Towards Energy Efficient Mobile Data Collection In Cluster-based IoT Networks

Sana Benhamaid (1) , Hicham Lakhlef (1) , Abdelmadjid Bouabdallah (1)

1 Heudiasyc - Heuristique et Diagnostic des Systèmes Complexes [Compiègne]

Mobile data collection is a very efficient solution to gather information from spatially distributed IoT nodes. In order to enhance the efficiency of mobile data collection, the trajectory planning of the mobile node has been widely studied. In the literature, most of the solutions proposed use static and non learning-based approaches. In our paper, we opted for a learning based approach in which we study a trajectory planning problem in mobile data collection for IoT where IoT nodes are organized in clusters. A relay node is chosen from each cluster in order to collect data from IoT nodes and transmit it to the mobile node in its range. In order to plan the trajectory of the mobile node, we train a deep Q-learning network (DQN) with combined experience replay. This solution will allow us to maximize the amount of data collected and reduce the energy consumption. It will also allow us to adapt the trajectory of the mobile node to the environment parameters without doing expensive recomputations and learning.

Mots clés: Internet of Things; mobile data collection; trajectory planning; deep reinforcement learning

Domaines: Informatique [cs]

Sana Benhamaid, Hicham Lakhlef, Abdelmadjid Bouabdallah. Towards Energy Efficient Mobile Data Collection In Cluster-based IoT Networks. 2021 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops), Mar 2021, Kassel, Germany. pp.340-343, ⟨10.1109/percomworkshops51409.2021.9431037⟩. ⟨hal-03519678⟩