Toward Edge General Intelligence With Multiple-Large Language Model (Multi-LLM): Architecture, Trust, and Orchestration

К периферийному общему интеллекту с использованием нескольких больших языковых моделей (Multi-LLM): архитектура, доверие и оркестрация
Xianbin Wang, Dusit Tao Niyato, Hongfang Yu, Xuemin Shen, Zehui Xiong, Jiacheng Wang, Ruichen Zhang, Yinqiu Liu, Gang Sun, Haoxiang Luo
2025-09-22

dynamic orchestrationedge computingmulti-LLM systemsmultimodal architecturestrustworthy AI
Edge computing enables real-time data processing closer to its source, thus improving the latency and performance of edge-enabled AI applications. However, predictive AI models often fall short when dealing with complex, dynamic tasks that require advanced reasoning and multimodal data processing. This survey explores the integration of multi-LLMs (Large Language Models) to address these challenges in edge computing, where multiple specialized LLMs collaborate to enhance task performance and adaptability in resource-constrained environments. We review the transition from conventional edge AI models to single LLM deployment and, ultimately, to multi-LLM systems. The survey discusses enabling technologies such as dynamic orchestration, resource scheduling, and cross-domain knowledge transfer that are key for multi-LLM implementation. A central focus is on trusted multi-LLM systems, ensuring robust decision-making in environments where reliability and privacy are crucial. We also present multimodal multi-LLM architectures, where multiple LLMs specialize in handling different data modalities, such as text, images, and audio, by integrating their outputs for comprehensive analysis. Finally, we highlight future directions, including improving resource efficiency, trustworthy governance multi-LLM systems, while addressing privacy, trust, and robustness concerns. This survey provides a valuable reference for researchers and practitioners aiming to leverage multi-LLM systems in edge computing applications.
1
Collaborating specialized LLMs can improve task performance and adaptability in resource-constrained edge-computing environments.
2
Dynamic orchestration, resource scheduling, and cross-domain knowledge transfer are identified as key enabling technologies for deploying multi-LLM systems at the edge.
3
Multi-LLM systems are presented as a way to overcome conventional edge AI models’ limitations on complex, dynamic tasks requiring advanced reasoning and multimodal processing.
4
Multimodal multi-LLM systems assign different modalities, including text, images, and audio, to specialized models and integrate their outputs for comprehensive analysis.
5
Trusted multi-LLM architectures are emphasized as necessary for reliable decision-making when privacy and robustness are critical.

multi-LLM systems integrated into edge computing environments

their architecture, trustworthiness, orchestration, resource efficiency, multimodal collaboration, and adaptability under resource, privacy, and robustness constraints

Publication Details
Publication Date
2025-09-22
Journal
Publisher
ISSN
Cited by
94
Access Type
Author Information
Authors
Xianbin Wang
Dusit Tao Niyato
Hongfang Yu
Xuemin Shen
Zehui Xiong
Jiacheng Wang
Ruichen Zhang
Yinqiu Liu
Gang Sun
Haoxiang Luo
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%