Demodulation constraint-aware assignment method for LoRa multi-gateway IoT networks

Facing a massive increase of the amount of IoT devices, many research efforts are directed towards Long Range (LoRa)-based technologies for future Internet-of-Things (IoT) applications. Although various works have considered the design of multi-Gateway (GW) IoT networks, most of them have overlooked the LoRa-specific fundamental constraint of a limited number of demodulators available at each GW. In this work, we aim at maximizing the total amount of demodulated uplink transmissions, while accounting for the hard constraint on the available number of demodulators at each GW. The rationale of the proposed strategy is to pre-assign each end-device to a unique GW, so as to avoid multiple redundant data frames at the network server. To solve this problem, an auction-based algorithm was proposed, as well as its computationally-efficient variant, which partitions end-devices into multiple auction groups that can be run in parallel. Numerical evaluations indicate that the proposed auction-based algorithm outperforms the best conventional algorithm, achieving a gain of at least 90% in the amount of demodulated data frames in all network configurations, while approaching the upper bound in dense GW configurations. In addition, our algorithm is shown to be highly tolerant to errors pertaining to imperfect device ID identification.