Children safety is a big concern to parents. Whether their child is feeling depressed, being attacked or bullied by other kids, at health or physical risk or even being accompanied by an unauthorized person. The currently available smartwatches designed for kids are either using a threshold-based approach which requires pre-configuration or they require intervention to activate the emergency reaction. These approaches lack in providing personalization for each user data and their protecting actions are based on the preconfigured rules only. In this research, we discuss the child safety issues and introduce a design for a solution based on anomaly detection approach through a smartwatch which is designed for children protection. Anomaly detection approaches can be seen in domains such as big data, intrusion detection and in sensor networks which are discussed in this research. In our approach, we are utilizing the anomaly detection approach for the data collected from the hardware non-invasive sensors embedded into the smartwatch to create personalized rules for each user. These rules are continuously compared to the live data to isolate the discovered anomalies. This approach design ensures much better protection decisions comparing to the currently available solutions. © 2018 Institution of Engineering and Technology. All rights reserved.