AI inference is moving to the network periphery. This dataset covers 60+ patent and literature records spanning 2014–2026, mapping the architectures, assignees, and emerging directions defining edge AI sensor processing.
filing and publication date range covered in this dataset
Explore in Eureka
~17
Indian jurisdiction patent records in this dataset
Explore in Eureka
5
key named assignee clusters with multi-jurisdictional filings in this dataset
Explore in Eureka
Published byPatSnap Insights Team··9 min readVerified by PatSnap Eureka Data
Technology Overview
From Cloud Offload to On-Device Intelligence
Edge AI real-time sensor data processing embeds AI inference, anomaly detection, and feature extraction directly on gateways, microcontrollers, field-programmable accelerators, and edge servers rather than routing raw data to centralized cloud platforms. The foundational rationale documented across multiple literature sources in this dataset is that cloud-centric architectures introduce unacceptable latency for time-critical applications while consuming excessive bandwidth.
Within this dataset, the field spans five identifiable sub-domains: embedded AI inference engines on constrained hardware, edge-cloud hybrid orchestration with dynamic task allocation, multi-sensor fusion combining camera, LiDAR, thermal, IMU, and environmental streams, federated and continuous learning frameworks for privacy-preserving model updates, and sensor-adaptive control loops balancing the accuracy-latency-energy trilemma.
Top Assignees by Patent Filing Count — Edge AI Sensor Processing (Dataset Snapshot)
↗ Click bars to explore
The innovation timeline in this dataset runs from 2014 through 2026. IBM's foundational wireless sensor network edge decisioning patent (2014, US) anchors the earliest layer, while the 2017–2020 period saw hardware architecture and commercial platform patents consolidate — notably Tyco Fire & Security GmbH's edge intelligence platform family and ETRI's IoE edge computing system (2020, US).
The most recent filing cluster in this dataset, comprising 17 patent records dated 2025–2026, is concentrated in Indian jurisdiction filings from engineering colleges, university spin-outs, and individual inventors. In this dataset, India accounts for approximately 17 of 25+ records with explicit jurisdiction data, making it the highest-volume jurisdiction by filing count in retrieved records.
PatSnap Eureka Source: PatSnap Eureka retrieved patent and literature records, edge AI sensor processing dataset snapshot, 2014–2026. Counts reflect records within this dataset only.Explore the data ↗
Data & Trends
Filing Activity and Technology Cluster Distribution
Patent activity in this dataset spans 2014–2026 with a pronounced acceleration after 2021. Technology clusters range from embedded inference engines — the most densely populated in this dataset — to emerging digital twin and federated learning filings concentrated in 2025–2026.
Technology Cluster Distribution — Edge AI Sensor Processing (Dataset Snapshot)
Embedded AI inference engines represent the largest identifiable cluster in this dataset, followed by edge-cloud hybrid orchestration and multi-sensor fusion, while federated learning and sensor-adaptive control are smaller but rapidly growing clusters in retrieved records.
↗ Click bars to explore
Filing Activity by Period — Edge AI Sensor Processing (Dataset Snapshot)
In this dataset, filings were sparse in 2014–2018, consolidated through 2019–2022, and accelerated sharply in 2023–2026, with 17 of 25+ patent records dated 2025–2026 concentrated in Indian jurisdiction filings from universities and individual inventors.
↗ Click bars to explore
PatSnap Eureka Source: PatSnap Eureka retrieved patent and literature records, edge AI sensor processing dataset snapshot, 2014–2026. Period counts are approximate based on explicit filing dates in retrieved records.Explore the data ↗
Application Domains
Key Deployment Domains for Edge AI Sensor Processing
Edge AI sensor processing patents in this dataset address at least six distinct application domains, from industrial predictive maintenance and smart cities to healthcare wearables, security surveillance, smart agriculture, and data center triage. Each domain is represented by named filings with specific technical configurations.
Edge Sensor Actuation · Predictive Maintenance
Industrial IoT & Machine Health
Tata Consultancy Services' 2021 patent (US/EP/IN) describes continuous health and process monitoring using camera, current, microphone, ultrasound, radar, and temperature sensors processed at the edge to prevent downtime without cloud connectivity. Johnson Controls Tyco IP Holdings' 2022 IN filing explicitly targets industrial machine sensor data analytics in distributed IoT environments. These filings cover oil and gas, mining, steel, and power plant contexts.
Dr. M.G.R. Educational & Research Institute's 2026 IN patent deploys a lightweight ML model on a microcontroller for thermal hotspot inference with noise filtering, frame normalization, and ambient temperature compensation for all-weather vehicle and pedestrian detection. Dusitech Co., Ltd.'s 2026 US patent processes object detection and metainformation generation onboard a drone's mission apparatus, transmitting only low-capacity structured data to ground controllers.
Akumen Artificial Intelligence Private Limited's 2025 IN patent captures wearable sensor streams — visual, audio, and depth perception — and performs denoising, contrast alteration, and normalization on edge devices before further analysis. The 2025 IN filing by Venkatesh Prabu Parthasarathy explicitly lists healthcare monitoring as a target domain alongside smart cities and industrial automation within a federated edge AI analytics pipeline.
