Reduced state computation of distance spectrum for certain optimal convolutional codes

For the class of rate 1/2 N-state linear convolutional codes with complementary structure, which includes all of the presently known optimal codes, the authors show that no more than 3N/4 states are required to compute d/sub free/ and the distance spectrum. Similar state reductions can also be achieved (although to a smaller extent) for rate 1/n codes, n even. A recursive algorithm is presented for distance spectrum calculation using the reduced state transition matrix.