New tailbiting encoders

In a tailbiting trellis encoder, the starting state of the encoder is set to the state at which the machine will be at the end of the encoded frame. By this means the probability of decoding error at the end of the frame can be kept low without the addition of termination bits and the consequent rate loss. We report on an extensive search for short and moderate-length convolutional encoders for tailbiting trellis representations of block codes at rates 1/4, 1/3, 1/2, and 2/3. The short tailbiting representations found are typically as good as the best known block codes.