To enhance the efficiency of different subsystems within embedded devices, thereby augmenting their capabilities, we incorporate cloud computing architecture within these devices to create an in-device micro-cloud platform. This allows subsystems, which were originally independent, to collaborate. Tasks on the device can be managed within a unified scheduling framework, thus increasing task flexibility and reducing task interaction overhead. Simultaneously, tasks on embedded devices often possess real-time requirements, an aspect often overlooked in previous cloud computing research. Accordingly, this paper prioritizes the real-time nature of tasks, studying a fair scheduling algorithm for tasks within the micro-cloud platform to address this hitherto neglected issue.