant's comparative advantage;
• § 101 eligibility, which is not an IPR ground at all. Given the Federal Circuit's treatment of network-monitoring claims in Packet Intelligence LLC v. NetScout Systems, Inc., 965 F.3d 1299 (Fed. Cir. 2020), where granular traffic-classification claims survived Alice, a § 101 attack here is not a gimme — but the claimed "statistics virtualization module" and "pCollector tuner" are markedly more functional and less structurally detailed than the parser/lookup-engine hardware recited in Packet Intelligence.

Pattern signals: none, because there is no pattern. The same petitioner has not filed multiple IPRs — nobody has filed one. The patent owner has not appealed anything to the Federal Circuit — there is nothing to appeal. No defensive aggregator appears in the proceeding chain, because there is no chain. The only aggregator presence is off-docket: AST's IP3 2023 ownership, followed by the announced 2025 portfolio divestiture through Tangible IP.

Recommended next steps

1. Verify current ownership before anything else. The assignment of record to IP3 2023, Series 923 (REEL/FRAME 066395/0826, effective 2023-11-17) may not reflect a post-Tangible-IP buyer. Run the chain in USPTO Assignment Search and Patent Center. Standing defects in an assignment chain are a cheap, high-leverage early defense if the plaintiff is not the record owner.
2. Treat the absence of PTAB activity as a signal about timing, not about strength. Well-asserted patents eventually attract IPRs. The null result is best explained by the patent's having spent its first years in a defensive aggregator's inventory. If an assertion letter arrives, expect the IPR clock to start contemporaneously — and note you have exactly one year from service of a complaint to file under § 315(b), with no ability to cure a late filing.
3. Commission a prior-art search now, while you have the full menu of grounds. Because there is no estoppel, litigation counsel can run system-art and § 112 theories in parallel with any IPR, and can hold back a ground as an IPR fallback. Do not burn your best references in a petition that a district court judge could have addressed.
4. Attack claim 10 on § 112(b) first, on the face of the claim. It recites the pCollector filter determining "an execution cycle of each of the physical statistics filter and physical statistics tuner" — a self-referential limitation with no antecedent support in the specification's own description of the same function ("determines a period of the pCollector"). Indefiniteness here is a claim-level kill that does not depend on any prior-art search.
5. Check the KR102322454B1 prosecution and the KR file history for art. The Korean counterpart is expired, so its file is a public, free, high-signal source — and Korean PTO examination often cites art the USPTO examiner never saw.
6. Monitor the PTAB docket going forward via USPTO PTAB Center and the ODP API. If a petition against this patent is filed, trial milestones are fixed: a § 314(b) institution decision within six months of the petition, and a Final Written Decision within 12 months of institution absent good cause.

Confidence and caveats

• High confidence that no AIA trial proceeding is on file: the canonical ODP-derived block says so, and independent searches corroborate it.
• Medium confidence on the negative for a very recently filed petition: I did not query PACER, PTAB Center, or the ODP API directly in this session, and a petition filed within the last few weeks may not yet be surfaced. Re-verify against the ODP API before relying on the null result in a filing.
• Not established: whether US 11,563,655 was included in a lot sold by Tangible IP in 2025, and to whom. I found the portfolio sale announcement but could not confirm this patent's disposition. Do not represent its current ownership as verified.
• Correction flagged to the earlier section: "IP3 2023, Series 923 of Allied Security Trust I" is AST's collaborative defensive-acquisition vehicle, not a monetization/assertion entity. The earlier section's inference to the contrary should not be relied on.

Ownership chain analysis — US 11,563,655 B2

Sources used: authoritative full text you supplied (https://patents.google.com/patent/US11563655/en, legal-events block), Google Patents "Also Published As"/family status data, the AST IP3 program record, the standard AST IP3 Patent Assignment Agreement form (Law Insider), Patentcloud weekly USPTO assignment digests (Jan/Feb 2024), RPX Insight, and AST/Unified Patents entity pages. I could not access the per-record correspondent field through the Assignment Center/UPC interfaces from this session — that gap is flagged rather than filled.

Inventors

Pattern assessment: no red flag. The inventor→university assignment was executed 2020-12-18 and recorded 2020-12-21, i.e., within three days of the 2020-12-21 US filing date — the ordinary "assign at filing" sequence, not a post-hoc cleanup. There is no evidence of inventors departing the assignee within 12 months, and no inventor appears in the later transfer to AST. The relevant abnormal signal for this patent is not inventor behavior but assignee behavior: a university research foundation sold the US patent ~10 months after grant without any intervening license, product, or spin-out recorded.

Original assignee

Korea University Research and Business Foundation (KURBF), Seoul, Republic of Korea — the technology-transfer / IP-holding arm of Korea University.

• Primary line of business: university technology commercialization and IP management. It is not an operating company and does not make network equipment or SDN software.
• Product embodying the claims: no evidence of one. The specification describes a research testbed (an OpenVirteX-style network hypervisor, ONOS tenant controllers, iperf3-generated TCP connections, linear and 4-ary fat-tree topologies, 20 switches). The Figures are experimental results, not product literature.
• Current status: operating (a live university foundation). It is the assignor, not the current owner.
• Sibling-asset signal: the Korean counterpart KR 10-202322454 B1 is listed by Google as "Expired – Fee Related." A university letting the home-jurisdiction grant lapse while selling the US counterpart is a hallmark of monetization-by-sale rather than commercialization.

Assignment timeline

Two recorded assignments. Both appear in the Google Patents legal-events block with reel/frame.

• 2020-12-18 (executed) / recorded 2020-12-21 — Reel 054705/0591

  • Conveyance: Assignment (Assignment of Assignors' Interest)
  • Assignor: YANG, Gyeongsik; KANG, Minkoo; YOO, Hyuck (individually)
  • Assignee: KOREA UNIVERSITY RESEARCH AND BUSINESS FOUNDATION, Republic of Korea
  • Correspondent: not retrievable in this session — data gap, not a negative finding. (Institution-to-university filings of this vintage are typically recorded by Korean outside counsel; I will not name one without the record.)
  • Context: Standard inventor-to-employer/university IP assignment at filing.

