SecDL: QoS-Aware Secure Deep Learning Approach for Dynamic Cluster-Based Routing in WSN Assisted IoT

Abstract

In WSN-assisted IoT, energy efficiency and security which play pivotal role in Quality of Service (QoS) are still challenging due to its open and resource constrained nature. Although many research works have been held on WSN-IoT, none of them is able to provide high-level security with energy efficiency. This paper resolves this problem by designing a novel Secure Deep Learning (SecDL) approach for dynamic cluster-based WSN-IoT networks. To improve energy efficiency, the network is designed to be Bi-Concentric Hexagons along with Mobile Sink technology. Dynamic clusters are formed within Bi-Hex network and optimal cluster heads are selected by Quality Prediction Phenomenon (QP2) that ensure QoS and also energy efficiency. Data aggregation is enabled in each cluster and handled with a Two-way Data Elimination then Reduction scheme. A new One Time-PRESENT (OT-PRESENT) cryptography algorithm is designed to achieve high-level security for aggregated data. Then, the ciphertext is transmitted to mobile sink through optimal route to ensure high-level QoS. For optimal route selection, a novel Crossover based Fitted Deep Neural Network (Co-FitDNN) is presented. This work also concentrates on IoT-user security since the sensory data can be accessed by IoT users. This work utilizes the concept of data mining to authenticate the IoT users. All IoT users are authenticated by Apriori based Robust Multi-factor Validation algorithm which maps the ideal authentication feature set for each user. In this way, the proposed SecDL approach achieves security, QoS and energy efficiency. Finally, the network is modeled in ns-3.26 and the results show betterment in network lifetime, throughput, packet delivery ratio, delay and encryption time.