This Ph.D. dissertation addresses the Many-Objective Optimization Problem (MaOP) study to reduce the inter-cell interference and the power consumption for realistic Centralized, Collaborative, Cloud, and Clean Radio Access Networks (C-RANs). It uses the Cell Switch-Off (CSO) scheme to switch-off/on Remote Radio Units (RRUs) and the Coordinated Scheduling (CS) technique to allocate resource blocks smartly. The EF1-NSGA-III (It is a variation of the NSGA-III algorithm that uses the front 1 to find extreme points at the normalization procedure extended in this thesis) algorithm is employed to solve a proposed Many-Objective Optimization Problem (MaOP). It is composed of four objective functions, four constraints, and two decision variables. However, the above problem is redefined to have three objective functions to see the performance comparison between the NSGA-II and EF1-NSGA-III algorithms. The OpenAirInterface (OAI) platform is used to evaluate and validate the performance of an indoor coverage system because most of the user-end equipment of next-generation cellular networks will be in an indoor environment. It constitutes the fastest growing 5G open-source platform that implements 3GPP technology on general-purpose computers, fast Ethernet transport ports, and Commercial-Off-The-Shelf (COTS) software-defined radio hardware. This document is composed of five contributions. The first one is a survey about testbed, emulators, and simulators for 4G/5G cellular networks. The second one is the extension of the KanGAL's NSGA-II code to implement the EF1-NSGA-III, adaptive EF1-NSGA-III (A-EF1-NSGA-III), and efficient adaptive EF1-NSGA-III (A$^2$-EF1-NSGA-III). They support up to 10 objective functions, manage real, integer, and binary decision variables, and many constraints. The above algorithms outperform other works in terms of the Inverted Generational Distance (IGD) metric. The third contribution is the implementation of real-time emulation methodologies for C-RANs using a frequency domain representation in OAI. It improves the average computation time 10-fold compared to the time domain without using Radio Frequency hardware and avoids their uncertainties. The fourth one is the implementation of the Coordination Scheduling (CS) technique as a proof-of-concept to validate the advantages of frequency domain methodologies and to allocate resource blocks dynamically among RRUs. Finally, a many-objective optimization problem is defined and solved with evolutionary algorithms where diversity is managed based on crowded-distance and reference points to reduce the power consumption for C-RANs. The solutions obtained are considered to control the scheduling task at the Radio Cloud Center (RCC) and to switch RRUs.