Artificial Intelligent Techniques for Wireless Communication and Networking

12Supervised Learning Approaches for Underwater Scalar Sensory Data Modeling With Diurnal Changes

J.V. Anand¹, T.R. Ganesh Babu², R. Praveena^2* and K. Vidhya³

¹Department of Electronics and Communication Engineering, Siddartha Institute of Science and Technology, Chittoor, Andhra Pradesh, India

²Department of Electronics and Communication Engineering, Muthayammal Engineering College, Namakkal, India

³Department of Electronics and Communication Engineering, Saveetha School of Engineering, Saveetha University, Chennai, India

Abstract

Spatial patterns of underwater temperature data, its impact of depth and associated temperature are taken from SEAONE obtained from Temperature Data Logger UA-002. The different geographical locations of data logger and different depths namely 5 and 12 m have been modeled for classification using a supervised learning approach. The first approach deals with inter depth variation profile in line with latitude and longitude in a particular area. Classification is obtained incorporating multilayer perceptron with dependant variables and its covariates to intercept the impact of temperature variations in day and night cycles. The second approach deals with calculating attenuations the temperature coefficient of real data sets and attribute of attenuation in a frequency dependant loss. Thus the idiosyncratic nature of underwater and coefficients of temperature and attenuations is systematically analyzed using realistic temperature data in simulation ...