Detection and blind identification of m-sequence codes using higher order statistics

The blind identification of m-sequences from short noisy intercepts remains intractable using second-order statistics. Previous work has shown that the triple correlation function provides a basis for such identification. By extending the field theory of m-sequences, new techniques are developed which allow the automatic derivation of their generator polynomials. Statistical theory is presented for the triple correlation of m-sequences with discrete errors resulting from hard decoding. Efficient C-implemented algorithms enable extended simulation to assess the new technique's performance for varying noise level and intercept length.