The reliability and quality of reception of decameter-range signals depend on many factors, the main one being the optimal consideration of the noise environment. A popular model of additive noise is a random process with a Gaussian distribution law, which, however, does not always fully describe the processes in the communication channel. This paper proposes considering a non-Gaussian noise model, described by a finite sequence of cumulants of the 2nd and 3rd orders. In this case, the noise power (2nd-order cumulant) is assumed constant, while the asymmetry coefficient can take arbitrary values and therefore must be measured. The additive interaction of noise with an unknown asymmetry coefficient and a sounding radio signal with known parameters is considered. The method of polynomial maximization is used to synthesize computational algorithms for estimating the asymmetry coefficient. Polynomial algorithms of degrees 3 and 4 are constructed. It is shown that, with increasing polynomial degree, the accuracy of the algorithms improves. Although the obtained algorithms become bulkier and more computationally complex for higher degrees, they provide better accuracy than “classical” algorithms. This improvement in accuracy is especially pronounced when the asymmetry coefficient significantly differs from zero.