US 5712976

I'll run targeted searches for the patent number and 2026 docket activity.

Technical Summary — U.S. Patent 5,712,976

Searches run (2026-09-23): (1) patent-number-specific queries on USPTO-derived mirrors and Google Patents; (2) CAFC 2026 docket/Federal Circuit decision queries restricted to "5712976" / "5,712,976."

CAFC 2026 docket result: no hits attributable to this patent. Every 2026 Federal Circuit item returned involved other patents (Ironburg '525; Digital Turbine '256; Hafeman '122/'393/'287; TrackTime '638/'978; Berkeley*IEOR '521/'137/'316; AGI SureTrack). A separate query for IBM-related 2026 appellate activity surfaced only LZLabs GmbH v. IBM, Fed. Cir. No. 24-2130, which concerned U.S. 8,713,289 B2 (mainframe condition-code emulation) and was dismissed by stipulated order on 2026-01-13 — a different patent. No petition, appeal, or 2026 docket entry referencing 5,712,976 was found. This is consistent with the earlier sections of this file and with the patent's 2015 expiration.

1. Bibliographic record

Sibling filings cross-referenced on the face of the patent (all filed 1994-09-08, all IBM): 08/302,616 → US 5,603,058; 08/302,619 → US 5,586,264; 08/302,625 (Belknap et al.); 08/302,626 (Henley et al.); 08/303,190 (Saxena et al.). Foreign family members per the Google Patents legal-event feed: CA 2,154,511 A1; CN 1122985 A; SG 32444 A1; BR 9503820 A; JP 3096409 B2; WO 96/08112 A1; KR 0184627 B1.

2. Abstract (verbatim, as printed)

A media streamer (10) includes at least one storage node (16) including mass storage for storing a digital representation of at least one video presentation. The mass storage is comprised of a plurality of mass storage units. The at least one video presentation requires a time T to present in its entirety, and is stored as a plurality of N data blocks. Each data block stores data corresponding to approximately a T/N period of the video presentation. The media streamer further includes a plurality of communication nodes (14) each having at least one input port that is coupled to an output of the at least one storage node for receiving a digital representation of a video presentation therefrom. Each of the plurality of communication nodes further includes a plurality of output ports individual ones of which transmit a digital representation as a data stream to a consumer of the digital representation. The N data blocks of the digital representation are partitioned into X stripes, wherein data blocks 1, X+1, 2*X+1, . . . etc., are associated with a first one of the X stripes, data blocks 2, X+2, 2*X+2, . . . etc., are associated with a second one of the X stripes, etc., and wherein individual ones of the X stripes are each stored on a different one of the plurality of mass storage units. The plurality of mass storage units preferably store a single copy of a digital representation of a video presentation. The X stripes are read out in such a manner as to enable a plurality of data streams to simultaneously convey a same one of the N data blocks, or are read out in such a manner as to enable a plurality of data streams to simultaneously convey a different one of the N data blocks.

3. Claim set — there is exactly one independent claim (claim 1); claims 2–4 depend from it

Claim 1 — the sole independent claim (plain language)

A "media streamer" (a video server) system, comprising five cooperating elements:

1. Storage node with striped mass storage. At least one storage node holds a digital video presentation in mass storage made up of multiple storage units. The presentation takes time T to play and is stored as N blocks, each block holding roughly a T/N slice of the video.
2. Switch means coupled to the storage node (a switch fabric — the spec's low-latency crossbar).
3. Communication nodes. Multiple communication nodes, each with (a) at least one input port coupled via the switch to the storage node output for receiving and buffering the video's data blocks, and (b) multiple output ports, each output port transmitting an isochronous digital representation as a data stream to a consumer.
4. Round-robin striping. The N blocks are partitioned into X stripes, where stripe 1 = blocks 1, X+1, 2X+1, …; stripe 2 = blocks 2, X+2, 2X+2, …; etc., with each stripe stored on a different mass storage unit.
5. Control means. Control logic operates the switch in accordance with consumer inputs to route data blocks to selected communication nodes, such that the stripes are read out so a plurality of streams can simultaneously convey the same data block, or different data blocks, to each of a plurality of communication nodes.

In plain terms: split one movie into N chunks, stripe those chunks across a set of disks, and put a switch in front of a pool of playback nodes so that adding disks (not duplicating content) is what lets you serve more viewers — from a single copy — either all watching the same chunk or each watching a different chunk.

Dependent claims

• Claim 2 (depends from 1): the T/N period is approximately 0.2 s to 2 s — i.e., each stored block corresponds to a fraction of a second to ~2 seconds of playback.
• Claim 3 (depends from 1): X is determined by an expression whose variables are r (nominal stream data rate), n (max simultaneously output streams at that rate), d (effective output rate of one mass storage unit), and m (max simultaneously output streams at that nominal rate from all mass storage units storing any part of the video).
• Claim 4 (depends from 3): combines claims 2 and 3 — the X expression plus the 0.2–2 s T/N window.

4. Contradictions found against previously generated sections — flagged per instruction

1. Claim-text availability (material contradiction). The Obviousness section states the claims "were not retrievable from the fetched record" and proceeds on a "reconstructed" claim that "should be re-verified." That premise is now superseded: the full claims 1–4 are available and verified in this analysis (USPTO-derived record at uspto.report/patent/grant/5712976, corroborated by EveryPatent and Justia). The verbatim claim 1 differs from the reconstruction in specific ways a litigator needs: claim 1 expressly recites "switch means" and "control means" (means-plus-function terms), requires the input ports to be coupled "via said switch means," requires "receiving and buffering", and requires isochronous transmission to "each of a plurality of communication nodes." The Prior Art section's four-claim summary is correct and should be treated as controlling; the Obviousness section's "reconstructed claim" caveat should be deleted.