• 2023-11-17 (executed) / recorded 2024-02-06 — Reel 066395/0826

  • Conveyance: Assignment
  • Assignor: KOREA UNIVERSITY RESEARCH AND BUSINESS FOUNDATION
  • Assignee: IP3 2023, SERIES 923 OF ALLIED SECURITY TRUST I, New Jersey
  • Assignee address of record (from the standard AST IP3 instrument): 100 Overlook Center, 2nd Floor, Princeton, New Jersey 08540, U.S.A. — i.e., Allied Security Trust's own headquarters, not a commercial registered-agent mill.
  • Correspondent: not retrievable in this session — data gap. Note for follow-up: AST's IP3 sales function (patent_sales@ast.com; Mihir Patel, VP Technology) is the business contact; the recording correspondent on reel 066395/0826 must be pulled from the Assignment Center record directly.
  • Context: Transfer-to-defensive-aggregator via AST's eighth annual Industry Patent Purchase Program (IP3 2023) — a fixed-price, fixed-term cooperative purchase. Not a distress sale; AST's IP3 2023 post-mortem reports $4.334M across 48 deals / 53 families / 80 US patents, average ~$81,774 per family, with sellers including individual sellers and universities, and intent-to-purchase notices issued mid-October 2023 — which brackets this patent's 2023-11-17 execution date precisely.

Corroboration that this is a batch program, not a one-off: Patentcloud's USPTO assignment digests show the same assignee taking other university portfolios in adjacent reel blocks and the same recording window — Reel 066114/951 (recorded 2024-01-12, executed 2023-12-18) assignor University-Industry Cooperation Group of Kyung Hee University, 6 patents; Reel 066346/752 (recorded 2024-02-05, executed 2024-01-02) assignor The Research Foundation of State University of New York, 5 patents. Reel 066395/0826 sits squarely inside this run. That is the single most diagnostic fact in this file.

If you need the correspondent: query https://assignmentcenter.uspto.gov/ by patent number 11563655 and open record 066395/0826; the "correspondent" block is the field to extract, then cross-check it against reels 066114/951 and 066346/752 (same assignee, same window) to test the recurrence signal.

Timeline diagram

timeline
    title Ownership of US 11563655
    2020 : Korean priority application filed
         : Inventors assign to Korea Univ Foundation
         : US application filed 21 Dec
    2023 : US patent issued 24 Jan
         : Foundation assigns to IP3 2023 vehicle
         : Assignment executed 17 Nov
    2024 : Assignment recorded 6 Feb
         : Entity status restated as large

NPE / troll-pattern signals

1. Shell-entity transfer — present, with an important qualifier.
   The patent left the university foundation for a single-purpose series cell of a Delaware statutory trust ("Series 923 of Allied Security Trust I"), a structure with no employees and no products. Supporting record: Reel 066395/0826, executed 2023-11-17, recorded 2024-02-06. The qualifier that prevents this from being an anonymous shell finding: the vehicle is administered by a publicly identifiable organization at its own address (100 Overlook Center, Princeton NJ), the series naming is a standard AST IP3 artifact, and the same assignee name appears across dozens of unrelated portfolios in the same recording window. This is a batch acquisition cell, not a straw-man LLC.

2. Known asserter in the chain — not present.
   Neither assignee matches the named NPE lists (Acacia, Marathon, IV, IPNav, Wi-LAN/Mosaid-Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Spangenberg entities). AST is the opposite category: a member-driven defensive cooperative whose documented policy is that "AST does not litigate" and which operates a "catch, license, and release" model (AST membership 2023 includes Cisco, Adobe, IBM, Intel, Meta, Microsoft, Oracle, Philips, Salesforce, ServiceNow, Snap, Sony, Spotify, Uber, Verizon). Record: Reel 066395/0826 assignee name on its face.

3. Repeat correspondent across the chain — unclear (data gap).
   The Assignor in link 1 and the Assignee in link 2 are unrelated organizations, so a single recurring recording attorney across both links would be unusual and is not established here — I could not extract the correspondent field for either reel. Do not treat this as "not present." The specific test worth running: whether the correspondent on 066395/0826 also appears on 066114/951 and 066346/752 (same assignee, same window). Recurrence across those three would be an AST-program artifact, not an NPE tell.

4. Cascading transfers — not present.
   Exactly one post-issuance hop: Korea University R&BF → IP3 2023 Series 923, executed 2023-11-17 and recorded 2024-02-06 (~10 months after the 2023-01-24 grant). There is no chain of chained LLCs, no second conveyance, and no common-principal pattern visible.

5. Pre-litigation transfer — not present on current evidence.
   I found no district court, ITC, or PTAB proceeding naming US 11,563,655 (consistent with the earlier litigation check in this analysis). There is therefore no suit to measure the transfer against. Unverified negative, not a clearance — I did not query PACER, the CAFC docket, or PTAB APIs.

6. Bankruptcy fire-sale — not present.
   The assignor is a solvent university foundation; there is no Chapter 7/11, no court-supervised 363 sale, no SEC-disclosed insolvency event. The transfer is a voluntary, fixed-price program sale (AST IP3 2023: sellers set a fixed price; AST reports average family price ~$81,774). Different mechanism, different economics.

