Intelligent Signal Processing at the Sensor Network Edge
Sensors are being used everywhere with new applications such as monitoring office rooms for building automation, increasing uptime in an industrial factory and monitoring the border for homeland security. Ideally, these new applications have potential to flood an area with an array of sensors. In order to do this an order of magnitude in cost reduction must be obtained while improving accuracy, robustness and future proofing the sensor network for changes in configuration.
While many technology standards are being introduced to address these challenges such as Wireless Sensor Networks using Zigbee for the RF protocol (based upon IEEE 802.15.4) and Sensor Plug and Play for a standard sensor interface (IEEE 1451.x), the goals of future sensor networks cannot be fully realized without going to a distributed processing network topology. This will require placing intelligent signal processing near or at the sensor before sending data across the communications link. In this paper industrial machine monitoring is used as an application example to illustrate why intelligent signal processing is needed at the edge and proposes a solution to the problem.