The Neural Network for Harmonic Analyses of Discrete Signals in a Residue Number

The majority of modern communication channels are series. In existing parallel channels of communication the series data, introduced in an item notation are transmitted. Accordingly such data are transmitted by signals, introduced in an item notation. Apparently, that for achievement of maximum productivity of parallel computing systems, such as neuro-networks, it is necessary to submit data in the parallel format, for example in a system of residual classes. Accordingly such data should be transmitted by signals, introduced in a system of residual classes. In the basis of the analysis and simulation of communication channels the harmonic analysis, in particular Fourier transforms and Walsh lies. Today these transformations are determined only for signals, introduced in an item notation.

Thus, there is a problem of definition of operations of decomposing of discrete parallel signals in Fourier serieses both Walsh and mining of a neuro-network for their implementation.

In a paper we considered the Fourier and Walsh transformation for discrete signals in a system of residual classes. The relation between these transformations in an item number system and in a system of residual classes was defined. The neuro-net for such transformation was defined