Online Optimization of Edge Empowered Human Digital Twin Deployment and Task Offloading
Yuye Yang, You Shi, Ruoyang Chen, Changyan Yi, Jiawen Kang
Abstract
Human digital twin (HDT) is an emerging paradigm that constructs powerful virtual twins (VTs) of physical twins (PTs) for assisting human-centric and complex task executions. In this paper, a two-timescale online optimization for HDT deployment under an end-edge-cloud framework is studied. We consider that PTs’ corresponding VTs are deployed on edge servers, including placing generic models by downloading experiential knowledge from the cloud and updating customized models by uploading personalized data from PTs. Taking into account HDT’s unpredictable mobility and status variation, we dynamically optimize VTs’ construction and PTs’ task offloading, together with communication and computation resource allocations to maximize task execution accuracy under stringent delay constraint. Observing the asynchronization of different decision variables, we propose a novel two-timescale accuracy-aware online optimization approach (TACO). Specifically, TACO employs an improved Lyapunov method to decompose the problem into multiple instant ones, and then addresses the two-timescale issue alternately by leveraging piecewise McCormick envelopes and block coordinate descent based algorithms. Theoretical analyses and simulations evaluate the performance of the proposed solution, and show its superiority over counterparts.