Self-healing mechanisms for IoT networks based on IA paradigms


Principal investigators

  • Rodrigue Domga Komguem, University of Yaounde I, Cameroon
  • Fabrice Valois, AGORA research team, Inria Grenoble – Rhône-Alpes


The Internet of Things (IoT) becomes more and more popular in our everyday life, with appli-cations ranging from smart cities to agricultural monitoring, environment observation, healthapplications, etc. If the IoT paradigm becomes a reality, it is due to the intensive efforts in thecloud computing field, and because the networking architecture and protocols become mature.Three typical wireless networking IoT architectures and protocols are available:i)short range(e.g.Bluetooth Low Energy, IEEE 802.15.4)ii)low power wide area networks (e.g.LoRa,Sigfox))iii)industrial (e.g.IEEE 802.15.4e TSCH, ISA100). Depending on the application re-quirements, these networking architectures and protocols should be optimized and provisionedconsidering a given scenario, in terms of network topology, network dynamics, required qualityof service, etc. A prior configuration phase is mandatory before the deployment. Unfortunately,because the network always operates in an open environment, especially when using the openISM band, the initial configuration leads to non optimal performance. For instance, when de-ploying a LoRaWAN, the network becomes more and more dense, with interfering gateways.The key question is: how to self-adapt the network configuration and the protocol setup tomaintain the required key performance indicators? In IOTA(i), we are focused on the design ofself-healing mechanisms to allow dynamic adaptation of the main parameters of the networkingprotocols. We claim that self-healing is a continuous process based on measurements and mon-itoring of the network behavior. We believe that machine learning will help us to continuouslyoptimize the network.

Key words: Dynamic adaptation, Internet of things, routing protocols, quality of service, performance evaluation

Website: in progress

Comments are closed.