Abstract
Caching and resource sharing (CRS) optimization is a promising joint technique to stimulate video streaming applications and services (VASs) in 5G networks. Solving the CRS optimization problem by using exhaustive algorithms (EAs) can provide exact optimal results, but it is not good enough to apply to a larger scale of 5G networks, i.e., 5G ultra-dense networks (UDNs) due to high memory and time complexity. In this paper, we study genetic algorithms (GAs) to solve the CRS optimization problem in 5G UDNs at reasonable memory and time complexity for approximated or exact optimal results. To do so, the GAs solution which often finds optimal real values under simple constraints in the form of lower and upper bounds is modified for finding optimal binary values under more complicated constraints by applying penalty method. Simulation results are shown to validate that the GAs are able to solve the CRS optimization problem not only for the optimal results at high accuracy (up to 99.99%) and low complexity compared to the EAs but also for more insightful analyses of system performance in 5G UDNs.