How Can Machine Learning Impact on Wireless Network and IoT? – A Survey
Swarnalina Laha, Nilanjan Roy Chowdhury, Raja Karmakar
Abstract
Machines can be trained by human beings by feeding them massive amount of data, like examples and instructions. They have to learn without human interventions. The Internet of Things (IoT) is still in its outset and is in the developing front. IoT consisting of devices/machines with diverse sensors associated together in wireless networks, needs great amount of security for avoiding unauthorized access to the network as it collects and transmit data packets. With industrial fields showing much interest in the forefront, intrusion threats can make it vulnerable. In this paper, we are addressing several positive measures regarding the use of machine learning (ML) modalities like supervised learning (SL), unsupervised learning (UL), semi- supervised learning (SSL), reinforcement learning (RL), support vector machines (SVM) which are used for training IoT devices. This paper also addresses a brief survey of the scenarios where machine learning algorithms have been applied in various IoT devices and networks such as signal authentication, security, network flow, information retrieval, traffic management, spectrum sensing. It also indicates the necessity of surveying the scattered works on machine and deep learning applications for various aspects like security, congestion control over wireless computer networks, embedded sensor systems for IoT applications. Several limitations of existing researches and a number of open research issues which fellow researchers may find useful in the future have also been discussed.