Interference management for coexisting Internet of Things networks over unlicensed spectrum

TL;DR

This paper investigates interference management strategies for IoT networks operating in unlicensed spectrum, comparing coordinated and uncoordinated access, and introduces reinforcement learning-based coordination to improve network capacity and reliability.

Contribution

It provides a comprehensive analysis of coordination schemes, including a novel reinforcement learning approach, with closed-form capacity expressions and performance tradeoff evaluations.

Findings

Coordination can double the number of connected devices at 1% packet loss.

Reinforcement learning enables distributed coordination among IoT devices.

Performance tradeoffs depend on system and traffic parameters.

Abstract

The main building block of Internet of Things (IoT) ecosystem is providing low-cost scalable connectivity for the radio/compute-constrained devices. This connectivity could be realized over the licensed spectrum like Narrowband-IoT (NBIoT) networks, or over the unlicensed spectrum like NBIoT-Unlicensed, SigFox and LoRa networks. In this paper, performance of IoT communications utilizing the unlicensed band, e.g. the 863-870 MHz in the Europe, in indoor use-cases like smart home, is investigated. More specifically, we focus on two scenarios for channel access management: i) coordinated access, where the activity patterns of gateways and sensors are coordinated with neighbors, and ii) uncoordinated access, in which each gateway and its associated nodes work independently from the neighbor ones. We further investigate a distributed coordination scheme in which, devices learn to coordinate their activity patterns leveraging tools from reinforcement learning. Closed-form expressions for capacity of the system, in terms of the number of sustained connections per gateway fulfilling a minimum quality of service (QoS) constraint are derived, and are further evaluated using simulations. Furthermore, delay-reliability and inter network interference-intra network collision performance tradeoffs offered by coordination are figured out. The simulation results highlight the impact of system and traffic parameters on the performance tradeoffs and characterize performance regions in which coordinated scheme outperforms the uncoordinated one, and vice versa. For example, for a packet loss requirement of 1%, the number of connected devices could be doubled by coordination.