This paper introduces SENMap, a mapping and synthesis tool for a scalable energy efficient neuromorphic computing architecture frameworks. SENECA a flexible architectural design optimized for executing edge AI SNN/ANN inference applications efficiently. To speed up the silicon tapeout and chip design for SENECA, an accurate emulator SENSIM was designed. While SENSIM supports direct mapping of SNNs on neuromorphic architectures, as the SNN/ANN grow in size, achieving optimal mapping for objectives like energy, throughput, area, and accuracy becomes challenging. This paper introduces SENMap, flexible mapping software for efficiently mapping large SNN/ANN applications onto adaptable architectures. SENMap considers architectural, pretrained SNN/ANN realistic examples, and event rate-based parameters and is open-sourced along with SENSIM to aid flexible neuromorphic chip design before fabrication. Experimental results show SENMap enables 40 percent energy improvements for a baseline SENSIM operating on timestep asynchronous mode of operation. SENMap is designed in such a way that it facilitates mapping large spiking neural networks for future modifications as well.

Intra-cortical extracellular neural sensing is being rapidly and widely applied in several clinical research and brain-computer interfaces (BCIs), as the number of sensing channels continues to double every 6 years. By distributing multiple high-density extracellular micro-electrode arrays (MEAs) in vivo across the brain, each with 1000's of sensing channels, neuroscientists have begun to map the correlation of neuronal activity across different brain regions, with single-neuron precision [1]. Since each neural sensing channel typically samples at 20 to 50kS/s with a > 10b ADC, multiple MEAs demand a data transfer rate up to Gb/s [2]. However, these BCIs are severely hindered in many clinical uses due to the lack of a high-data-rate and miniature-wireless-telemetry solution that can be implanted below the scalp, i.e., transcutaneously (Fig. 24.2.1). The area of the wireless telemetry module should be miniaturized to ~3cm² due to neurosurgical implantation constraints. A transmission range up to 10cm is highly desirable, in order to improve the reliability of the wireless link against e.g., antenna misalignment, etc. Finally, the power consumption of the wireless telemetry should be limited to ~10mW to minimize thermal flux from the module's surface area, avoiding excessive tissue heating. Most of the conventional transcutaneous wireless telemetry systems adopt inductive coupling, but the data-rate is limited to a few Mb/s. A near-infrared (NIR) optical transcutaneous TX using a vertical-cavity-surface-emitting laser (VCSEL) [2] demonstrated a data-rate up to 300Mb/s but suffers from a limited transmission range (4mm) and requires a sub-mm precise alignment between the implant TX and a wearable RX. Impulse-radio UWB (IR-UWB) is promising for the targeted requirements [3]–[5].