E. I. Shkelev, V. N. Bugrov, N. E. Zemnyukov

The discussed article based on the calculation of mutual function of uncertainty (VFN) approach to the treatment of complex noise-like signals (NLS) of rapidly moving objects in the environment of noise n(t) ( u(t) = s(t) + u(t) ), when due to the movement amplitude A(t) and phase (t) of a useful component of the signal s(t) = A(t)exp(j2f0t + (t)) at the receiving end are subject to strong changes. Processing method based on the adjustment of the amplitude U0(t) and phase 0(t) of the reference oscillation u0(t) = in the VFN in order to achieve compliance with the amplitude U(t) and phase (t) variation laws of the useful component s(t). Solved by adjusting the mechanism task is formulated as a task of finding (synthesis) of the reference signal u0(t) parameters, giving a maximum |R(, )| at zero bias with respect to time  and frequency . In anther words, it is determined a reference signal, in which the function R(, ) is close to or coinciding with the Woodward uncertainty function of s(t) signal. The problem of the reference signal u0(t) synthesis is solved by means of search parametric synthesis method. Synthesized parameters are n uniformly distributed reference frequencies fi (i = 1, 2, ? , n) on the time interval of the signal u0(t) existence. Starting fi values (a vector F = (f1, f2, f3,, fn) components) were chosen on the basis of time-frequency profile of the original reference signal, of a signal with a constant f = f0 frequency, in particular. At the each search step in accordance with the selected algorithm a reference signal u0(t) is generated, the distribution of the module VFN |R(, )| on the (, ) plane is calculated, its current maximum value |R(c, c)|, and the coordinates of the maximum c and c are determined. The obtained values of Rc, c, and c re-transmitted to the search program, and then the previously generated values of |R|, , and  are taking into account for definition of a new vector F. This continues until, until you find the reference signal u0(t), giving the zero - offset in the Rm = |R(0, 0)| maximum point. Thus, the search result is synthesized signal u0(t), which displayed typical useful signal s(t) characteristics. Numerical simulation confirmed the possibility of s(t) identification through the synthesis of the reference oscillation u0(t), that have similar to the s(t) properties. It should be noted that this method of determining the characteristics of s(t) requires a large amount of computation and is suitable primarily for signals analysis in those cases when not required to take decisions in real time, and the received signal u(t) may be previously stored in the memory. However, the problem of real time can be solved with the help of parallel computers, for example, matrix processors or multiprocessor systems.