Faultload time model of the MQTT protocol publish service

Nowadays, the Internet of Things touches all areas of our daily life, such as industry, economy, energy and agriculture. If we extend these domains to solutions related to smart homes and cars, we will count more than 50 billion connected devices in 2020. These applications transmit a high amount of data on the internet through IoT communication protocols. In some cases, the security aspect is required as the exchanged data can be sensitive. Therefore, it is necessary to develop a means to assess the confidence we can assign to such transmission protocols. In this context, the fault injection characterization mechanism speeds up the fault introduction into a transmission protocol to observe its reaction and to assess its resilience to application conditions with risks of errors occurring. This paper presents a systematic approach to identifying the moment of fault injection in the messaging protocol Message Queuing Telemetry Transport (MQTT). MQTT protocol handles exchanged messages across a distributed system where the injection instant cannot be defined through a time value as the synchronization of the distributed components is not guaranteed. New algorithms are introduced: (1) extract the send/receive messages' pairs, (2) timestamp the communication events using the vector clock, (3) filter the sending events and (4) generate alternate sent messages sequences. Events models for the publisher/broker provided services are generated. These services are: connect, disconnect and publish, obeying some required properties for services' quality.