Convolutional Codes Using Nonlinear Generators for Rate One-half and Memory Order Four

Performance of a convolutional code depends on the decoding algorithm and that code's distance properties. Performance bounds suggest constructing codes with the largest possible free distance. Published tables provide the feed forward encoding equations that are linear generator polynomials over a binary Galois field. This paper presents a new method for testing free distance of generators based on length k bounded synchronization codes. This paper also presents results for rate one-half, memory order four, convolutional codes with at least one feed forward encoding equation that is nonlinear over a binary Galois field.