Section: New Results

Sensor networks management

Participants : Cyril Auburtin, Alexandre Boeglin, Olivier Festor, Abdelkader Lahmadi, Emmanuel Nataf [contact] .

6LowPAN networks denotes many embedded devices interconnected by a variety of links ranging from wireless technologies such as 802.15.4, bluetooth, Low Power Wifi to wired technologies such as low power PLC. The common property of such networks is the limited resources of their nodes in terms of power, computing, memory and communication. The network could be described with thousands of devices with very limited internal and external resources and their communication channels are low-bandwith, high loss rate and volatile links subject to failure over time. These networks rely on the 6LowPAN protocol defined by the IETF as an adaptation layer for the IPv6 protocol to address their low power and lossy properties.

During the year 2011, we have started a research activity around the monitoring and security assessment of 6LowPAN networks. Our contributions are mainly as follows:

• We are developing a novel approach to assign monitoring roles in 6LowPAN networks using available local information provided by the routing layer. The resulting monitoring architecture is adaptive taking benefit from the reactivity of the routing protocol when dynamic changes occur due to connectivity or nodes mobility. Our first simulations results reveal that our assignment approach is more efficient, less aggressive and less resources consuming than its competitors.

• We have also designed and implemented a piggybacking technique to deliver monitoring report into existing packets traveling through 6LowPAN networks. In our solution, we have extended the IPv6 Hop-by-Hop extension header with a new option which contains status data of monitored nodes. This technique can reduce the number of packets and bytes sent across the network since there is no specific monitoring packets competing with existing traffic. Monitoring data shares the routing path of application data packets until it reaches a management node. We have applied our piggybacking technique to discover coap-enabled management agents. Each agent in the deployed wireless sensor network piggybacks its identifier into the RPL routing protocol messages until it reaches a manager node.

• Regarding security management of these networks, we have developed a stateless fuzzing tool for the 6LowPAN protocol [28] . The tool is build upon the Scapy packets manipulation library. It provides different mutation algorithms to be applied on 6LowPAN messages. These messages are defined by interaction scenarios described in an XML format.

• Related also to security, we have modelled an ontology for intrusion detection system in sensor networks [17] . The model exposes family of intrusions depending on their objectives. The service provided by the network, the communication channels and the security mechanisms are the main classes of the model.