Other Block Ciphers
In the late 1960s, led by Horst Feistel and later by Walt Tuchman, IBM initiated a research program in computer cryptography called Lucifer. Lucifer is also the name of a block algorithm that came out of that program in the early 1970s [1482, 1484]. In fact, there are at least two different algorithms with that name [552, 1492]. And  leaves some gaps in the specification of the algorithm. All this has led to more than a little confusion.
Lucifer is a substitution-permutation network, with building blocks similar to DES. In DES, the output of the function f is XORed with the input of the previous round to form the input of the next round. Lucifer's S-boxes have 4-bit inputs and 4-bit outputs; the input of the S-boxes is the bit-permuted output of the S-boxes of the previous round; the input of the S-boxes of the first round is the plaintext. A key bit is used to choose the actual S-box from two possible S-boxes. (Lucifer represents this as a single T-box with 9 bits in and 8 bits out.) Unlike DES, there is no swapping between rounds and no block halves are used. Lucifer has 16 rounds, 128-bit blocks, and a key schedule simpler than DES.
Using differential cryptanalysis against the first incarnation of Lucifer, Biham and Shamir [170, 172] showed that Lucifer, with 32-bit blocks and 8 rounds, can be broken with 40 chosen plaintexts and 229 steps; the same attack can break Lucifer with 128-bit blocks and 8 rounds with 60 chosen plaintexts ...