With the advent of 5G, there has been a significant surge in the deployment of Internet of Things (IoT) devices. This proliferation, while promising, has simultaneously highlighted crucial security vulnerabilities of IoT devices due to their inherent lack of robust security features. This research focuses on enhancing the security of the IoT ecosystem through continuous monitoring to identify potential network problems and cyber threats, which often present themselves as anomalous network patterns. This paper evaluates seven unsupervised Machine Learning (ML) models for anomaly detection in Wi-Fi networks, utilizing data from an IoT test bed under both normal and anomalous conditions. The models are trained on real, unlabeled IoT network data, aiming to detect irregularities in link quality and connectivity. The findings are based on the assessment of these models using a separate, labeled dataset for validation. The models are subsequently deployed and assessed within the IoT system through real-time evaluation and their performance is compared. This method improves both the security and operational reliability of IoT networks by proactively identifying and resolving issues, thereby reducing downtime and ensuring the continuous operation of IoT systems.