This paper presents an energy-efficient design of optical wireless communication (OWC) system for the indoor Internet of Things (IoT) with the assistance of machine learning (ML). A central coordinator (CC) has been proposed to interrogate the IoT devices and control the uplink formations with the prediction of transmission quality using ML classifiers. The passive reflective reflectors (PRF) are utilized in the IoT devices, which replaced the power-consuming active transmitters, formulate the zero-power consuming transmission links. The communication performance of the passive link establishments from the IoT devices have been investigated in terms of quality factor (Q-factor), bit error rate (BER), and signal-to-noise ratio (SNR) under different optical wireless channel conditions and link lengths. The ML classifiers have been evaluated on the prediction of transmission quality, and the results suggested the Euclidean K-nearest neighbor (KNN) with ten number of neighbors for the implementation. The IoT devices located within 1.2 m from the CC require a transmission power of 0.5 mW for links carrying 10 Gbps data, which increases the energy efficiency to 20 Gbps/mW with transmission energy consumption of 0.05 pJ/bit. This significant improvement in energy efficiency and passive communication ensures reliable, and green IoT links suitable for data-intensive indoor applications.

本文提出了一种借助机器学习 (ML) 的室内物联网 (IoT) 光无线通信 (OWC) 系统的节能设计。建议使用中央协调器 (CC) 来询问物联网设备并通过使用 ML 分类器预测传输质量来控制上行链路编组。物联网设备中采用无源反射反射器（PRF），取代了耗电的有源发射器，形成零功耗的传输链路。在不同光无线信道条件下，从质量因数 (Q 因数)、误码率 (BER) 和信噪比 (SNR) 方面研究了物联网设备无源链路建立的通信性能，链接长度。 ML 分类器对传输质量的预测进行了评估，结果建议使用具有 10 个邻居的欧几里得 K 最近邻 (KNN) 来实现。距离CC 1.2 m以内的物联网设备，承载10 Gbps数据的链路需要0.5 mW的传输功率，这将能量效率提高到20 Gbps/mW，传输能耗为0.05 pJ/bit。能源效率和无源通信的显着改进确保了适合数据密集型室内应用的可靠、绿色物联网链路。