ENHANCING DATA SURVIVABILITY IN UNATTENDED WIRELESS SENSOR NETWORKS - MACHINE LEARNING-BASED CH SELECTION AND OPTIMIZED HYBRID HOMOMORPHIC ENCRYPTION

Abstract

The research focuses on enhancing data survivability in Unattended Wireless Sensor Networks (UWSNs) through the combination of machine learning-based Cluster Head (CH) selection and optimized Attribute-Based Encryption (ABE) with Homomorphic Encryption. Additionally, the integration of blockchain technology is proposed to ensure tamper-proof data storage and provenance. The first aspect of the proposed solution involves utilizing machine learning algorithms, such as Deep Q-Networks (DQNs) or other suitable models, for intelligent and adaptive CH selection in UWSNs. This approach considers various factors, including energy efficiency, network coverage, communication reliability, and node characteristics, to dynamically select CHs. The second aspect focuses on data protection using optimized Attribute-Based Encryption (ABE) combined with Homomorphic Encryption. ABE allows fine-grained access control based on attributes, ensuring that only authorized entities can access specific data. Homomorphic Encryption enables the secure computation of encrypted data, preserving privacy and integrity. The optimization of these encryption techniques aims to strike a balance between security and computational efficiency, ensuring efficient data processing and transmission while maintaining data privacy and integrity. To ensure tamper-proof data storage and provenance, the integration of blockchain technology is proposed. Furthermore, the research incorporates a combination of optimization algorithms, specifically the Seagull Optimization Algorithm (SOA) and the Whale Optimization Algorithm (WOA), to optimize the performance of the proposed solution. These algorithms are utilized to fine-tune the system parameters, enhance the efficiency of CH selection, and improve the overall performance of the UWSNs. Through this comprehensive approach, the research aims to enhance the data survivability of UWSNs by leveraging machine learning-based CH selection, optimized ABE with Homomorphic Encryption, and blockchain technology for tamper-proof data storage and provenance. The integration of optimization algorithms further improves the effectiveness and efficiency of the proposed solution, ensuring robust data protection, latency, and energy consumption in UWSNs.