Introduction

Estimation theories and techniques are at the heart of data processing systems. The data processing methods can be traced back to Gauss, who originated the deterministic least-squares method and employed it in a relatively simple orbit determination problem. An estimator is a procedure that processes measurement to deduce an estimate of the state or parameters of a system from the prior knowledge of the system, measurement, and initial conditions. The most important probability density function of the random variable is the Gaussian or normal distribution. There are two commonly used models for the estimation procedure. The first model is called non-Bayesian or Fisher estimation assuming that the parameters to be estimated are nonrandom and constant in the observation interval but the observations are noisy. The second model is called Bayesian estimation assuming that the parameters to be estimated are random variables with a prior probability distribution, and the observations are noisy as well.