Device-to-Device (D2D) communication serves as a pivotal technology in the evolution of cellular networks, presenting opportunities to markedly improve spectral efficiency, minimize latency, and optimize energy consumption through direct interactions between user devices. Nonetheless, the implementation of spectrum reuse by D2D pairs presents significant interference challenges, particularly in underlay communication contexts. This paper presents a robust and energy-efficient resource allocation mechanism utilizing a polynomial-time proportional fair (PF) scheduling algorithm. The framework provides minimum rate assurances for cellular users while adaptively distributing multiple resource blocks to D2D pairs in response to fluctuating channel conditions. A greedy heuristic is utilized to address the mixed-integer nonlinear optimization problem in the context of real-time scheduling. Simulation outcomes indicate that the proposed methodology significantly enhances network throughput, decreases energy consumption, and mitigates packet loss, thereby confirming its relevance for 5G and future heterogeneous network environments.