An Efficient Approach for Discovering Objects in the Internet of Things using Clue-Based Search Engine
Raghu Nandan R, N. Nalini · International Journal of Engineering Trends and Technology · 2022
Internet of Things (IoT) technology is a form of a network employing different data-sensing devices that capture actual information and integrate any physical object with a pre-established protocol. Recently, Deep Learning and its variations have been used to detect objects in IoT. However, recognition of objects connected in the network still has major problems because of its troubled networks. Thus, accuracy is not attained in various methods due to their complex methodology. An effective method for discovering IoT objects using a clue-based search engine is proposed to address these concerns, mainly designed for identifying the objects connected in a network. In the initial phase, pcap files are gathered. The preprocessing phase includes flow generation missing field rejection, and normalization techniques. The RFO-based Deep Neural Network (DNN) methodologies are used to identify the objects connected to the network. The objects Discovery Search Engine in the IoT network are tested to verify its performance. Also, the proposed model is compared with several existing models, namely DBN, ANN, SVM and KNN, to estimate their effectiveness. The classification accuracies of DNN, DBN, ANN, SVM and KNN are 95.3%, 92.6%, 87.4%, 84.8% and 80.3%, respectively. The proposed work is based on the clues specified by the requester, which is the first of its kind so far. It is found that the accuracy of the DNN model gives better results when compared with the other models.