.png){kind=icon}
.png){kind=icon}
Return
Intelligent Computing Oriented Joint Model Deployment and Task Offloading in MEC Systems
.png){kind=icon}
PRE
AI
DOI：10.1109/tvt.2026.3679675.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A framework of intelligent computing oriented model deployment and task offloading is proposed for mobile edge computing (MEC) systems. In this framework, the cloud deploys intelligent models on MEC servers as needed, while terminals compress raw data and offload tasks to MEC servers with corresponding models, reducing transmission overhead and enhancing computation efficiency. A unified optimization framework that integrates task accuracy, delay, and resource constraints is proposed to advance the processing efficiency of intelligent computing oriented edge systems. The system utility based on task accuracy and delay is minimized by jointly optimizing model deployment, terminal association, and compression ratio. The resulting NP-hard mixed-integer nonlinear problem is addressed using a block coordinate descent-based algorithm, decomposing it into two subproblems solved via branch-and-cut and successive convex approximation. Finally, simulation results validate the performance improvement of the proposed algorithm over benchmarks.
Keywords:
Compression offloading
intelligent computing
model deployment
mobile edge computing
Journal
IF:
7.1
Papers:
1.8W
Citations:
6.6W
Organization
View more organizations