Self-managing Internet of Things

Internet of Things (IoT) are in full expansion. Applications range from factory floors to smart city environments. IoT applications consist of battery powered small computing devices (motes) that communicate wirelessly and interact with the environment through sensors and actuators. A key challenge that IoT engineers face is how to manage such systems that are subject to inherent uncertainties in their operation contexts, such as interferences and dynamic traffic in the network. Often these uncertainties are difficult to predict at development time. In practice, IoT applications are therefore typically over-provisioned at deployment; however, this leads to inefficiency. In this paper, we make a case for IoT applications that manage themselves at runtime to deal with uncertainties. We contribute: (1) a set of concerns that motivate the need for self-management for IoT systems, (2) three initial approaches that illustrate the potential of realising self-managing IoT systems, and (3) a set of open challenges for future research on self-adaptation in IoT.