An IoT-Enabled Edge Computing Framework for Precision Agriculture in Small-Scale Indian Farming

Abstract

Small-scale farming constitutes 86% of India's agricultural sector, yet suffers from
declining productivity (averaging 2.8 tonnes/hectare versus 4.5 tonnes globally)
exacerbated by climate variability and resource inefficiencies. This paper
presents KrishiEdge, a low-cost IoT-enabled edge computing framework
specifically designed for the resource-constrained, fragmented landholding realities
of Indian agriculture. Our system integrates custom-designed, solar-powered soil
sensor nodes (₹1,250 per unit) with an edge processing unit (Raspberry Pi with
custom AI accelerator) that executes lightweight machine learning models for real
time crop health monitoring, irrigation scheduling, and pest/disease prediction. A
novel aspect is the sparse data fusion algorithm that compensates for sparse sensor
deployment (1 node per acre instead of recommended 4 nodes) using spatial
correlation and historical patterns. The edge unit communicates via hybrid
LoRaWAN/2G connectivity, adapting to India's variable rural network
infrastructure. Field trials across 112 farms in Punjab, Maharashtra, and Tamil Nadu
(total 287 acres) over three crop seasons demonstrate water savings of 34–42%,
fertilizer reduction of 28%, and yield improvement of 18–27% for key crops (wheat,
rice, cotton). The system operates at ₹3.2 per day energy cost (including solar
charging) and achieves 94.3% accuracy in detecting fungal infections 5–7 days
before visual symptoms appear. Comparative analysis shows 78% lower operational
cost than cloud-based alternatives while maintaining data sovereignty—a critical
concern for Indian farmers. The framework's modular design and multilingual
interface (Hindi, Tamil, Punjabi, English) enable adoption by farmers with minimal
digital literacy, addressing India's pressing need for scalable, sustainable agricultural
technology.