AlienSense Limited's 2026 US and WO patents process object recognition at the edge device level to construct a real-time digital twin of a physical environment, reducing backend computation by using synchronized object libraries. Cresfree Co., Ltd.'s 2021 KR patent combines GPS, tilt sensors, and image processing on an AI camera to detect dangerous situations at precisely determined target positions for smart surveillance applications.
PatSnap Eureka Source: PatSnap Eureka retrieved patent records, edge AI sensor processing dataset snapshot, 2014–2026. Domain classifications derived from explicit application descriptions in retrieved filings.Explore insights ↗
Key Assignees
Key Patent Assignees in Edge AI Sensor Processing (Retrieved Records)
In this dataset, the Tyco Fire & Security GmbH / Foghorn Systems / Johnson Controls Tyco IP Holdings family accounts for 5 filings across US, IN, WO, and AU jurisdictions — the largest single assignee cluster in retrieved records. Tata Consultancy Services Limited follows with 3 filings spanning US, EP, and IN, while Muthayammal Engineering College and Venkatesh Prabu Parthasarathy each contribute 2 filings in retrieved records.
Top Assignees by Filing Count in Retrieved Records (Dataset Snapshot)
↗ Click bars to explore
Semantic CEP Platform · ML Edge Deployment
Tyco Fire & Security / Johnson Controls
This assignee family holds 5 filings in this dataset across US, IN, WO, and AU jurisdictions, spanning 2018–2025. Core patents include the Edge Intelligence Platform (2018, US) introducing a semantic expression language (Vel) for sensor stream matching, the Edge Computing Platform (2019, US) adding time-series databases and CEP engines, and the Intelligent Edge Computing Platform with Machine Learning Capability (2020–2024, US) extending ML model conversion for on-device execution. The Foghorn Systems WO filing (2020) is now part of this family. Multiple filings carry active legal status.
Tata Consultancy Services Limited holds 3 filings in this dataset spanning US, EP, and IN jurisdictions from 2021–2022. The core patent family covers edge-based sensor actuation and control in IoT networks for event monitoring, using camera, current, microphone, ultrasound, radar, and temperature sensors processed at the edge to prevent downtime without cloud connectivity. Applications explicitly target oil and gas, mining, steel, and power plant environments. The EP filing (2021) extends the US priority claim internationally.
This assignee holds multiple active patents in this technology cluster with filings spanning several jurisdictions and covering key process steps.
Undisclosed
Patent Assignee · Filing Activity
Additional Assignee
This assignee holds multiple active patents in this technology cluster with filings spanning several jurisdictions and covering key process steps.
Undisclosed
🔍
Unlock 8 More Assignees in This Edge AI Dataset
Additional named assignees in this dataset include MatrixSpace, Inc. (multi-sensor collaborative fusion, US 2024), AlienSense Limited (AI video sensor for digital twin, US/WO 2026), and Dusitech Co., Ltd. (autonomous drone edge AI, US 2026). Sign in to PatSnap Eureka to explore their full filing profiles.
MatrixSpace sensor fusionAlienSense digital twin IP+ more
PatSnap Eureka Source: PatSnap Eureka retrieved patent records, edge AI sensor processing dataset snapshot, 2014–2026. Assignee filing counts reflect records in this dataset only.Explore players ↗
Emerging Directions
Five Forward Signals from 2025–2026 Filings
Among the most recent filings in this dataset, five directional signals are identifiable: digital twin integration with edge AI video sensors, autonomous drone and aerial edge AI, federated learning for privacy-preserving edge processing, hybrid edge-fog-cloud tiered orchestration, and cognitive protocol-layer adaptation at the edge node.
Digital Twin Integration with Edge AI Video Sensors
AlienSense Limited's 2026 US and WO patents signal a convergence between edge sensor processing and real-time physical-world digital twin construction. Object recognition libraries are synchronized between edge devices and backend systems to minimize transmission overhead. This positions edge AI as the foundational enabler of physically grounded digital intelligence rather than merely a processing optimization.
Venkatesh Prabu Parthasarathy's 2025 IN filings (two records) specifically claim federated learning frameworks integrated into edge processing pipelines, enabling collaborative model improvement without centralizing raw data. A secure data management layer and an intelligent decision engine accompany the federated architecture. As data sovereignty and privacy regulations tighten globally, federated training capability is identified in this dataset as a key platform differentiator.
The hybrid edge-fog-cloud tiered AI direction (Dr. Shambhu Shankar Rai, 2026, IN) introduces a three-tier architecture with cross-layer feature fusion and continuous model updates. Sign in to PatSnap Eureka to read the full emerging directions analysis with linked patent records.
Hybrid fog-cloud tiered AIIIT Delhi cognitive protocol+ more
PatSnap Eureka Source: PatSnap Eureka retrieved patent records, edge AI sensor processing dataset snapshot, 2025–2026 filing cluster. Directional signals derived from explicit claims and abstracts in retrieved filings.Explore emerging trends ↗
Technology Comparison
Embedded AI Inference vs. Edge-Cloud Hybrid Orchestration
Click any row to explore further.
