A Remote FPGA-based Experimental Teaching System Design Supporting Single-board Multi-user and Multi-board Single-user Operations in MOOCs

Unlike traditional offline courses, where the enrollment number is generally fixed, massive open online courses (MOOCs) often exhibit significant disparities in the enrollment number across different sections. Namely, this number can vary by several or even hundreds of times depending on the MOOC sections. To this end, this study proposes a remote FPGA-based experimental teaching system with two main innovations. First, a software-hardware co-work framework is designed to divide a single physical FPGA into multiple independently virtual FPGAs, allowing for multiple users to use the same physical FPGA concurrently. Second, an "X86 CPU+Multi-PYNQ" collaborative computing system supporting up to 16 PYNQs for parallel computing is designed. This system uses an X86 CPU as a main processor to distribute computing tasks to the PYNQ-cluster for scheduling FPGA parallel computations, which enables a single user to use multiple FPGAs concurrently. Therefore, the proposed system can effectively address the problem of underutilized boards in MOOCs where the enrollment number is smaller than the number of available FPGA boards. In summary, the system proposed in this paper can effectively mitigate the conflict between the number of students and the number of FPGA boards in MOOCs.