The Clutter Parameter Estimation of K-distribution Based on Origin Moment Derivation

‍ ‍The parameter estimation of K-distribution had great significance for the prediction and estimation of the radar clutter characteristics. Therefore， it was very important to research the parameter estimation of the clutter distribution. The probability density function expression of K-distribution contained the complex function， so the maximum likelihood parameter estimation of K-distributed clutter was difficult to solve. The moment estimation method had the advantage of simple solution. The K-distribution parameter estimation method based on moment estimation often solved the parameters by utilizing equations with different order origin moments， such as the method of second- / fourth- order moment estimation and second- / fractional- order moment estimation. The combination of these different order origin moments would produce errors in the case of limited data length. It was found that the origin moment of K-distribution contained the product of exponential function and gamma function， and the derivatives of these two functions were related to themselves. By calculating the relationship between the origin moment deviation and the origin moment， a radar clutter parameter estimation of the K-distribution method based on the origin moment deviation was proposed. This method estimated the parameters under the condition of the same order origin moment， so it could avoid the estimation error between different order origin moments， and had better estimation performance. By utilizing the simulated data and measured clutter data， the efficiency and accuracy of parameter estimation in this method were analyzed and compared with other moment estimation methods. Both of the experiments showed that this method had 100% estimation efficiency and the higher estimation accuracy. In the moment estimation method， the high-order moment is sensitive to data， and the low-order moment should be selected as far as possible. By reasonably selecting the order k， the ideal estimation results and accuracy can be obtained.