2. Claim 3's expression — the formula was missing from the summary of the invention. The specification's summary recites the X expression without printing it. EveryPatent renders claim 3 verbatim as "X = maximum (r*n/d, r*m/d)". This matters for the § 103 analysis: the Obviousness section treated element (e) as a restatement of "n·r ≤ X·d" only. The literal claim takes the maximum of two ratios — the per-video stream count (n) and the all-videos-on-those-disks stream count (m) — which is exactly the "consider other content on the same disks" teaching the '976 specification gives in its worked example (s = ceiling(r·n/d) rises from 3 to 4 stripes when m = 45). The claim is therefore slightly narrower than the summary-based reconstruction assumed. (Minor caveat: the "X = maximum (…)" string is a third-party OCR rendering of a mathematical expression; verify the literal claim text in USPTO Patent Center before quoting it.)

3. Prior-art reference list — confirmed, with one addition to note. The Prior Art section's list of 26 U.S. references, 2 foreign documents, and 6 NPL items is corroborated by the fetched record. Two clarifications: the near-miss reference the Prior Art section flagged as "US 5,526,507 (Hill)" correctly matches the record's "5526507 (June 1996, Hill)" — no correction needed; and the two purely § 102(a)-type references (EP 0 529 864 A1, WO 93/16557) remain unverified as to content, as that section already cautioned.

4. No contradiction with the Litigation, PTAB, or Assignment sections. My CAFC 2026 docket search independently returns the same negative, and the assignment record remains a single inventors→IBM conveyance.

5. Analyst notes and uncertainty disclosures

• Means-plus-function exposure. For a patent filed 1994-09-08, "switch means" and "control means" in claim 1 are governed by pre-AIA § 112 ¶ 6 (now § 112(f)) and are to be construed as the corresponding structure disclosed in the specification and its equivalents. The disclosed structures are the low-latency crossbar switch (12) with crossbar switch cards 20 and packet-header routing, and the control-node/communication-node control hierarchy (control node 18, communication-node threads, the scheduler 104, and the request queues of FIGS. 3–4). No structure is named in the claim itself. This is a scope-limiting fact for any infringement theory and a natural indefiniteness/§ 112 attack vector — and it dovetails with the Federal Circuit's 2026 § 112(f) activity (e.g., TrackTime v. Amazon, Fed. Cir., July 2, 2026, remanding for a Dyfan-based § 112(f) analysis; see the CAFC 2026 materials returned in my search).
• Claims never adjudicated. No IPR/PGR/CBM was ever filed and no district-court or Federal Circuit case naming this patent was found, so claims 1–4 remain exactly as issued, untested, and expired (2015-01-27, Expired – Fee Related).
• Search-source limitation (explicit). I do not have live API access to USPTO Patent Center, the Open Data Portal, PTAB E2E, or an assignment-center abstract of title from this environment. The bibliographic data above is drawn from USPTO-derived third-party mirrors (uspto.report, EveryPatent), Google Patents' legal-event feed, FreePatentsOnline, and Justia, which agree with one another. My CAFC 2026 check was web-search based, not a direct CM/ECF or PACER docket query. I therefore cannot state a categorical negative as to every historical filing; I can state that no evidence of any 2026 — or other — CAFC activity involving 5712976 exists in any source I could reach, and that the patent's 2015 expiration makes 2026 appellate activity implausible. A definitive negative would require a paid PACER/CAFC docket query and a Patent Center lookup by patent number 5712976 and application 08/302,624.
• Family mapping caution. US 5,761,417 ("Video data streamer having scheduler …"), US 5,805,821 ("Video optimized media streamer user interface employing non-blocking switching …"), US 5,668,948, US 5,603,058, and US 5,586,264 all appear to belong to this same 1994-09-08 IBM media-streamer family (several share inventor Saxena). I have not verified which sibling corresponds to which of the five cross-referenced application numbers, so treat those mappings as unverified.

The initial searches returned no results. Let me try different search formulations.

Litigation Search Results for U.S. Patent 5,712,976

Based on the searches I was able to run, I found no known litigation involving U.S. Patent 5,712,976 (the IBM "Video data streamer" patent).

What I searched

• Unified Patents litigation portal (portal.unifiedpatents.com) — search for "5712976" returned no results.
• CourtListener / PACER-derived dockets — no docket entries referencing patent number 5712976 were found.
• CAFC docket searches — no Federal Circuit appeal involving this patent was found (consistent with the earlier analysis in this project: the patent's anticipated expiration was January 27, 2015, and it is recorded as Expired – Fee Related).
• General web searches for combinations of "5712976," "5,712,976," "video data streamer," and "IBM patent lawsuit" — no litigation hits.

Near-miss (excluded, not the same patent)

One search surfaced Broadcom Corporation v. Netflix, Inc. (N.D. Cal., Case No. 3:20-cv-04677), which references "the '976 patent." That case is not this patent: the claims at issue there are Broadcom/Avago claims directed to connecting to web services and server load balancing (see the CourtListener/Casetext excerpts), which has nothing to do with IBM's video data streamer, striping, or isochronous video delivery. That "976" patent is a different, similarly numbered patent, and per your instruction I am excluding it rather than treating it as a match.

