Performance enhancement for ansi x3.92 data encryption algorithm standard

Abstract

ANSI X3.92 Data Encryption algorithm is public knowledge, and is widely used where data security and integrity is vital, such as commercial banks, secret service organizations etc. Even though this algorithm is feasible to implement in software, it is impractical to achieve desired performance for most of the applications. Hence, a hardware solution is highly recommended. Prior art DES hardware in CMOS technology served performance needs of low-end and mid-range systems only, due to the technology constraints. However, some of these constraints are removed through the technology breakthroughs and the current CMOS is well suited for high performance applications. While prior art DES designs allowed one round per cycle to minimize the cell count, the current technology allows of multiple rounds per cycle due to the denser CMOS chip technology. In order to maximize the number of rounds for a given cycle time, careful study of algorithm to determine the critical paths from a logic implementation perspective is required. This invention identifies one such path in f-function of the DES algorithm, where the expanded data and the key is XOR'd prior to entering S-function. It is mathematically proven that if the left half of the input data is expanded and XOR'ed with the second key at the same time the right half of the input data which is still going through its XOR, S-function, permute and expansion, then this expanded result can be immediately XOR'ed with the left input data being expanded and XOR'ed in parallel. The resulting output can be used as input to the next S-function, thus eliminating a stage of expansion and XOR logic for all subsequent S-function inputs in rounds 2 through 15.