A Blockchain-Enabled Security Framework for Cloud-Based Sensor Systems with Deep Learning-Driven Attack Classification
TL;DR
A Weighted Symmetric Hashed Blockchain framework that integrates mutual-information-based feature weighting, deep-learning-based attack classification, AES-256-GCM authenticated encryption, cryptographic hashing, and permissioned-ledger logging, demonstrating the potential of WSHB as a reproducible framework for attack classification and secure event logging in cloud-integrated sensor environments.
Abstract
Cloud-integrated sensor and Internet of Things (IoT) systems enable scalable data storage, processing, and intelligent monitoring, but their distributed nature exposes network-flow and host-level data to unauthorized access, tampering, and cyberattacks. This study proposes a Weighted Symmetric Hashed Blockchain (WSHB) framework that integrates mutual-information-based feature weighting, deep-learning-based attack classification, AES-256-GCM authenticated encryption, cryptographic hashing, and permissioned-ledger logging. The framework was evaluated independently using HIKARI-2021 for network-based intrusion detection and ADFA-LD for host-based intrusion detection. A transparent comparative evaluation was conducted against LSTM and CNN–RNN baselines using identical data splits, preprocessing settings, input representations, hyperparameter-search budget, and repeated initialization seeds. The proposed WSHB-DNN classifier achieved macro-F1 scores of 0.6837 on HIKARI-2021 and 0.7914 on ADFA-LD, showing the strongest overall classification performance among the evaluated models. The cryptographic and permissioned-ledger components provide confidentiality protection, record-level authentication, integrity verification, and tamper-evident logging for confirmed attack-event records. These results demonstrate the potential of WSHB as a reproducible framework for attack classification and secure event logging in cloud-integrated sensor environments.