Joint Communication and Computing Optimization for Digital Twin Synchronization with Aerial Relay

Digital twin (DT) applications usually need to be synchronized in real time with the state of the physical system (PS). This process includes both synchronizing the data collected by sensors in the PS to a server and updating the DT at the server in adaptation to the dynamics of the PS. However, communicating with power- and rate-limited sensors and processing high-volume sensed data with limited computing resources present significant challenges for DT synchronization. In this paper, we tackle both bottlenecks by proposing a joint communication and computing design for DT synchronization. In particular, we exploit buffering at an aerial cluster head of the sensors and enable buffer-aided (BA) relaying to increase the communication throughput of the sensors during data synchronization. Moreover, we adopt a novel stream computing scheme, which allows for DT updating in parallel with data synchronization, to accelerate DT synchronization. To maximize the performance of the proposed approach, we jointly optimize the trajectory of the aerial relay and the communication and computing resource allocation for minimizing the DT synchronization time. The formulated problem is a mixed-integer nonconvex problem, which is generally intractable. To tackle this challenge, we propose a low-complexity two-layer iterative suboptimal algorithm. Our simulation results show that the DT synchronization time can be significantly reduced by up to 43.8%, through stream computing and the joint optimization of the relay's trajectory and the communication/computing resource allocation.