This paper presents a comprehensive study on the modeling of power line communication (PLC) channels in a MIMO (Multiple Input Multiple Output) environment. PLC systems utilize existing electrical infrastructure to transmit data, but the complexity of these channels, characterized by multipath propagation and inter-symbol interference (ISI), poses significant challenges for designing robust and efficient systems. In this study, we rigorously compare two approaches to PLC channel modelling, which are as follows: the empirical approach and the deterministic approach. The empirical approach relies on the analysis of experimental data to derive statistical models of the channel, offering a realistic representation based on concrete observations. In contrast, the deterministic approach employs theoretical principles and electromagnetic equations to model the channel behavior, providing a detailed description of propagation phenomena. The obtained results demonstrate the significant impact of multipath propagation on the performance of PLC communication systems, highlighting the limitations of empirical models in various scenarios and the increased accuracy of deterministic models. This comparative study conducted in this work highlights the advantages and limitations of each approach and proposes solutions to optimize the performance of power line communication networks.