Blockchain reinforced task distribution and secure deduplication using adaptive deep reinforcement learning in cluster based fog IoT

Abstract

The fog-assisted Internet-of-Things (IoT) is gaining interest due to its large number of devices, which can lead to more duplicate data transmission over the internet. This paper proposes task distribution and secure deduplication over Cluster-based IoT, implementing four layers: IoT Devices Layer, Fog Layer, Cloud Layer, and Service Layer. In the IoT devices layer, devices sense air pollutants and are authenticated to the cloud server using Edwards Curve-based Elliptic Curve Cryptography (EC-ECC). Adaptive Rewards Optimized Deep Reinforcement Learning (ARO-DRL) is used for cluster-head selection at the first layer. In the fog layer, SHA-3 is proposed for duplicate verification, and the Emperor Penguin Optimization Algorithm is used to choose the best fog node. Packet Scrutinization Algorithm is used in the fog node to analyze packet features, including DDoS attack packets. A proxy server is deployed between the cloud server and fog node for queue modeling. In the cloud layer, a hybrid cloud environment is used to protect organizations' data in a highly secure manner. IoT devices are divided into sensitive and nonsensitive devices, with sensitive data encrypted using RC6, AES, and Fiestel encryption schemes. The overall environment is assumed to be decentralized, with security invoked to IoT devices to provision Quality of Service (QoS) by avoiding attackers. Experiments were conducted and analyzed using NS3 with Java programming, and simulation results showed improvements in average latency, user satisfaction, network lifetime, energy consumption, and security strength.