Logo image
A Dynamic Quantum-Resistant Selective Encryption Approach for Agricultural Sensors with Limited Resources
Journal article   Open access   Peer reviewed

A Dynamic Quantum-Resistant Selective Encryption Approach for Agricultural Sensors with Limited Resources

Akshay Kappala, Eric Yocam, Nikil Kayastha, Sai Ram Vodnala, Varghese Vaidyan, Gurcan Comert and Yong Wang
IEEE access, Vol.14, pp.34845-34862
02/25/2026

Abstract

agricultural IoT security discrete wavelet transform Encryption Image encryption Image sensors lattice-based cryptography post-quantum cryptography Quantum computing Security security analysis selective encryption Sensor systems Sensors Smart agriculture Temperature sensors Agriculture Cryptography
Agricultural IoT sensors face dual challenges of severe resource constraints and emerging quantum computing threats. Existing encryption schemes either lack quantum resistance or impose excessive computational overhead that is unsuitable for battery-powered field devices. We propose a Dynamic Quantum-Resistant Selective Encryption (DQRSE) scheme that combines the 2D DiscreteWavelet Transform (2D-DWT) with NIST-standardized post-quantum cryptography. The approach selectively encrypts the LL2 sub-band using AES-256-GCM, while protecting high-frequency sub-bands using HMAC-SHA256. ML-KEM-1024 provides quantum-resistant key encapsulation, and ML-DSA ensures authenticated signatures. A novel content-adaptive key rotation mechanism triggers new key generation when Frobenius-norm image differences exceed threshold τ , balancing security against efficiency. Evaluation in the CottonWeedID15 dataset (5,187 agricultural images) demonstrates key selection diversity of 7.88 bits/byte (dynamic) vs 3.32 bits / byte (static), Number of pixels change rate (NPCR) 99.6%, unified average change intensity (UACI) 33.4%, encrypted pixel correlation coefficient <0.005 and minimal computational overhead compared to static approaches 3.32 bits / byte (key selection diversity). DQRSE represents the first quantum-resistant dynamic selective encryption specifically optimized for agricultural sensor networks. Performance evaluation conducted on Debian VMs (4GB RAM, 4 cores) represents idealized lower bounds; the target ARM Cortex-M agricultural sensors (<256KB RAM, 48-168 MHz) will exhibit 25 × higher latency and energy consumption. Future work includes hardware validation on embedded platforms.
url
https://doi.org/10.1109/ACCESS.2026.3668608View
Published (Version of record) Open

Metrics

1 Record Views

Details

Logo image