Model-based Network Fault Injection for IoT Protocols

Abstract

IoT devices operate in environments where networks may be unstable. They rely on transport protocols to deliver data with given quality-of-service settings. To test an implementation of the popular MQTT protocol thoroughly, we extend the model-based test framework “Modbat” to simulate unstable networks by taking into account delays and transmission failures. Our proxy-based technology requires no changes to the IoT software, while the model allows the user to define stateless or stateful types or fault patterns. We evaluate our methods on a client-server library for MQTT, a transport protocol designed for IoT.