Delay-Optimal Closed-Form Scheduling for Multi-Destination Computation Offloading

This letter studies a delay-optimal multi-destination computation offloading system, where the subtasks are transmitted to multiple Edge Computing Servers (ECSs) sequentially due to the limited communication resource, and different subtasks are computed in parallel at different ECSs. Consider the communication and computation capabilities, we jointly optimize the task assignment and the offloading scheduling for minimizing the total delay, which is the maximum completion time among all ECSs with their corresponding subtasks. We first consider a simplified case with equal task assignment, and then obtain the optimal offloading order according to the Lowest-Computation-First (LCF) rule. Under flexible task assignment, we derive the optimal offloading scheduling policy according to the Highest-Communication-First (HCF) offloading order with closed-form task assignment. Simulation results demonstrate the jointly optimized offloading scheduling policy reduces the delay by up to 62.4% compared with the non-scheduling offloading method.

Automated summary

The letter proposes a delay-optimal multi-destination computation offloading system by optimizing task assignment and offloading scheduling, which reduces the delay by up to 62.4% compared to non-scheduling offloading method.