This paper considers estimation of signal parameters from a sequence of independent, identically distributed, and quantized observations. The approach is based on discrete statistical modeling of random processes and channel statistics. Three best asymptotically normal (BAN) estimates are discussed. The maximum likelihood (ML) estimate of a constant signal in additive white Laplace noise is derived. Then, for small signal parameters, a Locally Efficient (LE) estimator is developed. This estimator, as an approximation to the ML estimator, is simple and nonparametric, and is shown to be unbiased, consistent, asymptotically normal, and efficient in the limit when the signal parameter vanishes.
|Original language||English (US)|
|Number of pages||6|
|State||Published - Dec 1 1982|
ASJC Scopus subject areas