I. J. Computer Network and Information Security, 2022, 3, 63-74 Published Online June 2022 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijcnis.2022.03.05 Copyright © 2022 MECS I.J. Computer Network and Information Security, 2022, 3, 63-74 Cascade Forward Neural Networks-based Adaptive Model for Real-time Adaptive Learning of Stochastic Signal Power Datasets Odesanya Ituabhor Department of Physics, Federal University Lokoja, Lokoja, Kogi State, Nigeria E-mail: Ituabhor.odesanya@fulokoja.edu.ng Joseph Isabona and Jangfa T. zhimwang Department of Physics, Federal University Lokoja, Lokoja, Kogi State, Nigeria E-mail: Joseph.isabona@fulokoja.edu.ng, Ituabhor.odesanya@fulokoja.edu.ng Ikechi Risi Department of Physics, River State University Port Harcourt, Nigeria E-mail: ikechi.risi@ust.edu.ng Received: 25 December 2021; Accepted: 06 February 2022; Published: 08 June 2022 Abstract: In this work, adaptive learning of a monitored real-time stochastic phenomenon over an operational LTE broadband radio network interface is proposed using cascade forward neural network (CFNN) model. The optimal architecture of the model has been implemented computationally in the input and hidden units by means of incremental search process. Particularly, we have applied the proposed adaptive-based cascaded forward neural network model for realistic learning of practical signal data taken from an operational LTE cellular network. The performance of the adaptive learning model is compared with a benchmark feedforward neural network model (FFNN) using a number of measured stochastic SINR datasets obtained over a period of three months at two indoors and outdoors locations of the LTE network. The results showed that proposed CFNN model provided the best adaptive learning performance (0.9310 RMSE; 0.8669 MSE; 0.5210 MAE; 0.9311 R), compared to the benchmark FFNN model (1.0566 RMSE; 1.1164 MSE; 0.5568 MAE; 0.9131 R) in the first studied outdoor location. Similar robust performances were attained for the proposed CFNN model in other locations, thus indicating that it is superior to FFNN model for adaptive learning of real-time stochastic phenomenon. Index Terms: Stochastic phenomenon, Neural networks, Adaptive modelling, Adaptive learning, Practical SINR. 1. Introduction Unlike the wired or cable communication channel that is stationary and relatively predictable, the wireless communication channel is unstable and difficult to analyze due to its stochastic nature [1, 2]. Particularly, radio frequency wireless broadband channels are complex phenomenons that are largely impacted by diverse radio frequency propagation mechanisms, such as scattering, diffraction, and reflection. For instance, reflection occurs when a travelling radio wave comes in contact with a large obstruction. For diffraction, it is experienced when the obstructing object blocks the propagated radio wave path, thus resulting in the meandering of the radio wave around the obstructing object. On the hand, scattering occurs when the communication channel in which the radio waves broadcast are non- uniformities and irregularities owing to different environmental blockades. Owing to these phenomena, copies of received signal and signal quality are random variables and these make them very difficult to be modelled [3]. This phenomenon is also accompanied by the wireless radio communication channel that is frequency selective, time-varying and space-varying. Thus, to effectively model and characterize the signal quality and the entire wireless communication systems performance, the aforementioned possible variation must be well articulated. The concept of applying parametric and non-parametric mathematical models to examine the behavior of time- dependent and stochastic phenomena has been widely established globally. Neural Networks models belong to a family of nonparametric models with massively parallel architectures and computational tools that can adaptively learn and solve complicated engineering problems. Till date ANNs remained a broad and robustly applied learning and modeling tools, ranging from catering for simple to complex function approximation problems, pattern recognition problems,