Band Assignment in Ultra-Narrowband (UNB) Systems for Massive IoT Access

TL;DR

This paper proposes a learning-based band assignment strategy for ultra-narrowband IoT networks with hardware constraints, improving packet decoding probability over random assignment through simulation.

Contribution

It introduces a novel learning-based algorithm for band assignment in UNB IoT systems, addressing hardware limitations and maximizing packet decoding probability.

Findings

Learning-based algorithm achieves near-optimal PDP

Outperforms random assignment significantly

Heuristic based on BS locations also outperforms random assignment

Abstract

In this work, we consider a novel type of Internet of Things (IoT) ultra-narrowband (UNB) network architecture that involves multiple multiplexing bands or channels for uplink transmission. An IoT device can randomly choose any of the multiplexing bands and transmit its packet. Due to hardware constraints, a base station (BS) is able to listen to only one multiplexing band. The hardware constraint is mainly due to the complexity of performing fast Fourier transform (FFT) at a very small sampling interval over the multiplexing bands in order to counter the uncertainty of IoT device frequency and synchronize onto transmissions. The objective is to find an assignment of BSs to multiplexing bands in order to maximize the packet decoding probability (PDP). We develop a learning-based algorithm based on a sub-optimal solution to PDP maximization. The simulation results show that our approach to band assignment achieves near-optimal performance in terms of PDP, while at the same time, significantly exceeding the performance of random assignment. We also develop a heuristic algorithm with no learning overhead based on the locations of the BSs that also outperforms random assignment and serves as a performance reference to our learning-based algorithm.