Sleep disorders are increasing worldwide and are associated with significant health risks, including cardiovascular disease, anxiety, and reduced cognitive performance. Polysomnography (PSG) remains the clinical gold standard for sleep assessment; however, its cost, technical complexity, and laboratory dependence limit continuous home-based use. Consumer wearable devices provide partial alternatives but often rely on proprietary algorithms and demonstrate moderate agreement with PSG, particularly in detailed sleep staging. This paper presents a low-cost, open-source IoT-based multisensor system for real-time home sleep stage monitoring. The proposed system integrates motion sensing (MPU6050), heart-rate monitoring using photoplethysmography (MAX30100), and infrared temperature sensing (MLX90614) on an Arduino Mega platform. A lightweight rule-based embedded algorithm performs real-time classification of wakefulness, light sleep, and deep sleep without reliance on cloud-based analytics. Preliminary controlled experiments demonstrate the feasibility of basic sleep-stage differentiation under simulated conditions. While the evaluation does not constitute clinical validation, the results support further development through extended overnight testing and comparative analysis with reference systems.