Interfacing PDM sensors with PFM spiking systems: application for Neuromorphic Auditory Sensors

In this paper we present a sub-system to convert\n\t\t\t\t audio information from low-power MEMS microphones with\n\t\t\t\t pulse density modulation (PDM) output into rate coded spike\n\t\t\t\t streams. These spikes represent the input signal of a Neuromorphic\n\t\t\t\t Auditory Sensor (NAS), which is implemented with Spike\n\t\t\t\t Signal Processing (SSP) building blocks. For this conversion, we\n\t\t\t\t have designed a HDL component for FPGA able to interface\n\t\t\t\t with PDM microphones and converts their pulses to temporal\n\t\t\t\t distributed spikes following a pulse frequency modulation (PFM)\n\t\t\t\t scheme with an accurate configurable Inter-Spike-Interval. The\n\t\t\t\t new FPGA component has been tested in two scenarios, first as a\n\t\t\t\t stand-alone circuit for its characterization, and then it has been\n\t\t\t\t integrated with a full NAS design to verify its behavior. This\n\t\t\t\t PDM interface demands less than 1% of a Spartan 6 FPGA\n\t\t\t\t resources and has a power consumption below 5mW.