The relentless advancement of computing technologies in the era of the Internet of Things has escalated the demand for high-speed real-time data processing. Conventional cloud-centric architectures face inherent limitations in latency, energy efficiency, and data security, while edge computing emerges as a transformative solution by proximal data processing. Here, we fabricate a high-performance ultraviolet photodetector based on WTe2/GaN van der Waals heterojunction and employ it as the core sensing unit to construct an edge-native intelligence computing system for localized low-latency solution concentration analysis. The photodetector with type-II band alignment shows high responsivity, fast response time, and low dark current simultaneously. Based on that, this system is capable of conducting real-time calculations on the sensing data near the photodetector. By deploying optimized tiny algorithms for the microprogrammed control unit on this system, a high identification accuracy of 92.78% is achieved for Vitamin B2 solutions. The power consumption and latency of this system are 65.94% and 78.9% lower than the results of the cloud computing system, which effectively avoids data security issues through near-sensing calculations. This work not only provides a viable pathway for high-accuracy and low-latency solution concentration sensing but also reveals the great potential of 2D materials in edge computing systems.
Wu et al. (Mon,) studied this question.