7. Privateering — unclear.
   Classic privateering is operating company → NPE asserts for it. Here the seller is a non-practicing university, and the buyer is a consortium of large operating companies' defense vehicle. The beneficiaries (Cisco, Google, IBM, Intel, Meta, Microsoft, Oracle…) are potentially adverse to any future asserter, which cuts against privateering. But the literature is explicit that AST's mandate is to "release" acquired portfolios, and there is documented downstream assertion of AST-sourced patents — RPX reported that GeoSymm Ventures LLC received a portfolio from AST in early June 2023 and promptly sued IBM (2:23-cv-00325) and Samsung (2:23-cv-00324) in E.D. Tex. and PTC (6:23-cv-00493) in W.D. Tex. So the released asset class is assertable; the record for this patent does not yet show that step.

8. Defensive aggregator / anti-NPE — present.
   The chain terminates at Allied Security Trust's IP3 2023 program vehicle (Reel 066395/0826). This is the inverse signal: the patent has been pulled into a defensive pool funded by 30+ large operating companies. Note that AST simultaneously lists its IP3 2023 acquisitions for resale and licensing, so this is neutralization-by-custody, not permanent neutralization — the asset can leave the pool.

Verdict

Defensive / non-asserting (chain terminates at a defensive aggregator) — with a documented downstream-risk caveat.

Justification: the only post-issuance conveyance is Reel 066395/0826 (executed 2023-11-17, recorded 2024-02-06), which moves the patent from Korea University Research and Business Foundation to IP3 2023, Series 923 of Allied Security Trust I — the acquisition vehicle of Allied Security Trust, a member-driven defensive cooperative that by policy does not litigate and that acquired this patent inside the same recorded transaction batch as other university sellers (Reels 066114/951 and 066346/752, recorded Jan–Feb 2024). The single most important caution against reading this as permanent neutralization: AST's model is "catch, license, and release," it publicly offers its IP3 2023 deals for resale, and RPX has documented AST-sourced patents reaching an asserting plaintiff (GeoSymm Ventures, June 2023). If this patent is later released, the next hop — not this one — is where an NPE verdict would be earned.

Explicit contradiction to flag with the earlier section: the prior section characterized the assignee name as "an entity name pattern commonly associated with patent monetization/assertion vehicles." That was a naming heuristic, and this research refines it: "IP3 2023, Series 923 of Allied Security Trust I" is not a monetization front — it is AST's defensive IP3 2023 purchase vehicle, resident at AST's headquarters (100 Overlook Center, 2nd Floor, Princeton NJ 08540), identifier corroborated by the Suny/ Kyung Hee co-assignments in the same reel range. The prior section's litigation conclusion (none found) stands; its ownership-characterization inference should be superseded by this section.

Verify at: https://assignmentcenter.uspto.gov/ (search patent number 11563655, records 054705/0591 and 066395/0826) and https://patents.google.com/patent/US11563655/en (Legal Events tab). Pull the correspondent field on 066395/0826 to close the one open signal (#3).

I'll research the specific records for US 11,563,655 and each of its cited references.

Let me research the remaining cited references in parallel.

Scope note before the analysis

I attempted targeted lookups on several of the eleven cited references and was able to independently confirm content for a subset (US 2006/0250959 A1, US 9,935,854 B2, US 2018/0287905 A1). The session hit its step limit before I could pull the full texts of the six Korean references (KR 2014‑0051776 A, KR 101485161 B1, KR 2016‑0041631 A, KR 2016‑0062688 A, KR 2016‑0107920 A, KR 101676570 B1, KR 20120122513‑family, KR 2019‑0063231 A). For those, the citation, priority date, publication date, assignee, and title are taken directly from the authoritative full text of US 11,563,655 supplied in the prompt (the Google Patents "Citations (11)" table), and the description is inferred from the title and assignee only — I flag each such inference explicitly rather than presenting it as confirmed.

Per the strict rule: I did not auto‑correct any identifier. The record analyzed is US 11,563,655 B2, application 17/128,353 — not any similar number.

1. The patent under analysis (USPTO identity check)

§ 102 critical date. The effective filing date for § 102 purposes is the earliest priority date the patent is entitled to: April 28, 2020 (KR 10‑2020‑0051783, incorporated by reference in the "CROSS‑REFERENCE" paragraph). Every one of the 11 cited references published or granted before that date, so each is facially available as prior art under at least § 102(a)(1) (patented/published before) and, for the earlier ones, § 102(a)(2)/pre‑AIA‑style § 102(b). I note § 102(a)(1) vs. § 102(a)(2) posture per reference below.

2. Master table — the 11 references of record

3. Detailed per‑reference entries

1. US 2006/0250959 A1 — Haim Porat, "Quality of service network and method"

• Full citation: US 2006/0250959 A1 (Porat), Quality of service network and method, published 2006‑11‑09; priority 2005‑02‑23.
• Confirmed content (verified this session): A configurable edge device for a carrier packet network implementing admission control and static/quasi‑static routing based on global network statistics. It discloses a data collector that collects and stores traffic data from edge devices and a calculator that computes traffic statistics for the entire network and derives reconfiguration data. (Source: https://patents.google.com/patent/US20060250959A1/en.)
• § 102 analysis: Facially available under § 102(a)(1). Potentially anticipates claim 1 only in a very strained reading — it teaches central collection of traffic statistics across edge devices, which maps loosely onto the "statistics virtualization module… receives a request… provides statistics" concept, but it is not an SDN/virtualization reference, has no virtual‑flow‑entry or virtual‑port statistics, and no notion of determining whether a physical resource is shared among tenants. I assess it as not anticipatory of any of claims 1, 8, 10, or 12. It is better characterized as background art on centralized statistics collection.