Conclusion

Caveat on confidence: My searches were web-based (CourtListener, Unified Patents, general web/Google Patents). I did not have direct access to the full USPTO Patent Center assignment/ litigation data or a comprehensive PACER full-text search across all district courts, so I cannot categorically prove that no historical (pre-2015) district-court case ever existed. However, all available evidence — including the patent's 2015 expiration and the absence of any docket or CAFC hits — points to no known litigation. If you need a definitive negative, the next step would be a paid PACER/USPTO litigation-search query across all district courts for the patent number 5712976.

Proceedings overview

Zero (0) AIA trial proceedings are on file for U.S. Patent 5,712,976 — no active trials, no claims invalidated, no claims sustained, no settlements, and no institution denials, because no IPR, PGR, or CBM petition was ever filed against this patent. The USPTO Open Data Portal (ODP) API returns no AIA trial proceedings for this patent as of the most recent ingest, and my web searches (PTAB decision databases, Unified Patents, CourtListener, general web) corroborated that negative — every "976 patent" hit surfaced (e.g., SAS Institute v. Iancu on the '936 patent; Broadcom v. Netflix on an unrelated Avago '976 patent; Apple v. Universal Secure Registry on the '539 patent) is a different patent. Defensive posture: neutral-to-favorable in the sense that no claim has ever been tested or narrowed before the PTAB — but the patent expired on 2015-01-27 (Expired – Fee Related), so a defendant's leverage comes from expiration, the pre-2015 damages window, and unrestricted prior-art defenses in district court, not from any IPR scalp.

Proceedings (none to report)

There are no individual proceedings to list. Per the canonical ODP block in this prompt, the USPTO Open Data Portal API returns no AIA trial proceedings for US 5712976 as of the most recent ingest. Web search confirmed the absence:

• Searches for "5712976" IPR, "5,712,976" "final written decision", "5712976" site:uspto.report, and IPR2014/2015/2016 "5712976" returned no PTAB proceeding numbers tied to this patent.
• The only 5712976-related web hits were the patent's own specification text on uspto.report and third-party patents citing the IBM media-streamer family (e.g., US 5,586,264, CA2153445C, US 8,609,085, US 8,189,819) as prior art — not PTAB activity.
• No PTAB Final Written Decision, institution decision, or Federal Circuit appeal referencing this patent was found.

Caveat on confidence: The ODP negative is the authoritative signal and my searches corroborate it. I cannot categorically prove that a pre-2015 historical proceeding never existed from web-accessible sources alone; if a definitive negative is needed, a paid PTAB E2E / Patent Center query for patent number 5712976 would close the loop. But every available indicator points to no PTAB activity, ever.

Strategic summary

Claims status — all UNTESTED. Because no AIA proceeding was ever filed, no claim of 5712976 has been canceled, sustained, or even reached an institution decision. All claims remain exactly as issued, and none carries any PTAB-derived patentability gloss. For a defendant facing assertion today, the more decisive facts are (1) the patent's anticipated expiration on 2015-01-27 and its Expired – Fee Related status — meaning damages are confined to pre-expiration conduct and no injunctive relief exists — and (2) the complete absence of any § 315(e)(2) estoppel cloud: no prior-art ground has been consumed by an IPR, so every § 102/§ 103 theory remains available to any defendant who was not itself the petitioner in a (nonexistent) proceeding. The only real procedural trap is § 315(b): any defendant who was served with a complaint more than a year ago is time-barred from filing an IPR — but with the patent expired since 2015, an IPR's practical value (and the one-year bar's sting) is largely theoretical anyway.

Estoppel landscape. Vacant. § 315(e)(2) estoppel binds only IPR petitioners and their privies as to grounds raised or reasonably available in a filed and instituted proceeding. With zero proceedings, there is no estoppel against anyone. A defendant sued today can raise any § 102/§ 103 combination, any § 112 issue, and any eligibility/indefiniteness defense in district court without PTAB-estoppel constraint. Note the corollary: the patent owner likewise holds no PTAB "wins" to brandish — no claim has been confirmed by the Board, so there is no administrative-validation argument for the patentee to lean on.

Pattern signals. None. There is no repeat petitioner, no patent-owner litigation pattern before the Board, and no defensive aggregator (e.g., Unified Patents) in the chain — because there is no chain. The absence of any IPR against a prominent IBM video-server patent is itself mildly notable: the patent issued in 1998 and was assertable for 17 years, yet no one ever challenged it at the PTAB. That is consistent with the patent being cited but rarely (if ever) asserted in litigation — the earlier litigation search in this project found no district-court or CAFC cases involving 5712976. One family-level flag worth checking: the same 1994-09-08 priority date and IBM inventorship carry over to US 5,586,264 ("Video optimized media streamer with cache management") and CA2153445C ("Video optimized media streamer user interface"). If the demand letter cites those siblings rather than 5712976 itself, run the same ODP/PTAB check against those numbers — I did not verify their PTAB status in this analysis.

Recommended next steps

1. Confirm the negative on the record. Run a PTAB E2E / Patent Center search for patent number 5712976 (and application 08/302,624) so the "no proceedings" finding is documented in the file, and pull the USPTO assignment database to identify the current owner — Google Patents lists IBM as assignee, but title may have moved, and knowing who is actually asserting matters more than the PTAB record.

2. Anchor the defense on expiration, not IPR. The patent expired 2015-01-27. Any infringement claim is limited to acts before that date (and, if no litigation was ever filed, likely faces laches/limitations issues of its own). Quote the Google Patents legal-status line — "Expired – Fee Related," anticipated expiration 2015-01-27 — and the pre-AIA damages framework. There is no FWD to link because none exists; do not represent otherwise to the court or opposing counsel.

