Abstract

The implementation of a LoRa-based wireless sensor network, aiming to operate as an edge-based anomaly detection system is elaborated in this paper. The dominant prerequisites are defined as achieving a maximum possible sensing rate on edge, minimum Communication packet loss, the lowest possible false alarm, and the optimal resource management. In this regard, the developed sensor-side algorithm manages measurements accordingly to make effective communications with the center while transmitting either notification or requested compressed data packets. On the other side, a central algorithm keeps all the authorized sensors synchronized and updated. Besides, the center avoids network congestion primarily by queuing the received event notifications and later, by transmitting the data queries consecutively under supervision. The re-construction of the received anomaly-related data packets for further evaluations is another task to do by the center. Overall, the complementary proposal is experimentally tested and validated through the network operation in a worst-case scenario