Comparison of estimation schemes for the K-distribution shape parameter

Three commonly used estimators together with the maximum likelihood (ML) estimator for the shape parameter of the K-distribution are compared and evaluated. In particular the ML estimation scheme, which provides the most accurate estimate in theory, but which is rarely used in practice, is included and serves as a gage to the other schemes. Distributions of the estimates, as well as other statistical properties, are analysed. Advantages and shortcomings of the different estimators are discussed.