An evaluation of link estimation algorithms for RPL in dynamic wireless sensor networks Peter Ruckebusch 1 , Jens Devloo, David Carels, Eli De Poorter 1 , and Ingrid Moerman 1 Ghent University iMinds, Department of Information Technology (INTEC), Gaston Crommenlaan 8 Bus 201, 9050 Ghent, Belgium, peter.ruckebusch@intec.ugent.be Abstract. Link estimators are extremely important in dynamic wireless sensor networks for obtaining a good network performance because they drive the decisions made by the routing protocol. Many estimators exist but the quality of their estimation depends on the scenario at hand. In this paper, the impact of the estimator on the network performance is investigated in different networking scenarios. Also the influence of the underlying MAC protocol was evaluated. The evaluation was performed both in simulation and on a real-life testbed. The results clearly show that there is no link estimator that is best in all scenario’s. Another major finding indicates that the results obtained in simulator differ heavily from the testbed results. This illustrates that research findings cannot be solely based on simulation results but also requires real-world experiments. The influence off the underlying MAC protocol is limited on the choice of estimator. Key words: WSN, RPL, dynamic wireless sensor networks, routing, IoT, mobility 1 Introduction Link estimation is extremely important in dynamic Wireless Sensor Networks (WSN) since it drives the decisions made by the routing protocol. Selecting the best path towards the destination is crucial for increasing the overall network performance in terms of throughput, latency and reliability. Moreover, selecting the best links reduces the number of retransmissions and, consequently, the energy consumption. For this reason, many link estimators for WSNs have been proposed [3]. While complex algorithms, taking asymmetrical links into account, perform better com- pared to simple estimators, they cause a higher overhead [10]. More intelligent algorithms, such as Four-bit [7], seem to obtain higher Packet Delivery Ratio (PDR) in networks with a high interference level [9]. In dense networks, how- ever, these algorithms can encounter some scalability issues [5], due to variations in different sources of information causing constantly changing metrics [10].