CELL FOR DISCRETE TIME CELLULAR NEURAL NETWORK

摘要

PURPOSE: A cell, adopted in a discrete time cellular neural network, is provided to perform a multiplication operation by using a distributed process structure consisting of a memory, an adder and an accumulator so that it can reduce a complexity of the cellular neural network and realize a VLSI(Very Large Scale Integration) design. CONSTITUTION: The cell comprises a partial generator(102), an initial value register(104), an accumulator(106), a limiter(108), and an output register(110). The partial generator(102) generates N feedback data, consisting of a plurality of N-1 bit input data offered by adjacent N-1 cells and self feedback data, and a sequential partial multiplication result. The initial value register(104) stores a preset constant bias value. The accumulator(106) adds each partial multiplication result and the constant bias value or a previous addition result, and performs a shift operation for outputting an M bit multiplication result. The limiter(108) corresponds the M bit multiplication result to the bit number of the input data. The output register(110) stores the limited multiplication result, performs a self feedback on the stored limited multiplication result at a next multiplication operation or inputs the multiplication result to other adjacent cell as input data.