Path Extension Analysis of Peer-to-Peer Communications in 6LoWPAN/RPL Sensor Networks

Researchers and manufacturers are currently putting a lot of efforts to design, improve and deploy the Internet of Things, involving a significant number of constrained and low cost embedded devices deployed in large scales with low power consumption, low bandwidth and limited communication range. For instance we can easily build a network composed by multiple sensors distributed in a building in order to monitor temperature in different offices. This kind of architecture is generally centralized as all sensors are mainly programmed to periodically transmit their data to the sink. The specific IPv6 Routing Protocol for Low-power and Lossy Networks (RPL) had been designed in order to enable such communications. Support for point-to-point traffic is also available. In fact new applications may also consider peer-to-peer communications between any nodes of the network. In that case, RPL is not optimal as data packets are forwarded in respect with longer paths with larger metrics. In this paper we propose to study the effectiveness of RPL compared to a shortest path algorithm such like the Dijkstra's algorithm. We suggest to analyze peer-to-peer communications inside random wireless sensor network topologies with size limited to 250 nodes, corresponding to a reasonable cluster size. We have built a particular simulation environment named Network Analysis and Routing eVALuation (NARVAL). This toolbox permits to generate random topologies in order to study the impact of routing algorithms on the effectiveness of communication protocols. In our work, we first generated many random network topologies where we selected a sink node. We built the Destination Oriented Directed Acyclic Graph (DODAG) from the chosen sink in respect with the RPL algorithm. We finally performed all paths between each couple of two distinct sensor nodes and compared them to the corresponding shortest paths obtained by the Dijkstra's algorithm. This approach permits to retrieve some statistics on the path extension between RPL and the Dijkstra's algorithm. We also analyzed the impact of the sink position and the network size on this path extension.