Learning How to Configure LoRa Networks With No Regret: A Distributed Approach

Long range (LoRa) is one of the most popular technologies for low-power wide area networks. It offers long-range communication with a low energy consumption, which makes it ideal many applications in Internet Things. The performance LoRa networks depends on parameters used by individual nodes. Several works have proposed different solutions, typically running central network server, to select these parameters. However, existing approaches not addressed need (re-)assign when channel conditions suddenly vary due additional traffic, changes weather or presence obstacles. Moreover, allocation strategies that require entity decide do scale large number configuration packets must be sent To address issues, this article proposes NoReL , distributed game-theoretic approach allows nodes autonomously update their and maximize packet delivery ratio. based stochastic variant no-regret learning, proven reach an $\epsilon$ -coarse correlated equilibrium Extensive simulations show achieves higher ratio than state art both static dynamic environments, improvement up 12%.