Transmitter detection in shared spectrum band

A method comprising:

• generating at least one sample corresponding to at least one radar signal or at least one non-radar signal;
• forming at least one spectrogram using time domain and frequency domain characteristics of the at least one sample;
• wherein the at least one spectrogram is formed as a result of subdividing an observation window of the at least one sample into a number of time slots of a given duration, computing a power spectral density for a subset of the time slots having a higher determined energy relative to other time slots, and combining one or more computed power spectral densities of the subset;
• passing the at least one spectrogram to a model to detect a presence of the at least one radar signal and to further classify the at least one radar signal as either interference/noise or radar present, and to estimate a bandwidth of the at least one radar signal that has been detected;
• determining the at least one radar signal to be in-band or out-of-band relative to a shared spectrum band, based on the estimated bandwidth of the at least one radar signal;
• detecting the at least one radar signal exceeding a threshold value of peak-to-average signal-to-interference, per megahertz;
• and
• applying an upper bound to the given duration of the time slots so that the average-to-average signal-to-interference is above a threshold, the given duration defining a width of the at least one spectrogram to provide deep-learning-based transmitter detection capability instantiated for an environmental sensing capability sensor design for a shared spectrum band.