الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 183 |  |  |
| | | from | 183 | | |
Abstract
In the current literature, a variety of methods have been presented for detecting and estimating chirp signals, with deterministic or random parameters, in a noisy environment. Commonly used methods are the periodogram, the Maximum Likelihood (ML) method, the Maximum Entropy (ME) method, and the Wigner-Ville Distribution (WVD).
The periodogram, which is considered a classical method, has a poor performance and resolution, especially for short data records, and is inefficient to compute. The ML and the ME, which are considered as modern methods, have the problem of finding the probability density function of observations under the alternative hypothesis. The WVD requires the use of the analytic signal and tends to have broader spectral lines at the extreme of time segments and is also inefficient to compute.
The approach given in this dissertation is based upon the state space representation of chirp signals. An Autoregressive Moving Average (ARMA) model of order two, which describes chirp signals in an Additive White Gaussian Noise (AWGN) channel and in a Rayleigh Fading Channel (RFC), is obtained. The parameters of the ARMA model are functions of the chirp signal unknowns, the carrier frequency and the bandwidth. A sliding window least square estimate for the chirp signal parameters is obtained.
A comparison between the state space representation method and the previous work is presented for different Signal-to-Noise Rations (SNR). The state space representation method has an advantage over previous methods in that it does not require the use of all the previous data samples to get an estimate of the signal at any time. It uses a small number of data samples, depending upon the sliding window size.