3. Preserve all prior-art grounds. With zero IPRs, § 315(e)(2) estoppel does not apply. If the case is worth litigating, run a full § 102/§ 103 search (the specification itself concedes the field was crowded — VTR/tape systems, RAID, DVI/JPEG/MPEG, ATM — so 1990s video-server art is plentiful). Just be mindful of the § 315(b) one-year clock if a complaint has already been served: the window to file an IPR may be closed even though the patent is expired, so plan on district-court validity defenses rather than a PTAB challenge.

4. If the demand letter cites a family member, repeat the ODP/PTAB check for US 5,586,264 and CA2153445C before assuming the same clean record — and confirm whether the asserted claims are method claims, system claims, or the video-adapter claims, since the truncated fetched text did not include the full claim set. The claims of 5712976 have never been PTAB-tested; treat them as "untested, expired, and assertable only for pre-2015 conduct" — which is the accurate, and for a defendant a fairly comfortable, posture.

Inventors

Pattern note: No unusual pattern. Both inventors assigned their interest to IBM at filing in the normal course. I found no evidence (and no basis in the record) that either inventor departed IBM within 12 months of filing, and there is no portfolio fire-sale signature — the assignment record is a routine employer assignment.

Original assignee

• Entity: International Business Machines Corporation (Armonk, NY) — the entity named on the issued patent and the assignee of the only recorded assignment.
• Line of business: Operating company — diversified computer hardware, software, and services.
• Product embodiment: The patent's specification describes the "media streamer" / video-optimized stream server, and it is one of a same-day IBM family (US 5,586,264; US 5,603,058; US 5,668,948) covering the same architecture. IBM did market video-server products in this era (the IBM media streamer / VideoCharger line grew out of this architecture), but I could not verify from my web-accessible sources that a specific shipped product embodied these exact claims. Treat "product shipped" as probable but unverified.
• Current status of the company: Operating (one of the world's largest technology companies). Current status of the patent: Expired – Fee Related, anticipated expiration January 27, 2015, per Google Patents legal-status data.

Assignment timeline

I was unable to run a direct query of the USPTO Assignment Center (assignmentcenter.uspto.gov) from this environment, and no web-accessible source surfaced the reel/frame or correspondent for this patent's assignment. I will not fabricate those fields. What follows is what the record affirmatively shows.

• ~1994-09 (executed, with filing 1994-09-08) / recorded 1994-11-09 — Reel/frame not retrievable from sources searched (do not treat as absent — treat as unverified)
  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Falcon, Lorenzo, Jr.; Saxena, Ashok Raj
  • Assignee: International Business Machines Corporation
  • Correspondent: not retrievable from sources searched
  • Context: Routine employer assignment from the two inventors to IBM at filing (this is the only legal event of record per Google Patents' legal-event feed for this patent).

No post-issuance assignments, security agreements, mergers, name changes, or licenses were found. Google Patents' legal-event feed for US5712976A shows only the 1994-11-09 assignment to IBM; the patent then remained in IBM's hands until its fee-related expiration on 2015-01-27. If the USPTO Assignment Center confirms this, it means IBM — an operating company — owned the patent for its entire life and no NPE ever entered the chain.

Caveat: Because I could not pull the Assignment Center abstract of title directly, I cannot categorically rule out a later recorded conveyance that Google Patents' feed does not surface. The affirmative evidence, however, points entirely to continuous IBM ownership.

Timeline diagram

timeline
    title Ownership of US 5712976
    1994 : Filed by IBM inventors
         : Assigned to IBM
    1998 : Patent issued
    2015 : Expired fee related

NPE / troll-pattern signals

1. Shell-entity transfer — not present. No transfer from IBM to any LLC / IP / Licensing / Holdings entity appears anywhere in the record. The patent never left IBM.
2. Known asserter in the chain — not present. No Acacia, Marathon, Intellectual Ventures, Wi-LAN, Conversant, or any Unified Patents / RPX-listed asserter appears as assignee or assignor. The only assignee of record is IBM.
3. Repeat correspondent across the chain — not present / not assessable. There is only one recorded assignment, and its correspondent could not be retrieved from my sources. Recurrence is impossible with a single link, so this signal cannot be scored.
4. Cascading transfers — not present. No chained LLC transfers exist in any time window.
5. Pre-litigation transfer — not present. No infringement suit naming this patent was found (see prior litigation summary), and no transfer exists to be timed against one.
6. Bankruptcy fire-sale — not present. IBM has not been in Chapter 7/11, and no bankruptcy-proceeding sale of this patent appears.
7. Privateering — not present. No operating-company-to-NPE transfer exists, and no SEC-filing or EFF/Patent Progress coverage of such an arrangement for this patent was found.
8. Defensive aggregator (anti-NPE) — not present. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. It terminates at IBM / natural expiration. (This is the inverse of an NPE signal, but it is not a defensive-aggregation outcome either.)

Verdict

Insufficient data — the record contains only the original inventors-to-IBM assignment (recorded 1994-11-09), and I could not retrieve the reel/frame or correspondent from web-accessible sources to complete the abstract of title.

That said, the available evidence is affirmatively clean: every source (Google Patents legal events, legal-status "Expired – Fee Related" as of 2015-01-27, and the absence of any litigation) indicates IBM, a large operating company, owned this patent for its entire life and never transferred it to any NPE. No shell LLC, no known asserter, no cascading transfers, and no privateering appear anywhere in the record. The "insufficient data" verdict is driven strictly by the definition (only the original assignment is on record) and by my inability to run a direct Assignment Center query — not by any affirmative NPE signal, of which there are none.

