The growing connection among IoT and AI poses many challenges that call for novel approaches and rethinking of the entire architecture, communication, and processing to meet the requirements in latency, reliability, and use of resources. Edge computing is a promising approach in this sense. Moreover, it can become beneficial to bring advanced, i.e., AI-based, IoT solutions in areas where connectivity is scarce and where in general resources are limited.In this paper we describe an edge/fog generic architecture to allow the adoption of edge solutions in IoT deployments in poorly connected and resource limited scenarios. To this end we integrated, using microservices, an MQTT based system that can collect ingress data, handle their persistency, and coordinate data integration with the cloud using a specific service called aggregator. The edge stations have a dedicated channel with the aggregator that is based on LoRa to enable long-range transmissions with low power consumption.Some details of the implementation aspects are described along with some preliminary results. The initial tests of the architecture indicated that is flexible and robust enough to provide a good platform for the deployment of advanced IoT services in contexts with limited resources.
CCS CONCEPTS• Networks → Peer-to-peer protocols; Network architectures; • Information systems → Collaborative and social computing systems and tools.