Numerič ne metode končnih razlik za modeliranje telekomunikacijskih kanalov: pristopi in izzivi.

Wireless networks of the next generation should optimally adapt themselves to the operating environment in order to meet the requirements of a significantly higher throughput at a lower energy consumption and at increased utilization of radio resources. The current methods used for the radio environment modelling are insufficiently accurate. Older empirical models can no longer be used to model channel spatial and temporal characteristics because of their inadequacy. The widespread deterministic modelling based on geometric optics significantly deviates from the radio frequency diffraction and scattering behavior. On the other hand, the numerical finite-difference techniques based on the Maxwell’s partial differential equations of electrodynamics do provide the required level of details but are limited to rather small computation domains. However, the increasing number of the access points and the associated reduction in the wireless cell reach suggest to use numerical techniques as a viable alternative to the ray optical modelling of the telecommunication channels. The paper reviews the current challenges and approaches to solving the large-scale telecommunication problems by using the numerical finite-difference techniques and proposes to adapt them because of their considerable acceptability. The techniques complemented with the use of hardware accelerators can become an efficient accuracy-improved alternative of the deterministic models. [ABSTRACT FROM AUTHOR]