2020 agenda


Sparse Sensing of RESM Data Using Gabor Transforms

18 Nov 2020
Track 2
The Gabor Transform (GT) representation was designed to represent signals which are relatively, but not absolutely, compact in both frequency and time.  This representation is equivalent to a short-time Fourier transform, but with better mathematical underpinning.

The paper will show that if the data is undersampled, GTs with a reduced frequency range can be created and that if two undersampled GTs are created with different undersampling factors the original data can be reconstructed subject only to the constraint that the signal bandwidth at each time step must be less than the undersampled bandwidth.  It will be demonstrated that wideband chirps, noise bursts and overlapping pulses which are close in frequency, such as are seem from marine radars, can all be sampled at just over one eighth of the Nyquist rate and then successfully reconstructed using this technique. 

The combination of CS and GT therefore has the capability of reducing processing data rates whilst retaining the flexibility needed to build systems with low SWAP-C, making it more practical and cost-effective to field them in order to give warfighters access to the information they need to understand the electromagnetic environment.