А.Yu. Savinkov - Dr.Sci. (Eng.), Associate Professor, Deputy Director of Programs ОАО - Concern "Sozvezdie". E-mail: a.savinkov@mail.ru
Yu.N. Pribytkov - Ph.D. (Phys.-Math.), Associate Professor, Head of Department JSC - Concern "Sozvezdie". E-mail: ypribytkov@kodofon.vrn.ru

In developing algorithms to receive signals of digital data processing systems only additive noise and fading models are often used. The algorithms synthesized for these models become inefficient in case of band-pass and narrowband interference. This is especially typical for wideband communication systems. One of the ways to tackle this problem is to use rejection filters to cancel signal spectral components affected by interference and the following signal processing by means of algorithms synthesized for additive noise and fading. In this paper obtaining potential performance of the digital wideband communication system using the above procedure is investigated. In the literature the synthesis and analysis of band-rejection filters performance is studied in detail for single signals. When operation of this filter is considered in the digital communication system intersymbol interference introduced by the rejector in case of rather narrow rejection band should be taken into account. Therefore the paper proposes a simplified model of the digital communication system signal including the data symbol signal whose parameters should be estimated, thermal noise, band-pass interference and the signal of adjacent data symbols. In the general case this signal can be approximated by a stochastic process. Thus this model contains not only additive interference but also the stochastic process which occurs only beyond the estimated data symbol. In synthesizing the rejection filter a definitive dispersion minimum of the digital communication system information parameter is proposed to be used as optimality criterion. In particular, the paper considers signal amplitude estimation. When this approach is used no knowledge of the band-pass interference distribution law is required. It is shown that the use of this criterion in case of Gaussian noise leads to shifted efficient estimates of signal amplitude. The paper demonstrates that when estimating the signal amplitude the activity of rejection filter and the following correlation processing if equivalent to signal multiplication by some weighting function and its subsequent integration. An integral equation is obtained the solution of which is the optimal weighting function. For the case of digital signal processing an explicit solution to this equation is found. For analog processing an approximate solution is obtain by successive approximation. Appropriate expressions for minimum definitive estimation dispersion are obtained. General solutions are obtained without restrictions of the band width and statistical properties of interference. They can be used to estimate potential performance on reducing the interference impact in the investigated algorithm with the rejection filter. As an example of practical application of obtained general expressions estimation of the data symbol amplitude of the digital signal processing system in the presence of narrowband interference is considered. An explicit form of the weighting function is found. The impact of weighting function duration on the interference cancellation efficiency is analyzed. Potential characteristics of the amplitude in the presence of narrowband interference is obtained taking into account intersymbol interference.