Resource Scheduling for Tasks with Internal Dependencies in User-Centric Cell-Free Network

To meet the increasing demand for communication and computing resources in delay-sensitive and computation-intensive applications, resource-limited users should take advantage of the extensive computing resources at the network edge. A user-centric cell-free network can fully utilize the communication and computing resources of multiple access points (APs) to provide better load balancing among users. Parallel offloading for divisible tasks is expected to offer greater flexibility and further reduce latency. When data dependencies between different tasks has to be considered, the strict temporal relationship makes the task offloading problem even more challenging, which has been ignored in existing studies. In this paper, we propose a task-dependent parallel task offloading strategy for the user-centric cell-free network. The total time latency is minimized by joint optimizing AP selection, power control, task assignment and computational resource allocation. The formulated problem is then solved by an intelligent optimization algorithm based on gray wolf optimization algorithm (GWO). Simulation results show that in user-centric cell-free networks with limited computational resources, the proposed scheme can effectively reduce user latency.