DSP PROJECT PRESENTATION

RECONSTRUCTION OF SIGNALS FROM

IRREGULARLY SAMPLED DATA

Nagamohan Kompella mohi@erc.msstate.edu

Vamsidhar Juvvigunta vamsi@erc.msstate.edu

Balakrishna Nakshatrala balki@erc.msstate.edu

ABSTRACT

This project aims at the implementation of some of the available algorithms for the reconstruction of irregularly sampled, band-limited signals. These algorithms include the Adaptive Weights Method (ADPW), the Projection onto Convex Sets (POCS) based method, and the Wiley/Marvasti (WILMAR) method. This work deals with one dimensional signals. The algorithms are tested on synthetic data generated by the computer as well as real signals (audio signals are considered). The algorithms are compared based on their speed of convergence and robustness. The influence of the auxilary parameters specific to each method is also examined.

DEFINITION OF PROBLEM

Data transmission in noisy channels introduces gaps between signal samples leading to Irregularly sampled signals.
Irregular sampling for compression of signals again leads to Irregularly sampled signals.
Hence, reconstruction of the signal assumes significance.

SOLUTION STRATEGIES CONSIDERED

Adaptive Weights Method (ADPW)

Projection onto Convex Sets (POCS)

Wiley/Marvasti Method (WILMAR)

(All of them fall under the category of Iterative Reconstruction Algorithms)

ITERATIVE ALGORITHMS MADE EASY

ADPW

(whatever balki wants to write here !)

POCS

(whatever vamsi wants to say here!)

WILMAR

RESULTS

CONCLUSIONS