A smart lighting system leverages the power of IoT (Internet of Things) for optimal management of lighting facilities and substantial energy savings. To ensure the uninterrupted functionality of these systems, it is essential to prioritize detecting anomalies in time. The manual identification of individual anomalous smart lights is challenging when thousands of lights are in the entire system. For this purpose, we have proposed the integration of the LDR sensing system into each smart light to predict the illumination failure in it. LDR module measures the intensity of light emitted by the bulb and facilitates the detection of idle lights and prompt maintenance interventions. This paper presents two approaches for smart light anomaly detection leveraging IoT. One is the point anomaly detection system, which indicates individual lights that are not functioning; the other is the collective anomaly detection system, which sends alerts whenever there is an illumination-related issue with one or more lights in the system but does not specify individual anomalous lights. The performance of the proposed methods is validated by detecting the illumination failure of smart lights in a smart lighting system simulated using the Porteous Professional 8.11 simulator.