Open RAN AI Patents: Who Leads, Where the Gaps Are 2026
- Small, concentrated field. just 26 patent families total, with filing activity peaking at 13 in 2023 and falling off sharply since — this is an early-stage cluster, not a mature one.
- One filer dominates citation weight. a single US filing on O-RAN performance optimization and configuration carries 51 citations, more than seven times the next-most-cited record.
- Filing is fragmented outside the leader. co-assignee pairs each appear once, and recent-year momentum across named assignees is flat or down 100% year-on-year — no firm is currently accelerating.
Filing growth compares 2021 (5 records) with 2024 (1) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field.
What this landscape covers
This dataset tracks patent families at the intersection of Open RAN architecture and machine learning-driven network functions — traffic prediction, RIC-hosted xApps, and AI-based network automation — filtered against IPC classes covering wireless networks (H04W), digital transmission (H04L) and AI computing models (G06N). It is a narrow, technically specific slice: 26 published families across roughly a decade of filing activity, most of it concentrated in the last five years.
Because publication lags filing by around 18 months, the apparent drop-off after 2023 likely overstates how much activity has actually stopped — 2025 and 2026 filings are still arriving. Even allowing for that lag, though, the underlying filer base is thin and no single assignee shows sustained multi-year output.
Filing trend and technology composition
Two views of the same 26-family dataset: how filing volume has moved year over year, and which IPC subclasses carry the claim weight.
Filing trend
Filings sat at zero through much of the 2017-2021 window, rose to a peak of 13 in 2023, then dropped to a single filing at the 2022 midpoint comparison — a pattern consistent with a short, concentrated filing burst rather than sustained growth. Treat the final one to two years as undercounted given typical publication lag.
IPC composition
H04W (wireless communication networks) leads with 19 records, followed by H04L (digital transmission, 14) and H04B (transmission generally, 11). G06N (AI computing models) appears in 9 records — meaning most filings anchor their claims in the wireless/network layer and treat the machine learning component as a supporting element rather than the primary claim subject.
Shares are the percentage of the 26 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Open RAN AI and Machine Learning with Eureka
This page is one run against one query. Ask Eureka your own question about open ran ai and machine learning and every answer comes back with the patent numbers behind it.
Try EurekaMost-cited and representative filings
Machine learning assisted RRM policies for O-RAN networks (US20250247304A1, Mavenir Systems)
The filing describes a distributed unit sending buffer-occupancy and per-bearer performance parameters to a traffic-prediction analytics module, which runs an LSTM neural network to forecast per-bearer or per-logical-channel data traffic at 5QI granularity — closing the loop between measured RRM inputs and predictive scheduling decisions inside the O-RAN architecture.Filed by Mavenir Systems, dated 2025-07-31 — one of the most recent records in the dataset and a useful marker of where claim drafting has moved toward specific ML architectures (LSTM) rather than generic 'machine learning' language.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20220116799A1 | Method and device for o-ran-based performance optimization and configuration | 51 |
| 2 | WO2021187871A1 | Method and apparatus for o-ran-based performance optimization and configuration | 7 |
| 3 | US20240422587A1 | Method and device for o-ran-based performance optimization and configuration | 4 |
| 4 | US12082006B2 | Method and device for O-RAN-based performance optimization and configuration | 2 |
| 5 | EP4598102A1 | Machine learning assisted radio resource management (RRM) policies for high data rate low latency and other a… | 1 |
| 6 | WO2024033545A1 | Method and system for intelligent data collection and management for open ran intelligent controllers | 1 |
| 7 | US20240378486A1 | Exposing a machine learning model in a near real time ric | 1 |
Citation counts favour older filings simply because they've had more time to accumulate references — read this as a signal of influence within the searched corpus, not a ranking of current technical importance.
Publication numbers are shown where the record carries one (7 of 7 rows); clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the numbers say
Three findings that shape how a freedom-to-operate or whitespace review should be scoped for this field.
The field had one filing burst, not steady growth
Activity was essentially flat through 2021, spiked in 2023, then fell — a pattern more typical of a short window of active prosecution around a specific technical push (likely tied to O-RAN Alliance specification milestones) than an emerging, compounding technology area.
One filing anchors the citation graph
US20220116799A1 on O-RAN-based performance optimization and configuration is cited 51 times — more than seven times the next record. Later filings from the same family (WO2021187871A1, US20240422587A1, US12082006B2) repeat the same core claim language, suggesting one applicant staked out this ground early and has been defending it through continuations.
