Situation Detection on the Edge Nikos Papageorgiou 1 , Dimitris Apostolou 1, 2 , Yiannis Verginadis 1 , Andreas Tsagkaropoulos 1 , Gregoris Mentzas 1 1 National Technical University of Athens, 9 Iroon Polytechniou str., 157 80 Zografou, Athens, Greece {npapag, jverg, atsagkaropoulos, gmentzas}@mail.ntua.gr 2 University of Piraeus, 80 Karaoli & Dimitriou str., 185 34, Piraeus, Greece dapost@unipi.gr Abstract. Situation Awareness in edge computing devices is necessary for detecting issues that may hinder their computation capacity and reliability. The Situation Detection Mechanism presented in this paper uses Complex Event Processing in order to detect situations where the edge infrastructure requires an adaptation. We designed the Situation Detection Mechanism so as it is modular and can be easily deployed as a Docker container or a set of Docker containers. Moreover, we designed it to be independent of Complex Event Processing libraries and we have shown that it can operate with both the Siddhi and Drools libraries. We evaluated our work with a real-world scenario indicative of the usage of our component, and its capabilities. Keywords: situation awareness, complex event processing, edge computing 1 Introduction Mobile Edge Computing (MEC) enables a computing and storage infrastructure provisioned closely to the end-users at the edge of a cellular network. Combining MEC in multi-cloud infrastructures can help to combat latency challenges imposed by cloud-centric architectures. However, edge devices are highly dynamic in nature: they are not as reliable as server and cloud computing resources; they computing capacity is limited and varies greatly depending on their workload; their operating environment (temperature, humidity, etc.) may impact their performance; they often include sensors which send data at a very high rate which can sometimes swamp the available network bandwidth. Perception of these elements in the environment of edge devices within a volume of time and space and the comprehension of their meaning is typically referred to as ‘situation’ [6], [7]. Situation detection can enhance the capacity to manage edge resources effectively as part of a computing environment. Situations in edge computing infrastructures are highly related to the current status and context of edge devices and behavior of deployed applications. Situations are in a rich structural and temporal relationship, and they are dynamic by definition, continuously evolving and adapting. To cope with the dynamicity of situations, one needs to sense and process data in large volumes, in different modalities [18]. To realize systems for Situation Awareness (SA) “individual pieces of raw information