978-1-4799-4786-7/14/$31.00 c 2014 IEEE Trickle-L 2 : Lightweight Link Quality Estimation through Trickle in RPL Networks Emilio Ancillotti, Raffaele Bruno, Marco Conti Institute for Informatics and Telematics (IIT) Italian National Research Council (CNR) Via G. Moruzzi 1, 56124 Pisa, ITALY Email: {a.ancillotti,r.bruno,m.conti}@iit.cnr.it Enzo Mingozzi, Carlo Vallati Dipartimento di Ingegneria dell’Informazione University of Pisa, Italy Email: {e.mingozzi, c.vallati}@iet.unipi.it Abstract—Lightweight link quality estimation is crucial in wireless sensor networks. Indeed, devices with limited capabilities shall trade off between consuming their resources to maintain a precise view of the neighbours’ link quality and to build routes almost blindly. For instance, the Routing Protocol for Low-Power and Lossy Networks (RPL), which has been recently standardised by the IETF to enable IPv6-based sensor networks, only estimates the quality of the links used to deliver data packets. However, this solution has been demonstrated to cause periods of routing instability and reduced packet delivery rates since it estimates only the quality of utilised links. To address this issue in this work we propose a lightweight link estimation procedure that exploits Trickle-based topology maintenance techniques to simultaneously estimate link qualities and propagate routing information. Our proposed scheme has been integrated in the Contiki’s RPL prototype implementation. Simulation results demonstrate that our proposal is capable of measuring the quality of the links to neighbours with small overhead, which results into better routing decisions and improved packet delivery rates. Keywords-RPL, LLN, Contiki OS, link quality estimation, probing, reliability, Cooja. I. I NTRODUCTION The idea of Wireless Sensor Networks (WSNs) as isolated groups of nodes that implement proprietary networking pro- tocols and are incapable of communicating with the rest of the world has come to an end. The Internet of Things (IoT) vision foresees sensor networks seamlessly connected and integrated into the Internet together with a new generation of daily objects empowered with communication capabilities, referred as smart objects [1]. A key driver of this integration is the development of network architectures and standards that ensure a full interoperability of interconnected devices, and allow smart objects to communicate using IP, the universal communication protocol. In the past decade it was often argued that a full IPv6 stack is unsuitable for the requirements of small embedded devices. Indeed, sensors and smart objects are usually in- expensive resource-constrained devices that are characterised by limited memory and low computational capabilities [2]. Furthermore, communications between IoT devices are usually performed through low-power wireless technologies, such as IEEE 802.15.4 [3]. On the one hand, those communication technologies guarantee limited energy consumption, which helps to maximise the lifetime of battery powered devices. On the other hand, they are typically characterised by low data rates, high packet loss rates, and instability. However, the increasing interest in connecting WSNs to existing Internet- based services and preliminary studies demonstrating the feasibility of using IPv6 in WSNs [4] has led the Internet Engineering Task Force (IETF) to create a Working Group with the goal of standardising IP-compatible protocols for the class of Low-Power and Lossy Networks (LLNs). One the main results of those standardisation efforts is a standard adaptation layer, called 6LoWPAN [5], which enables IPv6 frames to be delivered on networks characterised by lossy wireless links with low data rate and short frames, such as the IEEE 802.15.4 technology. However, 6LoWPAN is only an adaptation layer, which specifies fragmentation and reassembly mechanisms and compression techniques. As far as fundamental communication aspects, such as routing, forwarding and device configuration, are concerned, another working group from IETF, called ROLL, has recently standardised the IPv6 Routing Protocol for Low Power and Lossy networks (RPL), [6]. RPL is designed with scalability and energy efficiency as main requirements. As a matter of fact, those features are essential in large and dense sensor networks in which the overhead necessary to spread routing information can be so high to saturate the limited bandwidth offered by the wireless medium and to drain the limited amount of power available to each device. Minimising this signalling overhead is of paramount importance to ensure the feasibility of such de- ployments. Among the various techniques employed in RPL to ensure routing scalability of particular relevance is the Trickle algorithm, which is used to control the dissemination of routing control messages within the network. Specifically, Trickle aims at minimising the amount of route updates that are broadcasted by each node by suppressing the transmission of some updates when unnecessary [7]. At the same time, link quality assessment is a critical func- tionality for multi-hop wireless networks. Accurate estimation of the quality of the links is a prerequisite for optimised routing. Typically, most multi-hop routing protocols for sensor networks aim at selecting network paths that minimise the overall number of (re)-transmissions needed to deliver a mes-