2. KR 2014‑0051776 A — ETRI, "Apparatus for network monitoring based on flow and network monitoring system"

• Full citation: KR 10‑2014‑0051776 A, 한국전자통신연구원 (ETRI), Apparatus for network monitoring based on flow and network monitoring system (flow‑based), priority 2012‑10‑23, published 2014‑05‑02.
• Description (title‑based inference, not independently verified): Flow‑based network monitoring apparatus + system, i.e., collecting and observing per‑flow statistics in a network.
• § 102 analysis: Facially § 102(a)(1). This is the closest of the Korean citations to the statistics‑collection aspect of claims 8 and 9 (periodic monitoring and storage of physical statistics; request interval estimation using mean/variance per physical flow). However, the title indicates flow‑based monitoring at the network level, without the recitations that give claim 8 its scope: a physical statistics cache in a network hypervisor, calculation of virtual statistics for multiple separately managed virtual networks sharing the switches, and isolated per‑tenant service. I do not find it anticipatory of claims 8 or 9 on the information available; at most it is § 103 material on the monitoring primitive.

3. KR 101485161 B1 — KT Corp, "System and method for providing real time streaming service"

• Full citation: KR 10‑1485161 B1, 주식회사 케이티 (KT), priority 2013‑12‑27, registered/published 2015‑01‑22.
• Description (title‑based inference): Real‑time streaming service provisioning system/method.
• § 102 analysis: Facially § 102(a)(1). Not anticipatory of any claim. Topically remote from SDN network virtualization, virtual/physical flow‑entry mapping, and cache‑based statistics disaggregation. Likely cited on the "quality/real‑time delivery" theme or as a KT‑family reference.

4. KR 2016‑0041631 A — ETRI, "Apparatus and method for quality of service aware routing control"

• Full citation: KR 10‑2016‑0041631 A, ETRI, priority 2014‑10‑08, published 2016‑04‑18.
• Description (title‑based inference): QoS‑aware routing control apparatus and method.
• § 102 analysis: Facially § 102(a)(1). Not anticipatory. Concerns routing control, not per‑tenant statistics virtualization, caching, or collector aggregation. § 103‑only relevance, if any.

5. KR 2016‑0062688 A — ETRI, "Overlay network based on the original packet flow mapping apparatus and method therefor"

• Full citation: KR 10‑2016‑0062688 A, ETRI, priority 2014‑11‑25, published 2016‑06‑02.
• Description (title‑based inference): Overlay‑network apparatus/method built on mapping between original packet flows (i.e., mapping an overlay flow to an underlying/original flow).
• § 102 analysis: Facially § 102(a)(1). This is the Korean citation with the most direct § 102 relevance to claim 1's "mapping" concept — claim 1's virtualization module resolves a requested virtual entity to the physical resource and determines sharing; dependent claims 3–6 turn on virtual‑flow‑entry ↔ physical‑flow‑entry mapping. If this reference discloses mapping an overlay/virtual flow identifier to an original/physical flow identifier, it would be § 103 material on the mapping step, but it does not appear to disclose the isolated per‑tenant statistics provisioning that is the core of claim 1. Not anticipatory on available information.

6. KR 2016‑0107920 A — Sungkyunkwan University, "Autonomous control method in software defined network"

• Full citation: KR 10‑2016‑0107920 A, 성균관대학교산학협력단, priority 2015‑03‑06, published 2016‑09‑19.
• Description (title‑based inference): Autonomous (self‑driven) control method within an SDN.
• § 102 analysis: Facially § 102(a)(1). Not anticipatory. SDN autonomous control is not "responding with statistics of a plurality of switches to a single monitoring request" nor "adding delay to the first execution of each pCollector." Weak § 103 relevance for the claim‑10/12 SDN‑context preamble at most.

7. KR 101676570 B1 — KT Corp, "Method and system for providing virtual data center"

• Full citation: KR 10‑1676570 B1, 주식회사 케이티 (KT), priority 2015‑06‑18, registered 2016‑11‑15.
• Description (title‑based inference): Method/system for providing a virtual data center.
• § 102 analysis: Facially § 102(a)(1). A "virtual data center" reference is § 103‑relevant background for the virtualization‑environment preamble of claims 1/8/12, but the title discloses no statistics virtualization, no physical‑statistics cache, and no pCollector aggregation. Not anticipatory.

8. US 9,935,854 B2 — Uila Networks, "Infrastructure performance monitoring" ⭐ strongest US citation

• Full citation: US 9,935,854 B2 (Kuan, Wu, Nguyen), Uila Networks, Inc., filed 2014‑09‑23, granted 2018‑04‑03; published as US 2016/0087861 A1.
• Confirmed content (verified this session): An infrastructure‑performance‑monitoring system that provides network analysis for virtualized networks, "offering visibility into the virtualized network for an operator to enable analysis of the entire data center and/or individual tenants." It expressly criticizes prior tools because they "may not obtain individual tenant data in a multi‑tenant data center" and lack "virtual network visibility," then monitors at the virtual layer using virtual taps that duplicate packets, a virtual network packet broker (vNPB) that aggregates the duplicated packets, and monitoring tools that correlate/aggregate per flow, grouped by virtual entity (VM, virtual switch, "port group/tenant," virtual network). It determines virtual entities and then the physical entities associated with them and stores a mapping of the virtual infrastructure. (Sources: https://patents.google.com/patent/US9935854B2/en; https://patents.justia.com/patent/9935854.)
• § 102 analysis — the most serious § 102 candidate against claim 1 and 12:
  • Claim 1: Partial. Uila discloses (a) a monitoring platform for multiple virtual networks/tenants, (b) individual tenant traffic data ("monitoring tool for data center tenant 116 to monitor the individual tenant traffic data"), and (c) a mapping between virtual entities (port group/tenant, virtual network, virtual NIC) and physical entities (switches, hosts). What it does not appear to disclose is the claim‑1 pivot: determining whether "the at least one physical resource is shared between the first virtual network and at least one other" virtual network, and isolatedly providing statistics based on that determination. Uila derives per‑tenant visibility by tapping and grouping duplicated packets by the identified port group/tenant — a different mechanism than attributing a shared physical resource's statistics. On the available record I assess Uila as § 103 material on claim 1 (and claim 12's isolation element), not a clean § 102 anticipation.
  • Claim 8: Uila's per‑VM/per‑port‑group aggregation of duplicated flows is not the claimed "physical statistics cache that performs periodic monitoring of a plurality of physical switches and stores statistics of the physical switches collected" with virtual statistics calculated from the cache. Not anticipatory.
  • Claim 12: Same deficiency as claim 1 for the isolation element; no pCollector aggregation. Not anticipatory.
  • Caveat: I verified Google Patents and Justia text, not the full 19‑figure specification; a passage expressly teaching cache‑based derivation of per‑tenant statistics from shared‑resource counters cannot be ruled out.

