Machine Learning Algorithm Estimation for Spectrum Multiple Access under Simulation and Real Datasets

The paper deals with the main problem of the telecommunications world, the problem described as “the investigation of spectrum multiple access”, which involves different cases of user access to wireless networks (for e.g. Long Time Evolution or 5G New Radio, other wireless communications, etc.). In the present study, simulation and real data were obtained using different design models of signal propagation for the metropolis of Novosibirsk (Russia). Analysis results were obtained for the most used and maintained spectrum range from 2.400 to 2.527 GHz. This paper examines new datasets from real signal propagation for different design models. A comparison of the implementation of conventional access and machine learning algorithms shows the required result, which can be applied in multiple spectrum access technologies such as cognitive radio and others. In addition, the authors consider the probability of using each channel in the range and other parameters helping to estimate the channel or groups of channels for data transition in advance.