Characterization of Adaptive Implementation of Neuromorphic Spiking Sensory Systems On-Chip with Self-X Abilities

Efficient interfacing with an expanding variety of sensors is necessary for sensor systems to act as the interface between an artificial system and the real world. A sensor system's accuracy, robustness, and flexibility play crucial roles in the overall application system. Drawing inspiration from the impressive computational abilities found in biological systems, this project focuses on developing a neuromorphic spiking sensory system that emulates the efficient style of biology. By replicating the efficient design of biological machines, which have outperformed conventional technology in many ways, we pursue to create a highly effective sensory system that can function with the same efficiency as biology. Our neuromorphic spiking sensory system incorporates two levels of adaptation to address both static and dynamic variations. The first is self-adaptive, while the second is manually treated in our current design. The essential components of the adaptive neuromorphic spiking sensory system, including the synapse, neuron, adaptive coincidence detection (ACDs) and self-adaptive spike-to-rank coding (SA-SRC) in 4-bit resolution, were integrated into our chip (XFAB CMOS 0.35 mu m via EUROPRACTICE). The presented work focuses on the adaptation of the measurement approach to determine the characteristics, namely integral non-linearity (INL), differential non-linearity (DNL), of the SA-SRC on-chip and its particular information processing.