For verification, query the USPTO Assignment Center by patent number 5712976: https://assignmentcenter.uspto.gov/ (search "Patent Number" = 5712976); the legacy mirror is https://assignment.uspto.gov/patent/index.html.

I'll search for the patent's citation records and claim text. Let me run several targeted queries.

I have partial data. Let me pull the specific citation record for 5712976 and verify each reference's dates.

Let me try to obtain the actual claim text and verify key filing dates.

U.S. Patent 5,712,976 — Prior-Art Citation Analysis (§102)

A. Patent identification (search result — confirmed, not a similar number)

The search resolved to exactly one record: US 5,712,976 A, "Video data streamer for simultaneously conveying same one or different ones of data blocks stored in storage node to each of plurality of communication nodes," application 08/302,624, filed 1994-09-08, issued 1998-01-27, IBM, inventors Lorenzo Falcon, Jr. and Ashok Raj Saxena. Sources: https://patents.google.com/patent/[US5712976A](/patent/US5712976A), https://patents.justia.com/patent/5712976, https://www.freepatentsonline.com/5712976.html, https://uspto.report/patent/grant/5712976.

Near-miss numbers explicitly excluded (they surfaced in the same searches and are different patents): US 5,812,976 (Matsushita transport-decoder interface), US 5,715,750 (Goovaard), US 5,713,002/5,713,004, and EP 0529864 A1 (a separate prior-art document, not this patent). I have not auto-corrected any of these.

B. Correction to the previously generated sections — the claims ARE now available

The earlier sections stated the claim text was truncated and reconstructed a "likely independent claim" from the specification. That reconstruction is contradicted by the actual claim, and I flag it per the cross-reference rule:

Verified claims (source: FreePatentsOnline/Justia, both quoting the printed patent):

• Claim 1 (independent): storage node whose mass storage comprises a plurality of mass storage units storing a video presentation of duration T as N data blocks each ≈ T/N; switch means; a plurality of communication nodes, each with ≥1 input port coupled via the switch means to the storage-node output for receiving and buffering data blocks, and each with a plurality of output ports transmitting an isochronous digital representation as a data stream to a consumer; the N blocks partitioned into X stripes round-robin (1, X+1, 2X+1… / 2, X+2, 2X+2…), each stripe on a different mass storage unit; control means operating the switch means in accord with consumer inputs to supply blocks to selected communication nodes; the X stripes read out so plural data streams can simultaneously convey a same one or different ones of the N data blocks to each of a plurality of communication nodes.
• Claim 2: T/N duration ≈ 0.2–2 seconds (dep. claim 1).
• Claim 3: X = maximum (rn/d, rm/d), where r = nominal stream rate, n = max simultaneous streams at r, d = effective output rate of one mass storage unit, m = max simultaneous streams at r from all mass storage units holding any part of the video (dep. claim 1).
• Claim 4: T/N ≈ 0.2–2 seconds (dep. claim 3).

Gap: both sources truncate after claim 4. The Japanese counterpart (JP 3,096,409 B2 / JPH08107542A) shows further claims reciting "a plurality of disk data storage units" with the X = max(rn/d, rm/d) formula — suggesting at least one more independent claim in the U.S. set. Claims 5+ are unverified; pull them from USPTO Patent Center (app. 08/302,624).

C. The "References Cited" inventory (as it appears on the face of '976)

Retrieved from the Justia and FreePatentsOnline citation records: 26 U.S. patents, 2 foreign documents, and non-patent literature. Publication dates below are as printed in those records.

U.S. patents: 4,355,324; 4,604,687; 4,616,263; 4,679,191; 4,949,187; 5,089,885; 5,133,079; 5,200,989; 5,283,639; 5,289,461; 5,421,031; 5,442,390; 5,461,415; 5,471,640; 5,473,362; 5,487,167; 5,510,905; 5,513,011; 5,522,054; 5,526,507; 5,528,513; 5,530,830; 5,544,327; 5,550,982; 5,586,264; 5,606,359.

Foreign: EP 0529864 A1 (published 1993-03-03) and WO 93/16557 (published August 1993).

Non-patent literature: Intel, "Scalable Multi-server Technology for Interactive Multimedia Applications," May 1994, pp. 3–15; Thapar & Koerner (Hewlett-Packard), "Architectures for Video Servers," 1994 NCTA Technical Papers, pp. 141–148; IBM, "ALLICAT (0664) S10 SCSI Direct Access Device Functional Specification Release 4.50," Doc. No. AS01-0003-00, formatted 1993-05-06, pp. 1–291; Hosekote et al., "Scheduling continuous media in a video-on-demand server," IEEE, 1994, pp. 19–28; Kim, "Synchronized Disk Interleaving," IEEE, Nov. 1986, pp. 978–988; Rangan et al., "Designing File Systems for Digital Video and Audio," ACM (year/page cut off in the retrieved record).