9. US 2018/0287905 A1 / US 10,291,497 B2 — Juniper Networks, "Session‑based traffic statistics logging for virtual routers" ⭐ strongest § 102 candidate against claim 8/9

• Full citation: US 2018/0287905 A1 (Mehta, Viswanathan, Bhatt, Singh R.), Juniper Networks, Inc., filed 2017‑03‑31, published 2018‑10‑04; granted as US 10,291,497 B2 on 2019‑05‑14 (EP family EP3382959A3 / EP4020932).
• Confirmed content (verified this session): A virtual router in a cloud data center processes traffic flows for VMs to generate traffic flow statistics (5‑tuple, packet count, byte count); a virtual routing agent aggregates and identifies ingress/egress flow pairs for a common session and uploads session records to a traffic collector of the SDN controller, which stores them in an analytics database. The virtual router "periodically communicates" the generated statistics to the VR agent. It operates in an SDN/virtual‑network overlay environment with per‑tenant virtual networks ("each VM 48 is associated with one of the virtual networks VN0‑VN1 and represents tenant VMs"). (Sources: https://patents.google.com/patent/US10291497B2/en; https://uspto.report/patent/app/20180287905.) Its CPC class H04L43/062 and H04L41/142 exactly match two of the CPC classes on the patent under analysis.
• § 102 analysis:
  • Claim 8: Closest § 102 candidate. Every element is arguably present except one: claim 8 requires the cache to be in a network hypervisor and to compute virtual statistics for "multiple separately managed virtual networks that share the plurality of physical switches… based on isolated requests for individual physical resource consumption statistics." Juniper's collector computes statistics at the virtual router/VR‑agent level (per host), uploads to the SDN controller traffic collector, and its per‑flow statistics are the flows of that virtual router — there is no disclosure of one cache serving multiple tenants from shared physical‑switch counters and no "physical statistics cache" terminology or periodic hardware‑switch polling. I assess claim 8 as highly § 103‑relevant and borderline § 102; not a clean anticipation because the "cache at the hypervisor serving multiple sharing virtual networks" element is not shown.
  • Claim 9: Juniper's periodic aggregation/upload and its per‑flow byte/packet counting relate to claim 9's "request interval estimation… using a mean value and a variance value per physical flow," but Juniper's "diff bytes/diff packets" per monitoring period is not disclosed as a mean/variance estimation of request intervals. Not anticipatory of claim 9; § 103 material.
  • Claims 1, 10, 12: No virtual‑port statistics derived from virtual flow entries, no shared‑resource determination, no pCollector filter/tuner. Not anticipatory.

10. KR 2018‑0122513 A — Korea University R&B Foundation, "Method and framework for traffic engineering in network hypervisor of SDN‑based network virtualization platform" ⭐ most dangerous reference overall (same assignee, same lab)

• Full citation: KR 10‑2018‑0122513 A, 고려대학교 산학협력단 (Korea University Research and Business Foundation), priority 2017‑05‑02, published 2018‑11‑13.
• Description: By its title, a method and framework for traffic engineering in the network hypervisor of an SDN‑based network virtualization platform. This is the same technology base and the same assignee as US 11,563,655 (Korea University; inventor Hyuck Yoo is the thesis advisor credited in the related doctoral work "Programmable Network Virtualization for SDN‑based Cloud Systems," 고려대학교 대학원, 2019 — https://dcollection.korea.ac.kr/srch/srchDetail/000000084335). It is therefore the most likely § 102/§ 103 reference on the hypervisor‑side traffic‑engineering/monitoring concept.
• § 102 analysis: Facially § 102(a)(1) (published 2018‑11‑13, before the 2020‑04‑28 priority date).
  • Claims 1, 8, 9, 12: A network‑hypervisor traffic‑engineering framework plausibly discloses per‑virtual‑network isolation of network states inside the hypervisor, virt‑to‑phys mapping, and statistics used for traffic engineering — i.e., much of the environment of claims 1/8/12 and potentially the request‑interval/statistics machinery of claim 9. If it discloses populating a hypervisor‑side store from periodic physical‑switch polling and answering per‑tenant virtual statistics from that store, it could anticipate claim 8 and seriously threaten claims 1/12. I could not retrieve the Korean text this session to confirm, so I rate the anticipation risk as material but unverified — this is the reference I would pull first, including its Korean prosecution file and the family member(s).
• Caveat: The patent's own specification discusses OpenVirteX and "Libera" (from the 2019 PhD thesis of the same lab) as the state of the art, which raises the possibility that the claimed subject matter was distinguished from this reference during prosecution. The published Korean application is prior art regardless.

