The growth of IoT devices and the spread of remote Internet access technologies allow the emergence of several applications. As the connectivity areas expand and technologies evolve, so do the systems and devices that support its infrastructure. However, with the increase in its benefits, we can raise several concerns about network security. In this scenario, most devices have limited hardware resources and opaque security systems. Therefore, in this study, we implement and analyze the performance of a lightweight machine learning-based Network Intrusion Detection System. We adopted the AB-TRAP, which is a framework that enables the use of updated datasets and considers operational conditions, on a Raspberry Pi 4 device, evaluating the device’s CPU, memory, and network performance. The results showed an average CPU usage between 20% and 30%, and no memory overload for the NIDS implementation. Ultimately, the experiment results indicate that the framework implementation is suitable for the chosen device and that the lightweight detection system is viable. Additionally, we created a malicious traffic generation tool, which was used to generate the traffic used in the experiments.