| In wireless digital communications, the sender and receiver typically know the modulation scheme with which they will be communicating. Automatic modulation identification is the ability to identify the modulation in a communication system with little to no prior knowledge of the modulation scheme. Many techniques for modulation identification operate on many assumptions including that the input signal is base-banded, the carrier frequency is known and that the signal is narrow-band (i.e. neighboring signals in the wide-band are excluded). This work provides the blind processing of an arbitrary wide-band signal to allow such assumptions. The challenges of such a front-end or pre-processor include detecting signals which can appear at any frequency, with any band-width at any given time and for any arbitrary duration. This work takes as its input a wide-band signal with a random number of sub-signals, each turning on and off at random times and each at random locations in the frequency domain. The output of the system is a collection of signals corresponding to each sub-signal brought down to base-band, isolated in the frequency and time domains, nominally sampled and with estimates of key parameters. |
