New results on Q-routing protocol for wireless networks
Résumé
In the 90s, Q-routing assisted by reinforcement learning was introduced by Boyan and Littman with interesting results in terms of quality of service. Some recent works continue to promote the idea through improvement of the algorithm or specialized extensions. In this paper, we propose a simple modification to workaround the greedy behaviour of Q-routing by considering epoch notion. In comparison with the original Q-routing and the standard OLSRv2 under Qualnet simulator, we show that our extension provides an interesting improvement in terms of packet delivery ratio on the original irregular grid of Boyan and Littman with wireless links.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|