Claims sit in the network layer, not the ML layer
Nearly twice as many records classify under wireless communication networks (H04W) as under AI computing models (G06N). Drafters are largely claiming the network architecture and control loop, with the ML technique as a dependent or supporting element — this affects where a design-around should focus.
No repeat collaboration pattern yet
Every co-assignee pairing in the dataset occurs exactly once. Combined with flat-to-negative year-on-year momentum across named assignees, this points to a field still in early, exploratory filing rather than one with established multi-party research programmes.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to open ran ai and machine learning, with the prior art for and against each one.
Assignee landscape
Filing activity is spread thin: the named assignees in this dataset show zero filings in the latest tracked year, with several showing a full year-on-year drop from prior activity.
No assignee is currently accelerating
Every named assignee tracked for recent-year momentum shows zero filings in the latest year, several down from a prior year's activity by the full amount. This is consistent with the broader trend: the field's 2023 peak has not carried into sustained annual output.
Citation weight sits with one early filer
The most-cited record in the dataset, on O-RAN performance optimization and configuration, has been followed by multiple continuation-style filings from the same underlying invention — a sign the original applicant is actively extending protection around this specific claim territory.
A long tail of individual and small-team filers
Several records name individual inventors as co-assignees (for example T Venkata Subbamma paired separately with three different co-filers), alongside larger corporate names. This mix suggests the field is still open to individual or small-team filing rather than locked up by a handful of large incumbents.
| Assignee | Recent year | YoY |
|---|---|---|
| KDDI Corporation | 0 | -100% |
| Samsung Electronics Co., Ltd. | 0 | — |
| Dell Products L.P. | 0 | — |
| Mavenir Systems, Inc. | 0 | -100% |
| T VENKATA SUBBAMMA | 0 | -100% |
| SRIKANTH BHAT K | 0 | -100% |
| RAGHAVENDRA REDDY | 0 | -100% |
| NEC Laboratories Europe GmbH | 0 | — |
Where to take this analysis
This landscape is a starting point for scoping deeper freedom-to-operate or whitespace work in Open RAN AI.
Map the citation network around the top-cited family
US20220116799A1 and its continuations account for a disproportionate share of citation weight in this space. Understanding exactly which claim elements they cover — and which they don't — is the fastest way to find a genuine design-around.
Explore in EurekaTrack the G06N/H04W claim boundary
With most filings anchoring claims in the wireless layer rather than the AI layer, there may be room to file ML-architecture-specific claims (model type, training approach, feature set) that the current filer base has left comparatively thin.
Run a deeper search in EurekaCommon questions
This dataset identifies 26 published patent families matching Open RAN-specific machine learning claims — covering traffic prediction, RIC xApps and AI-based network automation — filed between 2015 and mid-2026. That is a small, technically narrow field compared to broader Open RAN or wireless-network patent counts generally. Because publication lags filing by roughly 18 months, the true count for 2025-2026 filings is understated in current data.
No single assignee shows sustained, growing output in this dataset — recent-year momentum for every named assignee tracked is flat or down 100% year-on-year from a prior filing. Citation weight, however, concentrates heavily around one applicant's family covering O-RAN-based performance optimization and configuration, which has been extended through multiple continuation-style filings. That combination — thin recent filing activity but strong citation concentration on one early filer — is the clearest signal of who currently controls the most defensible ground.
Claims split across four main IPC subclasses: wireless communication networks (H04W, the largest at 19 records), digital information transmission (H04L, 14), general transmission (H04B, 11) and AI computing models (G06N, 9). The pattern shows that most patent drafters treat the network architecture as the primary claim subject and the machine learning technique — such as LSTM-based traffic prediction — as a supporting or dependent element rather than the core inventive claim.
Filing activity peaked at 13 families in 2023 after being near zero through most of 2017-2021, then declined toward the most recent tracked years. This looks like a short filing burst tied to a specific technical or standards push rather than steady compounding growth. Readers should treat the apparent 2024-2026 decline cautiously, since recent filings are still being published and the true recent-year volume is likely higher than currently visible.
The clearest gaps sit in areas adjacent to the dense O-RAN performance-optimization cluster: per-bearer LSTM-based traffic forecasting at fine granularity, cross-vendor xApp interoperability, RIC-hosted policy conflict resolution, federated learning across distributed RAN units, and AI-driven energy-saving control loops. These branches show thin claim density relative to the core cluster, and the field's overall fragmentation — no repeated co-assignee pairs, no accelerating filer — suggests room for a well-drafted first claim in any of them.
Research Open RAN AI and Machine Learning in depth with Eureka
Go past this page: query the whole open ran ai and machine learning corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.