11. KR 2019‑0063231 A — Altist Inc., "Data center management system based on SDN, and method thereof"

• Full citation: KR 10‑2019‑0063231 A, 주식회사 알티스트 (Altist Inc.), priority 2017‑11‑29, published 2019‑06‑07.
• Description: Per the patent's own Background (authoritative text): "an SDN‑based data center management system includes an SDN controller, a network hypervisor manager, and a resource monitoring module for monitoring network traffic, and a method thereof." (See the "BACKGROUND" and "DEFINITIONS/Description" passages: "In Korean Patent Publication No. 10‑2019‑0063231, it is disclosed that an SDN-based data center management system…")
• § 102 analysis: Facially § 102(a)(1). This is the only citation the patent itself characterizes in the specification, i.e., the applicant conceded it as the closest known art. It discloses an SDN controller + network hypervisor manager + resource monitoring module — i.e., it maps onto the structural environment of claims 1/8/12 (a hypervisor implementing SDN virtualization with a monitoring/resource module).
  • Claim 1: May disclose the statistics/network‑monitoring module and the hypervisor‑manager environment, but the title/Background give no indication of determining whether a physical resource is shared among tenants or of virtual flow‑entry / virtual port statistics. Not a clean anticipation of claim 1.
  • Claim 8: No indication of a physical statistics cache periodic‑monitoring‑and‑store with hypervisor‑side virtual‑statistics calculation. Not anticipatory.
  • Claim 12: The three‑module combination (statistics virtualization + transmission disaggregation + physical statistics aggregation) is the crux; a generic "resource monitoring module" in an SDN data‑center manager does not disclose the disaggregation or aggregation modules. Not anticipatory; strong § 103 background.
  • Note also the applicant's own statement that "it is difficult to overcome the limitations of various monitoring methods" with this reference — consistent with it being non‑anticipatory.

4. The four "family cites" (listed below the 11 examiner citations)

These are cited in family‑member prosecution (not necessarily by the examiner in the US case), so they are § 102 prior art only to the extent a family member relies on them. Interpreted literally from the record:

5. Non‑patent literature of record

• Yang et al., 2020, "Network monitoring for SDN virtual networks." Listed in the record's "Similar Documents" and consistent with the inventors' own work (Korea University; inventor Gyeongsik YANG). If this was published before 2020‑04‑28, it is § 102(a)(1) prior art by the inventors themselves and is a critical reference on the statistics‑virtualization and pCollector concepts. Its 2020 date sits exactly on the priority boundary, so I flag it as date‑sensitive and unverified; it should be date‑checked against the April 28, 2020 KR priority date.

6. Synthesis — best § 102 candidates, ranked

1. KR 2018‑0122513 A (Korea University) — highest anticipation risk, on claims 8, 9, 1, 12; same assignee/lab. Unverified text — retrieve first.
2. US 9,935,854 B2 (Uila Networks) — strongest verified § 102 candidate on claim 1 / 12 (per‑tenant visibility in a multi‑tenant virtualized data center), but its tap‑and‑group mechanism likely fails the "determine whether the physical resource is shared, and isolate based on that determination" limitation → probably § 103, not § 102.
3. US 2018/0287905 A1 / US 10,291,497 B2 (Juniper) — strongest verified § 102 candidate on claim 8 / 9 (periodic per‑flow statistics in a virtual router, aggregated and uploaded to an SDN traffic collector), failing the "hypervisor‑side physical statistics cache serving multiple sharing virtual networks" element → § 103, borderline § 102.
4. KR 2019‑0063231 A (Altist) — the applicant‑acknowledged closest art; environment‑level disclosure only → § 103.
5. KR 2014‑0051776 A (ETRI) — flow‑based monitoring primitive → § 103.

References I would not press for § 102: US 2006/0250959 A1, KR 101485161 B1, KR 2016‑0041631 A, KR 2016‑0062688 A, KR 2016‑0107920 A, KR 101676570 B1 — each is topically adjacent at best.

Claims 10 and 11 have no cited reference that discloses their core. The pCollector filter (selecting a period value that maximizes the count of tiny pCollectors following a similar period, comparing that ratio to a threshold such as 70%) and the pCollector tuner (setting a starting delay of an aggregated pCollector to immediately precede the first merged virtual‑statistics request) appear to be the most novel and least anticipated subject matter in the patent, and none of the 11 citations (or the four family cites) discloses it. That is consistent with the fact that the claims were allowed. If validity is later challenged, claims 10/11 — and the claim‑12 combination that incorporates them — are the hardest to defeat on this art.

7. Important caveats to carry forward

• A citation is not an anticipation finding. Being listed in the "Citations (11)" table means these references were of record; it does not mean the examiner used them in a § 102 rejection. The prosecution history shows a non‑final action (2021‑11‑23), a response (2022‑03‑30), a final rejection (2022‑06‑27), an advisory action (2022‑09‑13), then allowance (2022‑10‑24) — the specific rejections and the art relied on are not in the supplied record, so I cannot attribute any particular reference to any particular rejection. That file wrapper (and the Korean counterpart KR 102322454 B1's file) should be obtained to see which references actually drove the rejections.
• Six Korean references were not independently verified this session. Their descriptions above are title/assignee‑derived inferences, explicitly flagged. Do not treat those descriptions as confirmed disclosures.
• Date‑sensitive item: the "Yang et al., 2020" NPL entry must be date‑checked against the 2020‑04‑28 priority date before any § 102(a)(1) reliance.
• This analysis builds on the earlier sections (bibliography, claim readings, and the unverified litigation/CAFC status of the IP3 2023, Series 923 of Allied Security Trust I ownership). Nothing here contradicts those sections; it extends them with the prior‑art layer.

I'll analyze obviousness under § 103 using the prior art of record in this patent's "Citations" section, plus the family/similar-document references. Let me verify details on the key references so the analysis is grounded.

Good — the family/"similar documents" listing surfaced the inventors' own V-Sight paper, which is important for the date analysis. Let me verify a few of the cited references so the claim charts are grounded.

Obviousness Analysis — US 11,563,655 B2 under 35 U.S.C. § 103

Building on the previously generated bibliographic summary (§1), abstract (§2), and independent-claim overview (§3). I do not repeat those; I take the claim scope as already summarized there and the prior-art-of-record list as given.

0. Governing framework and posture

AIA applies. The application was filed 2020‑12‑21 with a KR priority of 2020‑04‑28 — both after the March 16, 2013 AIA effective date. So § 103 analysis runs through AIA § 102/103, with the effective filing date = 2020‑04‑28 for subject matter supported by the KR priority application.

