In the dynamic realm of Internet of Things based Cognitive Radio (CR-IoT) networks operating within 5G framework, we confront challenges associated with Beam Division Multiple Access (BDMA) in the context of massive Multiple-Input Multiple-Output (MIMO) systems. BDMA, a contemporary concept facilitating the simultaneous transmission of multiple users’ data streams through distinct beams, brings forth substantial advantages. However, as the number of transmit antennas grows, BDMA may grapple with an amplified Peak-to-Average-Power Ratio (PAPR). In response to this challenge, our work introduces an innovative approach that explicitly integrates PAPR constraints into the BDMA design, leveraging the concept of mutually orthogonal Graeco-Latin squares. Our approach involves the simultaneous optimization of hybrid digital and analog precoding. To facilitate the hybrid precoder design, our approach harnesses the power of the Graeco-Latin Square Approach, crafting an orthogonal user beam scheduling scheme. For determining the most efficient combination of precoding techniques and beam scheduling strategies, orthogonal-triangular decomposition is used. This algorithm strategically prioritizes user selection, followed by the derivation of their corresponding hybrid precoders, all while conscientiously considering explicit PAPR constraints. Validation through simulation results underscores the effectiveness of our proposed PAPR-aware hybrid approach of Graeco-Latin Square and orthogonal-triangular decomposition scheme, highlighting its potential to alleviate challenges associated with increasing PAPR in massive MIMO systems. This research significantly contributes to ongoing endeavors dedicated to enhancing the performance and efficiency of CR-IoT networks within the dynamic landscape of 5G communication.