D. Legal framework applied

• '976 has a 1994-09-08 filing date → pre-AIA § 102/§ 103.
• A reference published before 1994-09-08 is unconditional § 102(a)/(b) art.
• A reference issued after 1994-09-08 (most of this list) can be prior art only under § 102(e), which requires the reference's U.S. filing date to be before the applicant's invention date (presumptively the 1994-09-08 filing date). I could verify the filing date of only one reference (Fitzgerald '362, filed 1993-11-30); the rest need a Patent Center/Google Patents pull. I have not fabricated filing dates.
• Being cited by the examiner is not itself a §102 determination.
• Same-day, commonly owned IBM family members (5,586,264 Belknap; and the non-cited siblings 5,603,058, 5,668,948, 5,761,417, CA 2,153,445C) are not § 102(e) art (not "filed before the invention") and are disqualified as § 103 art by pre-AIA § 103(c) (common ownership).

E. Reference-by-reference §102 analysis

E.1 Pre-1994-09-08 publications — unconditional § 102(a)/(b) art (highest risk)

E.2 Post-1994-09-08 issuances — § 102(e) art only if filed before the invention date

E.3 Non-patent literature

F. Bottom line — the references that actually matter

1. Most relevant single reference: US 5,473,362 (Fitzgerald) — striped data across plural storage devices with time-multiplex scheduling, filed 1993-11-30 (verified), and therefore valid § 102(e) art. It is the only cited reference that maps to most of claim 1 (E1, E2, E6, E8) and is the lead § 103 combination reference. Whether it anticipates claim 1 turns on the claim-1 elements it appears to lack — a switch means, a control means, and a plurality of communication nodes each with a plurality of output ports.
2. Most relevant pre-filing reference: EP 0529864 A1 (1993-03-03) — a network video server serving multiple clients with separate control and video channels; solid § 102(a) art for the multi-consumer/control-channel elements.
3. Claim-3 (X-formula) specialization: US 5,528,513 (Vaitzblit) — admission control on exactly the r/n/d parameter set.
4. Striping mechanics: US 5,471,640 (McBride) and US 5,530,830 (Iwasaki) for E6, plus Rangan and Kim (1986) as the foundational literature.
5. Switch/multi-node architecture: US 5,289,461 (de Nijs) and the Intel "Scalable Multi-server" paper.
6. The four unlabeled early patents (4,355,324; 4,604,687; 4,616,263; 4,679,191) are almost certainly background/signal-distribution art and are not credible § 102 references against claim 1 — but I could not retrieve their text, so I do not affirmatively clear them.

No single cited reference appears to disclose every element of claim 1 (the combination of X-striping across different mass storage units + a switch means + control means + a plurality of communication nodes with plural isochronous output ports + simultaneous same-or-different block conveyance). That is precisely why the earlier § 103 section concluded the claim is vulnerable to combination rather than clean anticipation — a conclusion the now-verified claim text supports, since the claim is narrower than the earlier reconstruction assumed (it adds "switch means," "control means," "receiving and buffering," "isochronous," and per-node plural output ports).

G. Verification gaps I am flagging rather than filling

1. Filing dates for 25 of the 26 U.S. references are unverified. Only Fitzgerald '362 (1993-11-30) was verified. Every post-1994-09-08-issuing reference depends on a § 102(e) filing date, so this is the gating fact for the entire anticipation analysis.
2. Claims 5+ of '976 were not retrieved; my claim mapping covers claims 1–4 only.
3. Titles/abstracts for US 4,355,324, 4,604,687, 4,616,263, 4,679,191 and WO 93/16557 were not retrieved — I did not guess them.
4. Rangan et al. page/year was truncated in the Justia record.
5. Reference full texts were not read. All anticipation conclusions above are preliminary, derived from titles, citation position, and the examiner's §102(e) citation, and must be confirmed against the full documents and their claim charts.

I need the complete list of prior art references cited on the face of US 5,712,976. Let me pull the full "References Cited" section from the patent record.

I have the key cited references. Let me try to get the complete citation list including foreign references and non-patent literature, and check the filing dates of the most relevant references for prior-art status.

Obviousness Analysis Under 35 U.S.C. § 103 — U.S. Patent 5,712,976

0. Preliminary caveats (read first)

Claim text. As documented in the prior sections of this file, the verbatim claims of US 5,712,976 were not retrievable from the fetched record (the Google Patents text truncates before the claims). This analysis therefore proceeds on the reconstructed independent claim derived from the specification's Summary of the Invention. Every element mapping below should be re-verified against the actual claim language (available from USPTO Patent Center, application 08/302,624) before being used in litigation.

Statutory framework. The application was filed September 8, 1994 — before the March 16, 2013 AIA transition — so pre-AIA 35 U.S.C. § 103 governs, applying the Graham v. John Deere, 383 U.S. 1 (1966) factors as refined by KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007).

Prior-art status of cited references. The references on the face of the patent were "considered" by the examiner; that does not by itself make each one § 102 prior art. Under pre-AIA § 102(e), a U.S. patent is prior art as of its earliest effective U.S. filing date. I verified the key dates where noted; several cited references have unknown filing dates and should be confirmed before reliance. The primary combination below rests on verified prior art.

