Abstract
The present paper has considered multithreshold decoders for self-orthogonal codes providing a near-optimal efficiency of the error correction under linear computational complexity. New divergence principle used within construction and decoding convolutional codes has been discussed. The paper has shown that usage of such principle allows significantly approximating an area of the decoder effective operation to the channel capacity. Application of obtained self-orthogonal codes within construction of concatenated convolutional codes where parity-check codes are used as the outer codes has been indicated, as well as simulation results for the obtained concatenated construction have been represented.