Dynamic link adaptation based on coexistence-fingerprint detection for WSN

Operating in the ISM band, the wireless sensor network (WSN) risks being interfered by other concurrent networks. Our concerns are the technologies that do not perform listening before transmission such as Bluetooth, and the ones that do not detect other technologies due to their channel sensing techniques like WiFi. To overcome this issue a WSN node should be able to identify the presence of such technologies. This will allow deducing the characteristics of the generated traffic of these technologies, and thus the behavior of the channel can be predicted. These predictions would help to trigger adequate reactions as to avoid or synchronize with the concurrent networks. Many works exist on link adaptation, but they concern blind adaptations which are unintelligent and solve momentarily the problem that may reappear over time. In this paper, we perform several experiments on a real testbed to categorize the model of the bit errors in corrupted received packets. These experiments are performed under different conditions of channel noise and interferences. This allows us to identify each corruption pattern as a fingerprint for the interfering technology. Then we propose the Fingerprint Identification Mechanism (FIM) to identify on the fly the source of the corruption. With an implementation on Tmote Sky motes using Tinyos1.x, We demonstrate the use of FIM for link adaptation in a coexistence environment. Our mechanism led to throughput improvements of 87%-100% depending on the transmission rate and channel quality.