1. The prior art record (from the patent's citation records)

The patent's own "References Cited" (FreePatentsOnline / Google Patents) and the specification cite the following, which constitute the available prior-art pool:

U.S. patents cited on the face of '976:

• US 5,473,362 (Fitzgerald et al., Microsoft) — Video on demand system comprising stripped data across plural storable devices with time multiplex scheduling — filed Nov. 30, 1993 (verified); § 102(e) prior art against the Sept. 8, 1994 filing.
• US 5,606,359 (Youden) — VOD with multiple data sources providing VCR-like services
• US 5,550,982 (Long) — Video application server
• US 5,544,327 (Dan) — Load balancing in VOD servers by buffer allocation
• US 5,528,513 (Vaitzblit) — Scheduling and admission control policy for a continuous media server
• US 5,530,830 (Iwasaki) — Disk array system
• US 5,522,054 (Gunlock) — Dynamic control of outstanding disk read requests
• US 5,510,905 (Birk) — Video storage server using track-pairing
• US 5,487,167 (Dinallo) — PC with generalized data streaming apparatus
• US 5,471,640 (McBride) — Programmable disk array controller with counters for striping data
• US 5,461,415 (Wolf) — Look-ahead scheduling for VOD
• US 5,442,390 (Hooper) — VOD with memory accessing
• US 5,421,031 (De Bey), US 5,289,461 (de Nijs), US 5,283,639 (Esch), US 5,200,989 (Milone), US 5,133,079 (Ballantyne), US 5,089,885 (Clark), US 4,949,187 (Cohen) — distributed/multimedia distribution systems
• US 5,586,264 (Belknap) — Video optimized media streamer with cache management — sibling application, same filing date (Sept. 8, 1994), commonly owned; excluded as § 103 prior art under pre-AIA § 103(c) (same-day, common-assignee; I do not rely on it).

Non-patent literature cited in the specification:

• Patterson, Gibson & Katz, "A Case for Redundant Arrays of Inexpensive Disks (RAID)," ACM SIGMOD, Chicago, June 1–3, 1988, pp. 109–116 (verified) — the foundational RAID striping paper, expressly incorporated into the '976 specification's own discussion of disk storage.

Uncertain dates to verify before reliance: Vaitzblit '513, Dan '327, Wolf '415, Long '982, Youden '359, Hooper '390, McBride '640 all issued after Sept. 8, 1994; each is usable only if its underlying application was filed before the invention date (likely for several, since the VOD art of 1993–1994 was active, but unverified here).

2. The reconstructed independent claim — element breakdown

Based on the specification's Summary, the principal system claim requires:

3. Person of ordinary skill in the art (POSITA)

A POSITA circa 1993–1994 would be an engineer with a B.S./M.S. in computer science or electrical engineering and roughly 2–5 years' experience in digital video-on-demand server design, disk-array/RAID storage systems, continuous-media I/O scheduling, and multimedia networking (SCSI, ATM, MPEG), familiar with the contemporaneous literature (RAID/SIGMOD, IEEE COMPCON multimedia sessions, the Microsoft/Bellcore/Oracle VOD server efforts). Such a person would routinely perform bandwidth engineering — computing how many disks are required to sustain N isochronous streams at a given bit rate.

4. Primary obviousness combinations

Combination A (strongest): Fitzgerald '362 + RAID (Patterson 1988) — striped single-copy VOD server

Fitzgerald '362 (verified § 102(e) prior art) discloses nearly the entire claimed system:

• Element (a) — plurality of storage units storing a video as blocks: Fitzgerald's VOD system has "a plurality of storage devices" (explicitly including magnetic disk drives) storing "data sequences" (video image sequences) as "blocks of data." ✓
• Element (c) — round-robin striping, each stripe on a different device: Fitzgerald: "each consecutive block of data for the video image sequence is stored on a next storage device in the predetermined sequence… The predetermined sequence wraps around to the beginning." This is exactly the claim's "blocks 1, X+1, 2X+1…" round-robin partition. Each block of a sequence resides on a different device (the default is one copy, with multiple copies optional). ✓
• Element (d) — simultaneous conveyance of different blocks: Fitzgerald's "bandwidth unit" model schedules each subscriber to a like-positioned time slot across the devices; during each consecutive slot a consecutive block is read from the appropriate device, so different subscribers simultaneously receive different blocks from the same single copy. ✓
• Element (d) — simultaneous conveyance of the same block: In Fitzgerald's model, subscribers assigned to the same bandwidth unit receive the same block stream; a block read once from a device during its slot can be delivered to all subscribers sharing that unit. The "same block to multiple streams" is thus inherent in (or trivially derived from) the time-multiplexed schedule. ✓
• Element (e) — stripe count as a function of bandwidth: Fitzgerald's entire scheduling architecture treats device I/O bandwidth as the limiting resource, allocated in bandwidth units sized to the stream data rate. The notion that the number of devices (stripes) must satisfy aggregate device bandwidth ≥ (number of streams) × (stream rate) is inherent in the model; the '976 formula (≈ n·r/d) is simply that constraint written down. ✓

RAID (Patterson 1988) supplies the general motivation and mechanism for striping data across "a plurality of disks" specifically "to improve bandwidth" and "scalability" — precisely the reason '976 gives for striping ("concurrency or bandwidth reasons," not parity). The '976 specification itself cites RAID as background, conceding the concept's presence in the art. McBride '640 (programmable disk array controller with counters feeding "every nth data segment" to each disk) supplies the round-robin controller implementation if a hardware limitation were argued.

Motivation to combine: RAID and Fitzgerald are in the same field (high-bandwidth disk storage; VOD servers). A POSITA implementing Fitzgerald's striped VOD storage would use the standard disk-array striping techniques of RAID/McBride to realize the striping hardware. This is a textbook combination of known elements with predictable results (KSR).

