This is a draft article. It is a work in progress open to editing by anyone. Please ensure core content policies are met before publishing it as a live Wikipedia article. Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL Easy tools: Citation bot (help) | Advanced: Fix bare URLs · Article logs · Draft logs. Last edited by Citation bot (talk | contribs) 2 months ago. (Update) Finished drafting? Submit for review

This article may incorporate text from a large language model, which is prohibited in Wikipedia articles. It may include hallucinated information, copyright violations, claims not verified in cited sources, original research, or fictitious references. Any such material should be removed. (March 2026) (Learn how and when to remove this message)

Edge intelligence (or Edge AI) is a paradigm in distributed computing that integrates artificial intelligence (AI) capabilities directly into edge devices and edge servers, enabling data processing and decision-making close to the data source ^[1]. It combines concepts from edge computing and machine learning to reduce latency, improve privacy, and enable real-time analytics in resource-constrained environments.

Traditional cloud-based AI systems rely on centralized data processing, where raw data is transmitted from end devices to remote data centers. In contrast, edge intelligence shifts computation toward the network edge—such as sensors, mobile devices, and embedded systems—allowing models to operate locally or collaboratively across distributed nodes.^[1]

This approach is particularly useful in applications that require:^[1]

• Low latency (e.g., autonomous driving, industrial automation)
• Bandwidth efficiency (reducing data transmission to the cloud)
• Privacy preservation (keeping sensitive data local)

A typical edge intelligence system consists of three layers:^[2]

1. Device layer Includes sensors, IoT devices, and mobile endpoints that collect and sometimes process data locally.
2. Edge layer Comprises edge servers or gateways that perform intermediate processing, model inference, or aggregation.
3. Cloud layer Provides large-scale training, global coordination, and long-term storage.

These layers often collaborate through distributed learning frameworks such as federated learning or split learning.^[2]

Edge devices have limited computational resources. Techniques such as pruning, quantization, and knowledge distillation are used to reduce model size and complexity.^[3]

Learning is distributed across multiple devices^[4]:

• Federated learning enables local model training without sharing raw data.
• Split learning divides a model between client and server to balance computation and privacy.

Models are deployed directly on edge devices for real-time decision-making, avoiding cloud latency.^[3]

Hybrid approaches dynamically partition tasks between edge and cloud depending on resource availability and task requirements.^[2]

Edge intelligence is widely used in ^[1]:

• Autonomous vehicles: real-time perception and decision-making
• Smart cities: traffic monitoring and infrastructure management
• Healthcare: wearable devices and remote diagnostics
• Industrial IoT: predictive maintenance and process optimization
• Smart homes: voice assistants and security systems

Benefits for end users ^[1]:

• Low latency: real-time processing without cloud dependency
• Improved privacy: sensitive data remains on-device
• Reduced bandwidth usage: less data transmitted to the cloud
• Scalability: distributed computation across many devices

---

Research Challenges ^[2] :

• Resource constraints: limited memory, compute, and power
• Model heterogeneity: varying device capabilities
• Security and privacy risks: potential leakage from models or communication
• System complexity: coordination across distributed nodes
• Non-IID data: data distributions differ across devices

---

Current research in edge intelligence focuses on:

• Efficient distributed training under resource constraints
• Privacy-preserving and secure learning methods
• Adaptive model partitioning and scheduling
• Robustness against adversarial attacks
• Integration with 5G/6G and next-generation networks

---

• Edge computing
• Federated learning
• Split learning
• Internet of Things
• Distributed machine learning

1. Raghubir Singh, Sukhpal Singh Gill, Edge AI: A survey, Internet of Things and Cyber-Physical Systems, Volume 3, 2023, Pages 71-92, ISSN 2667-3452, https://doi.org/10.1016/j.iotcps.2023.02.004.
2. Yin Zhang, Chi Jiang, Binglei Yue, Jiafu Wan, Mohsen Guizani, Information fusion for edge intelligence: A survey, Information Fusion, Volume 81, 2022, Pages 171-186, ISSN 1566-2535, https://doi.org/10.1016/j.inffus.2021.11.018.