Quantifying load imbalance: A practical implementation for data collection in low power lossy networks

Abstract

In many-to-one or many-to-few traffic scenarios, It is inevitable that `hot spots¿ will occur where traffic from a number of sources gets accumulated. While these hot spots can not be avoided, they can be mitigated by means of distributing the traffic across forwarders with load balancing techniques. In this paper, we define a load imbalance metric, which is applicable to any tree/hierarchy based data collection and/or dissemination.We show how current load balancing techniques existing in wireless sensor networks literature can not be applied to large scale Low Power Lossy Networks (LLNs) and the Internet of Things (IoT). We thus propose a greedy algorithm, that requires only partial topology knowledge, and works with the IETF standardized Routing Protocol for LLNs (RPL), without adding extra control overhead. By not requiring full topology information or all link states, this approach can work in highly varying link condition and large scale deployments. We also provide worst case run-time complexity of our heuristic and simulation results on realistic topology and traffic profiles to establish the validity of our approach.