Extended Topic Trees for Flexible Subscriptions with MQTT

The Message Queuing Telemetry Transport (MQTT) protocol is an established standard for messaging in the Internet of Things (IoT), As with mailing lists, users subscribe to well-defined topics of interest in order to receive desired data from publishers. Given a certain interest in mind, humans are able to reason and select a suitable topic, though the expression of interest may syntactically differ from the topic description. Increasingly, machines play the role of subscribers such as in in-dustry 4.0 and internet of vehicles. For machines, matching a sub-scription interest to an appropriate topic is possible provided the interest and the topic are syntactically identical. This matching becomes impossible in case of any syntactic difference between interest and available topics. In order to enable spontaneous messaging for machines, flexible matching of interest and topics is necessary. Unfortunately, automated and flexible subscriptions are insufficiently addressed in the current MQTT literature. Accordingly, we propose to automatically enrich topics with additional knowledge through open thesauri to create equivalence classes for topics. To ensure high responsiveness, we rely on a graph database for efficient management of the extended topics. Through simulations, we show that our approach enables to match up to 86 % of subscription interests, though they are syntactically impossible.