A Holistic Perspective on Next-Generation Wireless Networks: Harnessing Federated Learning for Computational Modelling

The potential of machine learning (ML) and federated learning (FL) to solve computational modelling issues in communication systems and networks has received a lot of attention in recent research. Now, FL is getting a great deal of attention. as a powerful machine learning method to move the processing load to edge devices and avoid security issues with the collection of raw data. This paper investigates the FL in small-cell networks where several base stations (BSs) and a uniform Poisson point process (PPP) with different densities are employed to distribute users. This study provides closed, tractable form expressions for coverage probability and outage probability. It further examines the construction of closed-form coverage likelihood computations utilising feasible approximations and determines the minimum BS density required in small-cell networks to achieve a desired model aggregation rate. Due to unequal sample distribution, user, and MBS deployment, including low performance, outage likelihood, coverage rate, transmission capacity, spectrum efficiency, and lack of data privacy, there is a need to model communication networks and systems. Along with the stochastic approach to designing and simulating communication systems and networks, an FL is designed to address the problem. A stochastic communication network optimisation problem is addressed using the proposed FL technique. In this study, the intractable problem is relaxed, and a closed-form formula for coverage probability is generated by utilising the moment-generating function (MGL) and probability-generating function (PGFL). The proposed scheme and other schemes are compared through numerical results. It becomes clear that the FL outperforms the other methods accordingly.