Hybrid Architecture for Handwriting Perceptual Service based on Edge Computing

Handwriting is a natural way to express idea and perceptual service of daily handwriting is in great need. However, current cloud computing paradigm suffers from slow response, excessive networking bandwidth surge and vulnerable security. We propose a hybrid computing architecture incorporating edge computing and wireless sensor network to deliver handwriting perceptual service. Instead of uploading all raw data to the cloud, we offload the handwriting detection and text recognition services to the edge devices. Most bandwidth can be saved. The edge devices are organized to form wireless sensor networks, which possesses high resilience to disasters or damages. Merely the semantic texts than handwritten images are submitted to the cloud and resultant prompts or decisions are sent back. Our computing architecture in general has a fast edge intelligence response with balanced energy utilization. We implement a specific case study to investigate its feasibility. The result show that our design works pretty well.