.png){kind=icon}
.png){kind=icon}
Return
Toward AI-Native Task Orchestration for Collaborative Computing in SAGSINs
.png){kind=icon}
PRE
AI
DOI：10.1109/MCOM.004.2400304.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The transition to 6th generation (6G) mobile networks is a crucial step toward ubiquitous coverage, ultra-broadband connectivity, and comprehensive support for Internet of Things (IoT) applications such as autonomous driving, remote control, and more. This leap will be facilitated by the integration of space-air-ground-sea integrated networks (SAGSINs), which improve coverage for high-throughput applications such as Earth observation and intelligent transportation. With the increasing demand for communication and computing resources, cloud and edge collaboration is increasingly being used to ensure low latency and highly reliable services. This article explores collaborative computing at the network edge and high-lights the critical role of artificial intelligence (AI) agents in enabling cross-domain collaboration. We present an AI-native task orchestration framework that allows us to improve the ecosystem and devel-op efficient edge-toedge collaboration protocols. In particular, we have developed a hierarchical framework for collaborative planning and reasoning based on large language models (LLMS) and hierarchical task networks (HTNs) called L-HTN. Furthermore, we present a multi-agent deep rein-forcement learning (DRL) approach for task execution tailored to the stringent latency and reliability requirements of edge network applications. This work provides insights into optimizing collaborative computing at the network edge and sets new benchmarks for operational efficiency and quality of service in the era of 6G networks.
Keywords:
Collaboration
Artificial intelligence
Resource management
Adaptation models
Planning
Real-time systems
6G mobile communication
Performance evaluation
Computational modeling
Bandwidth
Journal
IF:
8.2
Papers:
6.9K
Citations:
2.2W