Processing visual sensor data in low-power wide area networks 

Low-power wide area networks (LPWAN) have witnessed a rapid evolution, with many important actors massively investing research efforts in this area. These networks connect low-power sensing devices which continuously perform measurements and emit small amounts of data. Important recent LPWAN developments include the LoRa, SigFox and IEEE 802.11ah technologies. From a different perspective, as consequence of precipitous advancements and miniaturization in hardware, sensor devices can be equipped nowadays with low-power cameras operating in various spectra, such as visual or infrared. This has led to the development of visual sensors and brought a new wave of applications in surveillance, environment monitoring, incident detection, wearable sensor networks, and many more. However, visual processing is known to require very large bandwidths and large energy consumption. This lies in absolute contrast with the low-power lowbandwidth profile provided by LPWAN. The fundamental research problem addressed in this project is to find out how visual processing can be deployed over LPWAN. Fundamental research areas covered by the project include network protocol design, distributed visual processing, and data compression.