The method and multiple device of a matrix search for simple and complex samples

The research consists in the development of hardware pattern search methods that use the principles of shared access and parallel data processing inherent in associative memory. Reducing the search time is achieved through the formation and parallel processing of a binary (characteristic) matrix of comparisons of pattern symbols and text. A composite pattern has been introduced, it allows flexible description of search terms. The characteristic matrix is in the form of a parallelogram; it consists of rows shifted to the right, starting from the first row. This form allows you to conduct a parallel search for simple and composite patterns on the diagonal elements of the matrix. The developed method supports hardware search in the characteristic matrix. It is distinguished by the simultaneous consideration of local and distributed relationships between the elements of the diagonals and rows of the matrix. Separate calculation of starting values along the diagonals of the matrix allows you to independently search for two types of samples by calculating in the cells of the diagonals of the characteristic matrix of two output search functions. The method has linear time and quadratic hardware complexity. The clock pulse duration is determined by the sum of the delays of the comparison circuit for a pair of symbols, a D-flip-flop and a two-input element I. The homogeneous structure of the matrix search device, the use of standard operations allow the device to be implemented on a promising FPGA element base, which determines its use in high-performance systems for processing and transmitting heterogeneous information.