Combination B: Fitzgerald '362 + continuous-media admission-control art (Vaitzblit '513, Dan '327, Wolf '415) — the stripe-count formula and scheduling

If the examiner/patentee points to element (e) (the X-formula) as the point of novelty, the continuous-media server literature fills it:

• Vaitzblit '513 teaches "scheduling and admission control policy for a continuous media server": before admitting a stream, the server verifies sufficient disk bandwidth and buffer capacity; the sustainable number of concurrent streams at a given data rate is a function of disk count and per-disk effective throughput — the exact parameters (r, n, d, m) of the '976 expression.
• Dan '327 teaches allocating buffers to streams based on stream requirements — resource provisioning per stream at connect time, the same operation '976 performs ("calculations of sfc and dfc are made at connect time").
• Wolf '415 teaches look-ahead scheduling of disk reads to meet isochronous output deadlines — the "just-in-time" read scheduling of the '976 storage/communication nodes.

Motivation to combine: The problem the '976 formula solves — how many disks do I need to support n simultaneous streams at rate r? — is the central capacity-planning question of every continuous-media server. Vaitzblit/Dan/Wolf all address it from the admission-control/scheduling side. Writing the constraint as a simple expression of r, n, d, m is an "obvious to try" algebraic exercise with a predictable result (KSR; see also the '976 spec's own admission that the stripe count is "calculated from the formula m·r/d"). A POSITA combining Fitzgerald's striped layout with Vaitzblit's admission control would necessarily arrive at the claimed stripe-count determination.

Combination C: Fitzgerald '362 + distributed-server architecture art (Youden '359, Long '982, Hooper '390, de Nijs '461, Esch '363, Ballantyne '079) — the multi-node communication architecture

Element (b) — "communication nodes each having an input port coupled to the storage node output and plural output ports transmitting data streams" — is the conventional distributed-server decomposition, taught by:

• Youden '359 — VOD system with multiple data sources configured to provide VCR-like services to multiple outputs;
• Long '982 — a video application server with storage feeding networked video outputs;
• Hooper '390 — VOD with memory accessing, i.e., shared memory so multiple users access the same video data (also supporting element (d)'s same-block-to-many-streams);
• de Nijs '461 / Esch '363 — interconnection/switching methods for digital multimedia communications, i.e., a switch fabric between multimedia nodes;
• Ballantyne '079 — central movie-distribution source serving many user terminals.

Motivation to combine: Scaling a VOD server beyond one machine's bandwidth by separating storage nodes from communication/output nodes over a switch is the standard distributed-server design of the era (the '976 spec concedes the field was crowded with VOD, ATM, and multimedia-server work). A POSITA would combine Fitzgerald's striped storage with a switch-based node architecture to achieve the claimed scalability — again a predictable aggregation of known components.

5. Graham-factor and KSR analysis

1. Scope and content of the prior art. The prior art (Fitzgerald, RAID, Vaitzblit, Dan, Wolf, Youden, Long, Hooper, de Nijs) collectively teaches every element of the reconstructed independent claim: striped single-copy storage of video blocks across multiple disks (Fitzgerald + RAID), time-scheduled simultaneous delivery of different blocks to different streams and the same block to multiple streams (Fitzgerald + Hooper), multi-node server architecture with a switch (Youden/de Nijs/Esch), and admission-control/bandwidth-based stripe-count calculation (Vaitzblit/Dan/Wolf).

2. Differences between the claim and the prior art. The principal arguable differences are (i) the express multi-node "storage node/communication node" partitioning and (ii) the closed-form X-expression. Both are (i) a standard design decomposition and (ii) a routine algebraic restatement of the known bandwidth constraint — not structural or functional surprises.

3. Level of ordinary skill. High — VOD server design was an active, crowded field in 1993–1994 (Microsoft's Tiger, Bellcore, Oracle, IBM's own media-streamer family all being contemporaneous).

4. Secondary considerations. None found. The patent expired (Jan. 27, 2015) with no known litigation, no PTAB challenge, and no evidence of commercial success, long-felt need, or copying (see the litigation/PTAB sections of this file). The absence of any assertion history cuts against a non-obviousness showing.

KSR points: This is the paradigmatic KSR case — a combination of known prior-art elements (disk striping, time-slot scheduling, admission control, node-based server architecture), each performing its known function, with a predictable result (more concurrent streams by adding disks), and no teaching away in any reference. The "single copy, multiple stream" goal is expressly the design objective of Fitzgerald's bandwidth-unit scheduling, not a novel insight of '976.

6. Anticipated patentee counterarguments and rebuttals

7. Bottom line

The reconstructed independent claim of US 5,712,976 is highly likely obvious under pre-AIA § 103. The strongest combination is Fitzgerald '362 (verified Nov. 30, 1993 filing; striped single-copy VOD storage with time-multiplexed simultaneous stream delivery) + RAID/Patterson 1988 and McBride '640 (stripe implementation) + Vaitzblit '513 / Dan '327 / Wolf '415 (admission-control-based stripe-count determination and isochronous scheduling) + Youden '359 / de Nijs '461 (multi-node switch architecture). A POSITA had every motivation to combine these — the same field, complementary teachings, a known scaling problem, and predictable results — and no secondary considerations offset that conclusion.

Two cautions before reliance in any proceeding: (1) verify the verbatim claims (the analysis is keyed to the reconstructed summary-based claim); (2) confirm the § 102(e) filing dates of Vaitzblit '513, Dan '327, Wolf '415, Long '982, Youden '359, Hooper '390, and McBride '640 — if any filed after Sept. 8, 1994, drop it from the combination and rely on Fitzgerald '362 + RAID (verified) plus the verified pre-filing art (Ballantyne '079, Milone '989, Esch '363, de Nijs '461, Clark '885, Cohen '187), which is still a strong case for elements (a)–(d), with element (e) supported by the inherent bandwidth-constraint reasoning.