Standard: Graham v. John Deere factors (scope/content, differences, PHOSITA level, secondary considerations), applied through the KSR lens — a combination is obvious where (a) the references are analogous art, (b) the art teaches/suggests each claimed element, and (c) there is an articulated reason with a reasonable expectation of success (MPEP 2143; KSR rationales: predictable variation, known technique to improve similar device, obvious design need, market/technological trend).

Level of ordinary skill (my proposed PHOSITA): an engineer (BS/MS CS or EE) with ~2–3 years in SDN/OpenFlow control-plane engineering, familiar with network hypervisors (FlowVisor/OpenVirteX), OpenFlow multipart stats requests, and the standard monitoring latency/overhead tradeoff. This matters because much of what the patent calls "inventive" is conventional control-plane engineering.

Critical threshold point — the strongest reference is the patent's own publication, and it may be disqualified. The "Similar Documents" listing on this page includes Yang et al. 2020, "Network monitoring for SDN virtual networks" — which is V-Sight (IEEE INFOCOM 2020, https://dl.acm.org/doi/abs/10.1109/INFOCOM41043.2020.9155260), authored by Gyeongsik Yang, Heesang Jin, Minkoo Kang, Gi Jun Moon, Chuck Yoo — i.e., the named inventors. Its abstract recites the exact three mechanisms of the claims: "1) statistics virtualization for isolated statistics, 2) transmission disaggregation for reduced transmission delay, and 3) pCollector aggregation for efficient control channel consumption," implemented on OpenVirteX, reducing delay and control-channel consumption "up to 454 times" (the patent's §5.3 reports the same "454 times" figure).

• Its ACM publication date is 2020‑07‑06, which is after the 2020‑04‑28 priority date. Under AIA § 102(a)(1) it is therefore not prior art, and in any event § 102(b)(1)(A) excepts a disclosure made by the inventors within the grace period.
• Practical consequence: the single closest disclosure to the claims is unavailable as prior art. Any § 103 rejection must be built from the other references of record, and the examiner's allowance (after a final rejection of 2022‑06‑27 and an advisory action of 2022‑09‑13 — see §1) is consistent with that narrower art base. I could not retrieve the actual Office actions, so I cannot confirm the examiner's stated reasons; treat that as a flag.

1. What each reference of record actually teaches

I verified content for the U.S. references; for the Korean-language citations I was limited to titles/abstracts and the patent's own characterizations (which I flag as second-hand).

Admitted prior art (AAPA): The specification itself states that OpenVirteX, "a representative network hypervisor that realizes SDN network virtualization, also performs network monitoring" by "collecting statistics such as topology status, flow statistics for each switch, and port status in the physical network and transmitting the statistics to the SDN controller." This is a binding admission that periodic per-switch/per-flow/per-port statistics collection and delivery in an SDN hypervisor was known.

2. Combination analysis by independent claim

Claim 1 — statistics virtualization (per-tenant isolation)

Primary: US 9,935,854 (Uila), optionally + AAPA (OpenVirteX) + KR '122513.

Element-by-element:

• "network hypervisor implementing SDN-based network virtualization" — AAPA (OpenVirteX) + KR '122513 (same-assignee hypervisor).
• "receive a request for … statistics for a first virtual network of multiple separately managed virtual networks" — Uila's per-tenant monitoring tool (item 116) receiving tenant traffic data; Uila expressly serves "individual tenants."
• "determine whether the at least one physical resource is shared … and isolatedly provide … based on the determination" — Uila's "determin[e] the virtual infrastructure," "determin[e] physical entities … associated with the virtual entities," and "group the traffic data based at least on the virtual entities" (by port group/tenant). Determining whether a physical entity maps to one vs. many virtual entities is the same comparison, just with a cardinality test.
• "virtual flow entry statistics" + "virtual port statistics" — Uila's per-VM/per-port-group statistics + AAPA flow stats.

Motivation (strong): Uila's own Background supplies the rationale verbatim — multi-tenant data centers need per-tenant visibility, and existing tools "may not obtain individual tenant data." Attribution of shared physical resources to tenants is necessary for tenant billing, SLA enforcement, and isolation (§1 "problem 1"). Reasonable expectation of success: the mapping data needed for the cardinality test is already maintained by the hypervisor (KR '122513).

Weakness: Uila monitors via taps/vNPB rather than OpenFlow counter reads from physical switches. A strict reading could say Uila doesn't teach deriving virtual flow-entry statistics by reading physical flow entries. Patch: combine with the ETRI flow-monitoring reference (KR '1776) or with the admitted OpenVirteX behavior. This is the weakest link in the Claim 1 attack.

Claim 8 — transmission disaggregation (cache + per-tenant virtual stats)

Primary: US 9,935,854 or Altist KR '231 (periodic monitoring + monitoring module) + AAPA.

• "physical statistics cache … performs periodic monitoring of a plurality of physical switches and stores statistics … collected" — AAPA discloses the periodic collection; caching the result is the classic "known technique to improve similar device" (KSR): buffering/caching stale-but-fresh-enough data to avoid a round trip is a ubiquitous, predictable optimization. Juniper '905 reinforces it — the VR agent "periodically" receives stats and "maintains a similar record," i.e., a stored local copy used to answer without re-querying.
• "calculates virtual statistics for each of multiple separately managed virtual networks … based on the physical statistics in the cache" — Uila's grouping/aggregation of collected traffic data by virtual entity.
• "provides … based on isolated requests" — Uila.

Motivation (strong): the patent's own Background quantifies the problem — "physical statistics transmission time, n·d_p, should be reduced" — which is exactly the articulated reason a PHOSITA would adopt caching. The specification's own model, d_v + n·d_p + d_NH, assumes the desirability of eliminating the n·d_p term; that is an admission of the design goal.

