Sorry I cannot identify the paper for you but I can tell you a few things about Radio Frequency signal intercept and identification (ID). I have worked with this most of my professional career in both industry and academia. I think your question pertains to an area called emitter classification but I will answer the more general question about communication intercept.

To begin, there are three major aspects of communication intercept: 1) emitter identification (also called transmitter ID or user ID), 2) emitter classification, and 2) content identification (what information is being transmitted)

(1) Historically, emitter ID has been around since the early days of radio at the beginning of the 20th century. Emitters had characteristic waveform Idiosyncracies such a power supply hum, rise and fall times in the modulation scheme, frequency instabilities (FMing) etc. Even such things as identifying the human speaker or morse code operator or microphone mechanical sounds were used to identify emitters. Sometimes, capacitors in a transmitter can become "microphonic" and create vibration noise in the transmitter. In covert RF intercept operations, microphonic capacitors have been used to intercept conversations near a strictly digital transmitter. An on, and on. Today, the RFID tag in consumer products is intential emitter identification.

(2) The authors of the paper you referenced may be talking about the identification of a class of emitters (emitter classification). It is true that commercial communication devices such as AM-FM radio, Analog television, Digital Television, Cellular Radiotelephone, Radar, garage door openers, proximity devices (RFID), ect all have a characteristic RF spectral signature and can be classified. Advanced digital systems are harder to classify but it can still be done with commercial systems.

In military and covert communications, the signal waveforms and spectra are specifically designed to protect against emitter ID & classification, and content classification. Communications are also spoofed so that content cannot be determined.

(3) Content identification is very difficult to determine . In the early days of communication it was easier, You could simply intercept the signal and demodulation it to determine content. Imagine looking at the waveform and spectra of an analog TV signal, could you tell if it was I Love Lucy or Big Brother? Highly unlikely except in very special circumstances. One the comes to mind is the pattern recognition software in smart phones that can identify many musical tunes by analyzing them.

In more modern papers on the subject of communication intercept, terms similar to  "Device Identification via Analog Signal Fingerprinting" might be used.

Good luck

Is governed by the ITU, it is known in different ways: as frequency allocation table, table radio frequency or RF frequency table

Also please read this paper: 

Demonstration of Spectral Sensing & Signal Classification

Robert Newgard, Shubha Kadambe, Zaid Towfic and Christina Conway

http://cms.comsoc.org/SiteGen/Uploads/Public/Docs_DYSPAN_2008/RockwellCollins_Demo_IEEEDySPAN2008.pdf