Anomaly detection has various applications including condition monitoring andfault diagnosis. The objective is to sense the environment, learn the normalsystem state, and then periodically classify whether the instantaneous statedeviates from the normal one or not. A flexible and cost-effective way ofmonitoring a system state is to use a wireless sensor network. In thetraditional approach, the sensors encode their observations and transmit themto a fusion center by means of some interference avoiding channel accessmethod. The fusion center then decodes all the data and classifies thecorresponding system state. As this approach can be highly inefficient in termsof energy consumption, in this paper we propose a transmission scheme thatexploits interference for carrying out the anomaly detection directly in theair. In other words, the wireless channel helps the fusion center to retrievethe sought classification outcome immediately from the channel output. Toachieve this, the chosen learning model is linear support vector machines.After discussing the proposed scheme and proving its reliability, we presentnumerical examples demonstrating that the scheme reduces the energy consumptionfor anomaly detection by up to 53% compared to a strategy that uses timedivision multiple-access.
展开▼
