US 7046636

I'll run targeted searches for this patent number before summarizing.

Let me run one more check specifically for litigation/docket activity and the current legal status.

US Patent 7,046,636 — Analyst Summary

Search scope note: I searched Google Patents, USPTO-adjacent aggregators (uspto.report, FreePatentsOnline, Justia), and Federal Circuit / district-court docket sources. No litigation, IPR, or CAFC 2026 appeal involving US 7,046,636 was found. The CAFC and E.D. Tex. hits returned in my searches involve other patents (e.g., Hafeman v. Google, Enhanced Data Streaming LLC v. Cisco Systems, 2:25-cv-01038-JRG) and do not reference this patent. Absence of a hit is not affirmative proof of no litigation — I simply found no record.

Bibliographic data

Status caveat: Google Patents also lists an "Adjusted expiration" of 2024-06-08 (a computed full-term/patent-term-adjustment figure). That date does not mean the patent is in force. The recorded legal events show the 11.5-year maintenance fee was not paid and the patent expired effective 2018-05-16 under 37 C.F.R. § 1.362. Treat the patent as lapsed and unenforceable absent revival.

Identifier caveat (strictly literal): a JP 7046636 B2 exists — an unrelated Japanese patent (NTT / Keio University, "signal analysis device," filed 2018-02-16). It is not related to this US patent. Other "7046636" hits I saw (a Lithuanian registry notice, a wildlife-survey table row, a patents-review inventor ID) are also unrelated and should not be conflated.

Abstract (quoting the patent)

A method for adaptively improving voice quality during a communication session includes monitoring voice quality parameters associated with a communication session at one or more network components of a communication network. At least one of the network components is selected to perform packet replication, if a value of the voice quality parameters deviates from a predetermined range. In accordance with a particular embodiment, replicated packets are transmitted from the selected network component along an alternate communication path. In accordance with another embodiment, an improvement in performance associated with transmission of the replicated communication packets along the alternate communication path is assessed. Transmission of the replicated communication packets may be discontinued if the improvement is below a predetermined minimum.

Plain-language overview of each independent claim

Claim 1 — Method: monitor → replicate → switch over.
Watch voice-quality parameters at network components on the primary path. If a value strays outside a predetermined range, pick at least one component to replicate packets. Send the replicated packets down an alternate path, then stop sending on the primary path and continue on the alternate path. Note: this claim requires an actual cut-over to the alternate path, not mere duplication.

Claim 2 — Method: replication plus bandwidth reservation.
Same monitoring/selection/replication steps as claim 1 (replicating onto an alternate path), but instead of switching over it adds reserving bandwidth along the alternate path if that bandwidth is available (the spec points to RSVP as the example).

Claim 4 — Method: periodic monitoring + dynamic path selection.
Same up-through-replication steps as claim 1, with two added limitations: parameters are monitored periodically throughout the session, and the alternate path is dynamically selected so as to achieve voice quality within the predetermined range throughout the session.

Claim 5 — Method: endpoint-first detection, then escalate upstream (the core "find the problem" claim).
Receive communication packets of a session between two endpoints traveling a primary path. Monitor a performance parameter at one of the two endpoints; compare endpoint values against a predetermined threshold. Only if the endpoint measurement is below threshold, begin monitoring the same parameter at one or more of the network components, and then select which component(s) should perform replication onto an alternate path based on the network-component measurement. In other words, cheap endpoint-side detection triggers expensive in-network probing to localize and bypass the degraded node.

Claim 20 — System.
A system with multiple network components that receive/transmit session packets, plus a voice quality system that receives and analyzes voice-quality parameters measured at one or more of those components. The voice quality system includes a processor that analyzes the parameters and selects at least one component to perform packet replication when the parameters deviate from a predetermined range.

Claim 24 — Logic encoded in media (software counterpart of claim 5).
Same end-to-end steps as claim 5 (endpoint monitoring → threshold comparison → conditional monitoring at network component → selection for replication onto an alternate path), but recited as logic stored on media.

Claim 29 — Logic encoded in media (software counterpart of claim 4/1 hybrid).
Monitor voice-quality parameters at network components; select at least one component to replicate when a value deviates from a predetermined range; transmit replicated packets along an alternate path; and the alternate path is dynamically selected to achieve voice quality within the range throughout the session.

Claim 31 — Apparatus, means-plus-function (claim 5 in §112 ¶6 style).
"A means for…" recitations mirroring claim 5: means for receiving packets, means for monitoring at an endpoint, means for comparing to a threshold, means for monitoring at a network component if the endpoint value deviates from the threshold, and means for selecting a component for replication onto an alternate path.

Claim 36 — Apparatus, means-plus-function.
Means for monitoring voice-quality parameters at network components; means for selecting at least one component to replicate when a value deviates from a predetermined range; means for transmitting replicated packets along an alternate path; where the alternate path is dynamically selected to achieve voice quality within the range periodically throughout the session.

Observations worth flagging

• Claim family structure: the disclosure is drafted across four statutory buckets — process (1–4), process-with-escalation (5–19), system (20–23), software (24–30), and means-plus-function apparatus (31–37). Claims 1, 2, and 4 share nearly identical preambles and diverge only in their final limitation (switch-over / bandwidth reservation / dynamic periodic selection), which makes them a deliberately narrow graduated set.
• Drafting variance between parallel claims: claim 5 requires monitoring at the network component "if the performance parameter … is below the threshold value," while the parallel apparatus claim 31 says "deviates from the threshold value." That is a substantive difference, not a typo to normalize.
• Claim 33 anomaly: in the apparatus set, claim 33 recites "further comprising: monitoring the performance parameter … and reporting results …" without a "means for" — an apparent drafting artifact in a §112 ¶6 claim. I flag it because it affects how claim 33 would be construed; I make no representation about how a court would treat it.
• Prior art of record includes US 6,466,574 (IBM — redundant voice/media frames for QoS), US 6,741,569 (Telchemy — QoS monitor), US 6,850,601 (Cisco), US 6,512,746 (Nortel — voice grade of service in IP network), and codec/speech patents from Voxware, ANT, Conexant, and Siemens. The '574 and '746 references are the closest-adjacent art on its face.
• Forward-citation relevance: the patent is cited by later Cisco work (e.g., US 7,948,910, "Monitoring quality of a packet flow in packet-based communication networks") and by third parties such as Citrix's adaptive packet-duplication-on-alternate-links family (US 10,367,712 / US 11,075,831).

Sources: Google Patents US7046636B1; USPTO patent grant record via uspto.report; FreePatentsOnline; Justia Patents.