Dimension
Embedded AI Inference Engines
Edge-Cloud Hybrid Orchestration
Primary Goal
On-device classification and detection on constrained hardware with no cloud dependency
Dynamic task allocation across edge, fog, and cloud layers to balance latency and accuracy
Representative Filing
Intelligent edge computing device with AI-driven data preprocessing (Dr. Deepa Parasar, 2026, IN)
Hybrid AI models for real-time IoT data analytics on edge-cloud platforms (Dr. Shambhu Shankar Rai, 2026, IN)
Hardware Target
Microcontroller units (MCU), embedded modules (e.g. ESP32-CAM), edge nodes with high-speed memory buffering
Distributed edge nodes, fog intermediaries, and cloud back-ends with cross-layer orchestration engine
Latency Profile
Ultra-low latency — inference occurs locally without network round-trip
Tiered latency — immediate anomaly detection at edge, contextual inference at fog, deep analysis at cloud
Model Update Mechanism
Adaptive learning under resource constraints at the local device level
Continuous model updates with dynamic task orchestration and cross-layer feature fusion
Privacy Posture
Raw data stays at source device — no transmission required for inference
Federated learning frameworks in some filings enable decentralized training without centralizing raw data
Key Named Assignees (Dataset)
Dr. Deepa Parasar (IN, 2026), Ajay Kumar Garg Engineering College (IN, 2025), Dr. M.G.R. Educational & Research Institute (IN, 2026)
Dr. Shambhu Shankar Rai (IN, 2026), Venkatesh Prabu Parthasarathy (IN, 2025), Muthayammal Engineering College (IN, 2026)
Primary Application Domains
Smart manufacturing, environmental monitoring, all-weather vehicle and pedestrian detection
Healthcare monitoring, smart cities, industrial automation with federated analytics
PatSnap Eureka Source: PatSnap Eureka retrieved patent records, edge AI sensor processing dataset snapshot, 2014–2026. Comparison dimensions derived from explicit claims and technical descriptions in retrieved filings.Compare in Eureka ↗
Frequently asked questions
Frequently Asked Questions: Edge AI Real-Time Sensor Data Processing
Edge AI real-time sensor data processing refers to the convergence of AI inference, embedded machine learning, and distributed IoT sensor architectures to enable low-latency, on-device analytics at the network periphery — on gateways, microcontrollers, field-programmable accelerators, and edge servers — rather than in centralized cloud infrastructure.
In this dataset, the Tyco Fire & Security GmbH / Foghorn Systems / Johnson Controls Tyco IP Holdings family holds 5 filings across US, IN, WO, and AU jurisdictions, making it the largest single assignee cluster in retrieved records. Tata Consultancy Services Limited follows with 3 filings spanning US, EP, and IN.
The five sub-domains identified in this dataset are: (1) embedded AI inference engines on constrained hardware, (2) edge-cloud hybrid orchestration with dynamic task allocation, (3) multi-sensor fusion combining camera, LiDAR, thermal, IMU, and environmental streams, (4) federated and continuous learning frameworks, and (5) sensor-adaptive control loops addressing the accuracy-latency-energy trilemma.
In this dataset, approximately 17 of 25+ patent records with explicit jurisdiction data are Indian filings, predominantly from engineering colleges and individual inventors, with most carrying 'pending' status. This volume reflects active academic and startup innovation in India's engineering ecosystem during 2025–2026, though these filings have limited commercial-grade enforcement infrastructure compared to US-origin patents in the dataset.
Multiple patents in this dataset — particularly Tata Consultancy Services' sensor actuation series — explicitly frame hierarchical sensor selection, adaptive mode switching, and waveform control as the primary mechanisms for resolving the trilemma between achieving low latency, high inference accuracy, and minimal energy consumption simultaneously on constrained edge hardware.
Five directional signals are identifiable in 2025–2026 filings: digital twin integration with edge AI video sensors (AlienSense Limited, US/WO 2026), autonomous drone edge AI pipelines (Dusitech Co., Ltd., US 2026), federated learning for privacy-preserving edge processing (Venkatesh Prabu Parthasarathy, IN 2025), hybrid edge-fog-cloud tiered orchestration (Dr. Shambhu Shankar Rai, IN 2026), and cognitive experiential protocol-layer adaptation (Indian Institute of Technology Delhi, IN 2026).
Still have questions? PatSnap Eureka can answer them instantly from patent and research data.Ask Eureka ↗
PatSnap Eureka
Generate Your Edge AI Sensor IP Landscape Report
Join 18,000+ innovators using PatSnap Eureka to generate reports like this one for any technology area.
Data and insights on this page are based on a limited patent and literature dataset and are for reference only. Figures may not represent the complete technology landscape.
Ask me anything about this tech. PatSnap Eureka searches patents and research literature to answer instantly.