In Airshark: Detecting Non-WiFi RF Devices using Commodity WiFi Hardware (PDF), researchers from U Wisconsin (Madison) document a firmware for WiFi access points that can detect and dynamically adjust to interference from vacuum cleaners, baby monitors, and other non-WiFi devices that operate in WiFi's radio spectrum. This kind of thing is the backbone of the theory of cognitive radio: devices that can use software defined radio, phased-array antennas, and cleverness to route around other devices in the band, which may, eventually, enable the a lot more data to occupy the radio spectrum. In Airshark's case, the cleverness is in using the wireless cards on the computers and other devices as a sensing array to triangulate on interference.
Airshark taps into the application programming interface of wireless cards used on access points to gather data about radio frequencies in the surrounding environment. The software has been trained to recognize signatures of various devices, and can pick them out from the ambient radio noise with more than 90% accuracy even if signals from multiple such devices are present.
False positives were .39% for environments with four or more interfering devices and using various signal strengths. The researchers found the rate was .068% for signals stronger than -80dBm. "We also found its performance to be comparable to a commercial signal analyzer," according to their research paper "Airshark: Detecting Non-WiFi RF Devices using Commodity WiFi Hardware."