International Journal of Wireless and Ad Hoc Communication IJWAC 2692-4056 10.54216/IJWAC https://www.americaspg.com/journals/show/1416 2019 2019 Improvement and Enhancement of bandwidth of 5G Networks using Machine Learning Department of Engineering, University of Technology and Applied Sciences, Shinas, Sultanate of Oman Aaras Y Y.kraidi Department of Engineering, University of Technology and Applied Sciences, Shinas, Sultanate of Oman A. Rajalingam Radio-frequency-based systems are exhibiting severe bandwidth congestion as a result of the exponential development in the amount of data flow. Both cognitive radio technology and free-space-optical communication are examples of attempts to find solutions to the problems posed by high data rates and limited spectral bandwidth. Operating an optical wireless transmission system does not need the purchase of a license. Additionally, the accommodation of unlicensed users across the restricted frequency that is accessible to us is the foundation of the technology known as cognitive radio. Since Dynamic-Window Size systems do not need a license, they are very cost-effective, they can be readily deployed, and they provide a high bandwidth; hence, Dynamic-Window Size systems may be used to bridge with the existing Radio Frequency system. Within the framework of the proposed Dynamic-Window-Size system, the Radio Frequency link is modeled based on the Rayleigh distribution, whilst the Dynamic-Window-Size link experiences -IG composite fading. It is possible to determine both the moment-generating function (MGF) and its derivative. By making use of the formulas that were derived from them, various performance metrics, such as ergodic channel capacity, bit error rate (BER), and output power are calculated, along with the validations that are provided by asymptotic findings. In addition to this, a new closed-form identity is discovered that relates to a specific instance of Bessel's function. In addition to the convex optimization that was mentioned above for the purpose of optimizing the overlay and underlay power in the scheme that was presented, the performance of the Cognitive Radio network is evaluated by making use of a variety of pulse-shaping windows. Suppressing the side lobes of the primary users' (PUs') sub-carriers is a way to reduce the amount of interference that primary users cause for secondary users without harming the primary users' own transmissions. This study involves the creation of a variety of pulse-shaping windows across a variety of power allocation systems as well as an examination of how these windows compare to one another. 2021 2021 68 76 10.54216/IJWAC.020205 https://www.americaspg.com/articleinfo/20/show/1416