The stability of multiple objective RPL tree formation

We address the problem of RPL tree formation in self-organized, multi-hop, wireless sensor networks, where resource-constrained nodes may independently select their routing paths that maximize their performance. We study the result of the tree formation applying a non-cooperative game-theoretic model, and show that multiple objectives may lead to unstable Nash graphs with unwanted traffic cycling. To ensure stability we propose an extension of the node´s strategy space, denoted as selective routing, that efficiently eliminates non-acyclic formations from the set of Nash equilibria, while the resulting routing decisions comply with standard RPL.