Loading...

Improvement of Level Crossing Sampling’s Performance in Sample Reconstruction, Data Compression and Sampler Stages

Nasiri, Hossein | 2022

432 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55197 (05)
  4. University: Sharif University of Technolog
  5. Department: Electrical Engineering
  6. Advisor(s): Marvasti, Farokh
  7. Abstract:
  8. Level crossing sampling is a sampling method in which a sample is taken whenever signal crosses predefined and specific levels. In this dissertation, some recommendations are made in order to increase the sampler’s performance in the sampling, data compression, and reconstruction stages.The IMATMirror algorithm is introduced in the sample reconstruction stage. This algorithm is derived from the IMAT reconstruction method. However, additional data processing is done in each iteration, causing the reconstructed signal to satisfy some properties of the Level-Crossing samples.In order to solve the problem of level crossing sampling of extremely bursty signals (ECG signals for example), a sampler employing adaptive level crossing with adaptive levels is introduced in the sampler stage. When these types of signals are sampled using a conventional LC sampler, additional samples are collected in short intervals of extreme changes, while no samples are taken in the other intervals, resulting in a slow sample reconstruction. The distance between predefined levels in this proposed sampler is not fixed and adapts to signal changes. A few samples obtained from intervals where the signal is approximately constant will contribute to a more accurate reconstruction.In the data compression stage, two deep networks are introduced to estimate the time interval of the subsequent sample using the previous ones. As a result, the time intervals can be stored differentially utilizing the values estimated by the networks rather than being represented by their absolute values
  9. Keywords:
  10. Deep Learning ; Adaptive Level-Crossing Sampling ; Sparse Signal Reconstruction ; Level-Crossing Sampling ; Data Compression

 Digital Object List

 Bookmark

...see more