Visible light communication (VLC) has attracted significant research interest within the last decade due in part to the vast amount of unused transmission bandwidth in the visible light spectrum. VLC is expected to be part of future generation networks. The heterogeneous integration of radio frequency (RF) and VLC systems has been envisioned as a promising solution to increase the capacity of wireless networks, especially in indoor environments. However, the promised advantages of VLC and heterogeneous RF/VLC systems cannot be realized without proper resource management algorithms that exploit the distinguishing characteristics between RF and VLC systems. Further, the problem of backhauling for VLC systems has received little attention. This dissertation’s first part focuses on designing and optimizing VLC and heterogeneous RF/VLC systems. Novel resource allocation algorithms that optimize the sum-rate and energy efficiency performances of VLC, hybrid, and aggregated RF/VLC systems while considering practical constraints like illumination requirements, inter-cell interference, quality-of-service requirements, and transmit power budgets are proposed. Moreover, a power line communicationbased backhaul solution for an indoor VLC system is developed, and a backhaul-aware resource allocation algorithm is proposed. These algorithms are developed by leveraging tools from fractional programming (i.e., Dinkelbach’s transform and quadratic transform), the multiplier adjustment method, matching theory, and multi-objective optimization. The latter part of this dissertation examines the adoption of emerging beyond 5G technologies, such as intelligent reflecting surfaces (IRSs) and reconfigurable intelligent surfaces (RISs), to overcome the limitations of VLC systems and boost their performance gains. Novel system models for IRSs-aided and RISs-aided VLC systems are proposed, and metaheuristic-based algorithms are developed to optimize the configurations of the IRSs/RISs and, consequently, the performance of VLC systems. Extensive simulations reveal that the proposed resource allocation schemes outperform the considered benchmarks and provide performance close to the optimal solution. Furthermore, the proposed system models achieve superior performance compared to benchmark system models.