DQ-NN and phantom routing for enhanced source location privacy for IoT under multiple source and destination

Abstract

The Internet of Things (IoT) is now an essential component of our day-to-day lives. In any case, the association of various devices presents numerous security challenges in IoT. In some cases, ubiquitous data or traffic may be collected by certain smart devices which threatens the privacy of a source node location. To address this issue, a hybrid DL technique named Deep Q Learning Neural network (DQ-NN) is proposed for the Source Location Privacy (SLP) in IoT networks based on phantom routing. Here, an IoT network with multiple sources and destinations is considered first, and then the phantom node is chosen by analyzing neighbor list, energy, distance, and trust heterogeneity parameters. After that, multiple routes are created from the source node to the sink node via the phantom node. Finally, path selection is performed by the proposed DQ-NN. Moreover, DQ-NN is obtained by merging the Deep Q Learning Network (DQN) and Deep Neural Network (DNN). A simulation environment consisting of 150 nodes is created to study the effectiveness of performance and scalability. The proposed novel DQ-NN outperforms other existing algorithms, by recording a high network lifetime is 111.912, a safety period of 664970.7 m, an energy is 0.034 J, and a distance is 56.594 m.