Congestion-Aware Routing in Dynamic IoT Networks: A Reinforcement Learning Approach

TL;DR

This paper introduces a reinforcement learning-based routing method for IoT networks that dynamically balances load, significantly improving packet delivery and delay performance under heavy traffic conditions.

Contribution

It proposes a novel Q-learning framework integrated with RPL to optimize parent selection based on congestion, link quality, and hop distance in dynamic IoT networks.

Findings

Enhanced packet delivery rates

Reduced average delay

Marginal increase in signalling frequency

Abstract

The innovative services empowered by the Internet of Things (IoT) require a seamless and reliable wireless infrastructure that enables communications within heterogeneous and dynamic low-power and lossy networks (LLNs). The Routing Protocol for LLNs (RPL) was designed to meet the communication requirements of a wide range of IoT application domains. However, a load balancing problem exists in RPL under heavy traffic-load scenarios, degrading the network performance in terms of delay and packet delivery. In this paper, we tackle the problem of load-balancing in RPL networks using a reinforcement-learning framework. The proposed method adopts Q-learning at each node to learn an optimal parent selection policy based on the dynamic network conditions. Each node maintains the routing information of its neighbours as Q-values that represent a composite routing cost as a function of the congestion level, the link-quality and the hop-distance. The Q-values are updated continuously exploiting the existing RPL signalling mechanism. The performance of the proposed approach is evaluated through extensive simulations and compared with the existing work to demonstrate its effectiveness. The results show that the proposed method substantially improves network performance in terms of packet delivery and average delay with a marginal increase in the signalling frequency.