Signal synthesis with limited number of basis functions via particle swarm optimization
Feilner, William P.
MetadataShow full item record
The ability to reconstruct a given signal using sets of bipolar impulses, generally smaller in frequency, has been researched. An algorithm based on Particle Swarm Optimization (PSO) was used to calculate the temporal positions and amplitudes of each impulse to match a certain signal. These impulses were then transmitted in an antenna array in an attempt to reproduce the desired signal via constructive and destructive interference in free space. It has been shown, with about 1% accuracy, that a wide range of frequencies can be synthesized with this approach. Waveform shape has also been matched, though to a lesser degree as dictated by the setup. System limitations aside, the PSO is theoretically capable of matching nearly any waveshape given an ideal bipolar impulse as the main input function. It can thus be said that a number of possibilities arise in terms of applicability, which have been examined accordingly. PSO differs from the standard wavelet approach in that both ideal and non-ideal bipolar impulses can be used to simulate a desired waveform. This is advantageous because it allows the experimental setup to be properly accounted for.