Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Edge General Intelligence via Large Language Models: Opportunities and Challenges

Handi Chen, Weipeng Deng, Shuo Yang, Jinfeng Xu, Zhihan Jiang, Edith C. H. Ngai, Jiangchuan Liu, Xue Liu

TL;DR

This paper surveys the emergence of Edge General Intelligence (EGI) powered by Large Language Models (LLMs), distinguishing it from traditional Edge Intelligence (EI) and framing three architectural paradigms: centralized, hybrid, and decentralized. It surveys system designs, implementation examples, and feasibility across these architectures, and compares Small Language Models (SLMs) on edge devices to inform orchestration strategies. It also discusses future directions in efficient on-device deployment, latency optimization, domain-specific adaptation, security/privacy, and multi-agent collaboration, highlighting both opportunities and challenges for real-world edge deployments. The work provides a comprehensive vision for integrating LLMs in edge networks, offering concrete guidance for researchers and practitioners to advance scalable, private, and responsive edge intelligence.

Abstract

Edge Intelligence (EI) has been instrumental in delivering real-time, localized services by leveraging the computational capabilities of edge networks. The integration of Large Language Models (LLMs) empowers EI to evolve into the next stage: Edge General Intelligence (EGI), enabling more adaptive and versatile applications that require advanced understanding and reasoning capabilities. However, systematic exploration in this area remains insufficient. This survey delineates the distinctions between EGI and traditional EI, categorizing LLM-empowered EGI into three conceptual systems: centralized, hybrid, and decentralized. For each system, we detail the framework designs and review existing implementations. Furthermore, we evaluate the performance and throughput of various Small Language Models (SLMs) that are more suitable for development on edge devices. This survey provides researchers with a comprehensive vision of EGI, offering insights into its vast potential and establishing a foundation for future advancements in this rapidly evolving field.

Towards Edge General Intelligence via Large Language Models: Opportunities and Challenges

TL;DR

This paper surveys the emergence of Edge General Intelligence (EGI) powered by Large Language Models (LLMs), distinguishing it from traditional Edge Intelligence (EI) and framing three architectural paradigms: centralized, hybrid, and decentralized. It surveys system designs, implementation examples, and feasibility across these architectures, and compares Small Language Models (SLMs) on edge devices to inform orchestration strategies. It also discusses future directions in efficient on-device deployment, latency optimization, domain-specific adaptation, security/privacy, and multi-agent collaboration, highlighting both opportunities and challenges for real-world edge deployments. The work provides a comprehensive vision for integrating LLMs in edge networks, offering concrete guidance for researchers and practitioners to advance scalable, private, and responsive edge intelligence.

Abstract

Edge Intelligence (EI) has been instrumental in delivering real-time, localized services by leveraging the computational capabilities of edge networks. The integration of Large Language Models (LLMs) empowers EI to evolve into the next stage: Edge General Intelligence (EGI), enabling more adaptive and versatile applications that require advanced understanding and reasoning capabilities. However, systematic exploration in this area remains insufficient. This survey delineates the distinctions between EGI and traditional EI, categorizing LLM-empowered EGI into three conceptual systems: centralized, hybrid, and decentralized. For each system, we detail the framework designs and review existing implementations. Furthermore, we evaluate the performance and throughput of various Small Language Models (SLMs) that are more suitable for development on edge devices. This survey provides researchers with a comprehensive vision of EGI, offering insights into its vast potential and establishing a foundation for future advancements in this rapidly evolving field.

Paper Structure

This paper contains 36 sections, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background of EI and LLMs
  3. Edge Computing and EI
  4. Large Language Models
  5. LLMs: Evaluating the Next Level of EI
  6. Hierarchical Cognitive Abilities of EI
  7. LLM-enhanced EGI
  8. Edge General Intelligence Systems
  9. Centralized EGI System
  10. System Framework
  11. Implementation
  12. Discussion
  13. Advantages
  14. Disadvantages
  15. Feasibility Discussion
  16. ...and 21 more sections

Figures (2)

  • Figure 1: An illustration of increasing intelligence from narrow to general intelligence. The main capabilities of general intelligence include comprehension (e.g., information summarization), reasoning (e.g., logic inference), generation (e.g., writing a story), adaptability (e.g., compatibility with various scenarios), multimodality (e.g., images and audios), and contextualization (e.g. context-aware dialogue), supporting various use cases.
  • Figure 2: Architectures of centralized, hybrid, and decentralized EGI systems (LLM and SLM represent large language model and small language model, respectively). (a) Centralized EGI system. (b) Hybrid EGI system. (c) Decentralized EGI system.