This dataset, scoped to Open RAN and O-RAN architecture claims combined with RAN dimensioning, resource allocation, energy-saving or network design optimization language under IPC classes H04W16, H04W24 and H04W28, contains 24 patent families published between 2015 and mid-2026. That is a deliberately narrow slice of the much larger overall O-RAN patent literature, filtered to claims that explicitly address optimization rather than general architecture. A broader search string without the optimization keyword filter would return a substantially larger set.

The assignee list in this corpus is short and includes a Japanese carrier group (Rakuten Mobile, filing jointly across its Japanese and US entities), a major chipmaker (Intel), a European research lab (NEC Laboratories Europe) and named individual inventors. None of these currently shows filing activity in the latest tracked year, and the field as a whole peaked in 2021, so leadership here is better read as who holds the earliest, most-cited claims rather than who is currently most active.

Filings in this corpus rose from zero in 2017 to a peak of 12 families in 2021, then fell to 6 by 2022, which lines up with the period when RAN Intelligent Controller architecture and network-slice resource-allocation concepts were being formalized in O-RAN Alliance specifications. Once core architectural claims are staked out early, later filers tend to shift toward narrower continuations or adjacent branches rather than filing fresh foundational claims, which is consistent with the decline seen after 2021. The most recent one to two years should still be treated as undercounted because of publication lag.

US11917527B2, assigned to Intel Corporation, covers a Non-Real-Time RAN Intelligent Controller inside an SMO entity that allocates network slice subnet instance resources using an AI/ML model trained on collected performance measurements, timing the reallocation to the model's inference output. It is a continuation of an earlier 2021 filing that is the single most-cited record in this corpus at 165 citations. Anyone building AI-driven slice-resource allocation into an O-RAN Non-RT RIC should review this family's claim scope directly rather than relying on summaries, since it anchors much of the later citation activity in this space.

The IPC composition shows heavy overlap between H04W wireless-network claims and H04L transport-layer claims, but only 4 of 24 records reach into G06N AI-model classifications, which is thinner than the field's AI-heavy narrative would suggest. Energy-saving cell switch-off scheduling, cross-vendor RU/DU interface optimization, and conflict resolution between xApps and rApps all appear adjacent to the dense RIC and slice-allocation claims but show comparatively little direct coverage in this 24-family set. These are reasonable areas to prior-art-search before committing new claim language, though the small overall corpus size means any gap should be verified against a broader search before relying on it.