Claim 10 — physical statistics aggregation (single request → many switches; filter + tuner)

This is the claim with the least comfortable art base, and where the previously-flagged self-referential wording matters ("pCollector filter to determine an execution cycle of each of the physical statistics filter and physical statistics tuner").

• "respond … with statistics of a plurality of switches [to] a single monitoring request" — This is standard OpenFlow. A single OFPMP_FLOW (OFPT_MULTIPART_REQUEST) with an empty/ALL match returns flow statistics for all entries on a switch; polling the fabric via one request pattern is conventional (OFPMP_AGGREGATE, OFPP_ANY). The patent admits the mechanism: the aggregated pCollector "includes 'all' as its request." This element is squarely predictable-variation art.
• "pCollector filter to determine an execution cycle" — Juniper '905's "algorithm to provide a statistically accurate representation" for selecting which flows/statistics to collect, plus the general adaptive-polling-interval art, supply this. The patent's own filter uses a cumulative-distribution/hit-rate method and a similarity-count — a routine statistical tuning.
• "pCollector tuner to give additional delay to a first execution of each pCollector" — timing a periodic poll so its first run is phase-aligned just before expected requests is a design choice with a clear, predictable benefit (fresh cached data). KSR "obvious design need."

Assessment: Claim 10 as a structural/functional claim is arguably obvious over OpenFlow-all-request + Juniper '905 + a scheduler, but the "additional delay to a first execution" and dependent Claim 11's "starting delay" heuristic (95% of period for tiny collectors; immediately before the first merged request for aggregated collectors) are the most defensible non-obviousness redoubts — they encode a specific, non-trivial phase-alignment optimization rather than a generic timer. Expect a rejection here to be contested on the rationale for why the tuner's specific delay value would have been predictable.

Claim 12 — the three-module combination

Obvious as an aggregation of Claims 1/8/10 combinations. Under KSR, where each module addresses a distinct, independently recognized problem (isolation, latency, overhead — the three problems the Background enumerates), combining them is the predictable use of prior-art elements according to their established functions. The only meaningful § 103 question is whether the combination produces an unexpected result; the patent's own data (Table: 46×–454× delay improvement, 1.44×–1.9× control-channel saving, ≤2.5 s accuracy overhead) is asserted as additive, not synergistic, and the accuracy overhead is worse, not better — which undercuts a "unexpected results" rebuttal.

Dependents

• Claim 2 — packet/byte/duration counts: Juniper '905 (packet/byte counts) + Uila.
• Claim 3/4 — edge-switch flow entry substitution: AAPA + hypervisor mapping (KR '122513). Arguably the sharpest technical point; needs a reference teaching edge-vs-core handling.
• Claim 5/6 — deriving port TX/RX by accumulating virtual flow-entry stats: Juniper '905's ingress/egress pairing and per-flow counters.
• Claim 7 — tenant networks on a shared physical network: Uila (multi-tenant data center) + AAPA.
• Claim 9 — request interval estimation via mean/variance per physical flow: no cited reference squarely teaches this. KSR "obvious to try"/routine statistics may support it, but this element is a presumptive weak spot for any rejection.
• Claim 11 — see Claim 10 discussion.

3. Summary of the strongest § 103 attack

1. Claim 1: US 9,935,854 (Uila) in view of AAPA (OpenVirteX monitoring) and KR 10‑2018‑0122513; optionally + ETRI KR 10‑2014‑0051776 for flow-statistics derivation.
2. Claim 8: Uila '854 and/or Altist KR 10‑2019‑0063231 in view of AAPA, further in view of Juniper US 2018/0287905 for the periodically-updated stored-stats record; rationale = the patent's own admitted latency model.
3. Claim 10: OpenFlow multipart "all-entries" statistics request (admitted in the specification) + Juniper '905 (statistically-representative subset selection) + a periodic scheduler; rationale = known technique to reduce control-channel load, a goal the Background admits.
4. Claim 12: aggregation of the above, per KSR.

Bottom line: Claims 1, 2, 5–8, and 12 appear vulnerable to a well-supported § 103 rejection. Claim 10 is borderline (vulnerable on the structural elements, defensible on the "additional delay to first execution"). Claims 3, 4, 9, and 11 are the strongest candidates for nonobviousness absent references beyond the record — particularly Claim 9's mean/variance per-physical-flow interval estimation and Claim 11's starting-delay heuristic, for which I found no teaching among the cited art.

4. Secondary considerations (and why they likely don't save the claims)

• Unexpected results: not credibly shown. The measured improvements are directionally what the design predicts; the accuracy cost is a known tradeoff the patent admits.
• Long-felt need: mixed — Uila's 2014/2018 filing already identified per-tenant visibility as a need, which supports obviousness (the need was known and being addressed) more than patentability. This is a double-edged fact.
• Copying/commercial success: no evidence in the record. The 2024-02-06 assignment to IP3 2023, Series 923 of Allied Security Trust I is a monetization signal, not a secondary-consideration fact (and is not § 103 evidence at all).

5. Explicit caveats

• I could not retrieve full text of the Korean-language citations (KR '1776, KR '1631, KR '2688, KR '7920, KR '6570, KR '33179, KR '63231) or the actual Office actions. My characterizations of those rely on their titles and on the patent's own summary; they are second-hand. A definitive § 103 opinion requires the cited documents' full text and the prosecution history.
• The V-Sight paper (the closet disclosure) is the inventors' own work published after the priority date and is therefore not usable prior art; I flag it because any analyst encountering the "Similar Documents" entry might otherwise assume it is available.
• Claim 10's self-referential filter wording (flagged in the earlier summary) is a claim-clarity exposure; a § 112(b) challenge may be more effective against Claim 10 than § 103.
• The earlier-noted internal inconsistency ("Example 2" vs. "Comparative Example 2") is a specification defect and does not affect claim scope or this analysis.
• I did not independently verify the legal status or litigation posture of this patent here; the earlier section's litigation conclusion remains unverified, and none of it bears on obviousness.