No abstract

No abstract

Jamming is an easy to execute attack to which wireless sensor networks are extremely vulnerable. If the application requires reliability, jamming needs to be detected and reported in order to cope with this attack. In this article, we investigate different approaches to identify jamming. Available jamming detection schemes primarily suffer from the usage of fixed thresholds as well as required effort. We adapted a variance-based estimate of signal-tonoise ratio measurements, called significance analysis, to the minor resources and computing efforts of wireless sensor nodes. As a start, we used real measurement data for theoretical analysis of the methods under investigation. Independently of the location of the jamming device, our significance analysis approach provides an immediate indication of jamming and can in theory be run with almost least effort, i.e., with O(14). On top of that, we implemented this approach on our state of the art sensor node and tested it in a real world outdoor setting. Our jamming detection engine monitors the wireless channel with a sampling rate of 10 ms. It returns a jamming detection decision within less than 5 ms while though achieving a detection accuracy in between 84 to 99 percent.

Wireless Sensor Networks (WSN), the Internet of Things (IoT) etc. are built using resource constraint devices. The hardware is compiled of different types of micro-controllers and radio frontends on which a plethora of operating systems and protocols is deployed. This poses a huge challenge when developing an intrusion detection system (IDS) that shall be applicable for IoT and WSNs. In order to facilitate such an IDS that is independent of the target platform we propose a security interpreter. In this paper we introduce its concept and architecture and discuss performance parameters such as memory footprint and execution times of different virtualization techniques. Our measurement results indicate clearly that virtualization is feasible. Executing a single instruction takes only 4.2 micro seconds and 1.0 micro seconds in the worst case and the best case, respectively.

This paper shortly introduces our real-time jamming detection approach which can be executed on standard wireless sensor nodes. The benefits are that no thresholds need to be defined since it detects jamming based on deviations of the Received Signal Strength Indication (RSSI) and the fact that for doing so it needs only 422 Bytes of memory including execution code and stored RSSI values. Our mock-up demonstrator visualises how various attacks of permanent, periodic and random jamming effect RSSI values and how the sensor nodes independently of the location of the jammer reliably indicate ongoing jamming.