PSoC-Based Real-Time Data Acquisition for a Scalable Spiking Neural Network Hardware Architecture

Data acquisition for monitoring the spiky activity of large-scale SNN hardware architectures are a challenge due to their time constraints, complexity, large logic size, and so on. This paper presents a versatile PSoC-Based Data Acquisition prototype, where a specialized Master Device is used for this purpose. It benefits from the heterogeneous nature of SoC platforms that allows it to host programmable logic together with a hard-core ARM processor integrating memory and a variety of peripherals in a single chip. The presented design enables monitoring the performance of a multi-chip neural network through a single Ethernet interface in a hardware and software co-design, which is combined with an application developed in Python that allows the visualization on the PC of a dynamic raster plot of neural activity. In addition, an example of full platform functionality is shown.