Revolutionizing 5G: Cognitive Machine Learning
Arun Kumar Singh, Himanshu Katiyar, Rajeev Kumar, Saurabh Dixit · 2020
The usage of mobile phones has witnessed a revolutionary shift with an increasing number of users employing it for accessing the web. Consequentially, there has been an overwhelming increase in data usage. Data is being transmitted as data, voice is sent as data, and video is also transmitted as data. The phenomenal rise in data rates has been dictated by Edholm's law of bandwidth, which is an extension of all-pervasive Moore's law. The leaps taken forward in mobile communication by the advancement of technology have been characterized by transition from one generation to other. The demand for bandwidth has increased manifold times. The conventional method of providing a dedicated channel to a user is no longer in vogue. Instead, we need to devise ways that enable smart spectrum sharing and can help ease the bandwidth crunch. In this article, the key technology drivers for the development of the Fifth Generation(5G) are illumined as envisaged by the International Telecommunications Union (ITU). The role of machine learning in the task of dynamic spectrum sensing and management is being increasingly explored. With the enhanced complexity of 5G the conventional of spectrum sensing and assignment may not suffice. Although, the supervised algorithms have improved the performance, but the unsupervised machine learning algorithms need to be augmented. Hence the role of cognitive machine learning in smart spectrum sharing is envisaged and highlighted.