Abstract
Braille keyboards depend on external digital signal-processing circuits to interpret multiple tactile key presses and generate corresponding characters. In this work, we introduce a hardware-conscious neuromorphic design for Braille input detection, in which 8-bit Braille codes are represented by deterministic burst spiking patterns, identified by a Spiking Neural Network (SNN) with a differential resistive crossbar. Specifically, our design integrates D-flip-flop input latch, gated burst pattern generation, binary weight assignment, and discrete-time $1^{st}$ -order Leaky Integrate-and-Fire (LIF) neural dynamics. A 108-class Braille dataset representation scheme is tested at various noise levels to determine optimal LIF threshold and crossbar quantization. In addition, transient circuit simulations verify the presynaptic spiking process, which includes a Pierce gate oscillator, a Schmitt trigger, a MOD-3 counter, and pulse-shaping active RC components. The simulation results indicate the viability of a resistor-only SNN architecture for Braille code detection, paving the way for subsequent PCB prototyping and full-scale hardware verification.