On Interference Detection Using Higher-Order Statistics.

In applications with strict requirements regarding reliability and real time capability such as factory and process automation, it is critical to detect sources of interference that might corrupt data packets leading to retransmissions and delays. To tackle this issue we propose a method for measuring and quantifying radio interference using higher-order statistics. Unlike traditional energy detectors that suffer from several shortcomings in noisy environments, higher-order statistics are robust and do not suffer from threshold uncertainties. We present results of experimental measurements as well as simulations for detecting mutual interference using a normalized fourth-order moment named the kurtosis. The proposed method is independent of the received power and is efficient for detecting low-level interference compared to energy detectors. Results show that not only the presence of interference could be detected but also its strength. As a use case, the proposed method has been applied to Bluetooth mutual interference in this work. In can however be extended to other standards and scenarios.