Rider Optimization With Deep Learning Based Image Encryption for Secure Drone Communication

Raja, N. Kannaiya; Lydia, E. Laxmi; Acharya, Thumpala Archana; Radhika, K.; Yang, Eunmok; Yi, Okyeon · 2023 · DOAJ

DOI: 10.1109/ACCESS.2023.3324068

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the critical security vulnerabilities inherent in drone (Unmanned Aerial Vehicle) communication systems, particularly in emergency monitoring scenarios where conventional terrestrial networks fail. Due to their reliance on wireless connections and limited onboard resources, drones are susceptible to cyberattacks and data interception. To mitigate these risks, the authors propose the Optimal Deep Learning with Image Encryption-based Secure Drone Communication (ODLIE-SDC) technique. The primary objective is to ensure secure transmission of surveillance imagery while simultaneously enabling accurate classification of emergency-related content. The ODLIE-SDC framework integrates a multi-stage process involving encryption, key optimization, feature extraction, and classification. For image encryption, the method employs a hyperchaotic map-based scheme that utilizes five-dimensional chaotic sequences to perform pixel permutation and diffusion, ensuring high secrecy. The optimal keys for this encryption are generated using a Rider Optimization Algorithm (ROA), a metaheuristic inspired by the dynamics of motorcycle racing groups (followers, bypass riders, overtakers, and attackers). The ROA optimizes key selection by maximizing the Peak Signal-to-Noise Ratio (PSNR). For the classification component, the system uses an EfficientNet-B4 model enhanced with Convolutional Block Attention Module (CBAM) for feature extraction. Hyperparameters for this model are tuned using Bayesian Optimization (BO) to minimize classification error rates. Finally, an Enhanced Stacked Autoencoder (ESAE) is used for image classification, designed to preserve original information during feature extraction to improve accuracy. Experimental validation was conducted using the AIDER dataset. The study reports that the proposed ODLIE-SDC technique achieves enhanced performance compared to existing approaches in both secure communication and image classification tasks. The integration of ROA for key generation and BO for hyperparameter tuning allows the system to balance computational efficiency with security robustness. The hyperchaotic encryption ensures that transmitted images are resistant to unauthorized access, while the deep learning pipeline effectively categorizes emergency scenarios. The significance of this work lies in its comprehensive approach to securing drone-based emergency response systems. By combining advanced chaotic encryption with optimized deep learning models, the ODLIE-SDC technique provides a viable solution for protecting sensitive surveillance data in disaster management and border surveillance applications. The findings suggest that metaheuristic optimization algorithms can effectively enhance both the security parameters of encryption schemes and the predictive accuracy of deep learning classifiers, offering a robust framework for future secure UAV communication systems.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• distraction detection algorithms