A lightweight edge computing platform designed on embeded system

Edge computing platform is a critical component to form next-generation intelligent equipment. In an effort to support new military and industrial applications with more flexible computing power, a new lightweight edge computing platform is introduced in this paper. This platform is fully based on embedded hardware, such as ARM, DSP and FPGA. Platform's management and scheduling architecture is developed using Go language to minimize code size and dependence on three-party libraries. In addition, the management and scheduling commands transmit across the platform over the Ethernet, while data transmit among computing nodes via the RapidIO network to isolate the commands and data channels. An experimental letter count application was performed to validate the feasibility of the platform and also supply optimization hint with a data throughput test. This work has the potential to reduce developer's experience in parallel and distributed systems, and then the cost and time in development of embedded intelligent equipment.