This paper presents a comprehensive and practical optimization framework for the phased integration of fifth generation (5G) cellular networks over existing heterogeneous infrastructures (2G/3G/4G), utilizing the Syrian national cellular network as a large-scale real-world case study. The research is built upon a massive operational dataset collected, integrated, and geolocalized from the main network operators (Syriatel and MTN) after undergoing extensive preprocessing, cleaning, and geospatial filtering. The primary contribution lies in formulating the multi-criteria optimization problem into a unified scalar objective function using the Weighted Sum Scalarization method and designing and developing two parallel metaheuristic optimization algorithms to solve it: a proposed Adaptive Genetic Algorithm (AGA) and a proposed Binary Particle Swarm Optimization (BPSO), incorporating a robust mathematical constraint repair mechanism. The evaluation framework focuses on three crucial conflicting objectives: maximizing population coverage, minimizing capital expenditure (CAPEX), and optimizing annual energy consumption. Both models were evaluated on a real-world dataset containing over 79,000 active cellular sites (cells). Empirical results demonstrate that BPSO achieves superior performance in computational convergence speed, running 16% faster than AGA, and provides better annual energy savings. Conversely, AGA demonstrates higher statistical stability and lower standard deviation of solutions across multiple randomized runs. These findings validate the proposed approaches as highly effective, reliable decision-support tools for strategic network planning in developing countries with restricted capital budgets, aligning perfectly with recent ITU standards and 5G-Advanced specification.