• MATLOG: a Feistel based poly-alphabetic encryption

      Zaland, Zubair (University of Bedfordshire, 2012-05)
      One might question, "What makes MATLOG different from other feistel based ciphers?", the answer is the encryption methods that are both Feistel and non Feistel. First of all the idea of poly-alphabetic encryption is unique to feistel approach. The capabilities like generating round keys from a single key like Advance Encryption Standard (The Rijndeal Algorithm ) but that of variable lengths and the AES itself not being feistel based. The idea of security through obscurity, although not implemented with its true meaning but MATLOG does not use the user key directly for encryption and the computer generated keys do not directly encrypt the data not even letting the user himself know how it is done. The artefact is currently set to 10 rounds of encryption whereas within each round it performs 2 - 6 rounds of substitution and one round of transposition using two methods of transposition joined as one. In this document I am going to discuss the various standards of encryption and various modes of operation they work on. Then discuss the survey I conducted online in which the university students and other people on social networks having the knowledge of the subject took part, Followed by the design of the algorithm Itself which involves splitting of the plaintext first of all; Splitting the block here means creating two separate blocks out of random numbers and the user data, topic is discussed in further detail under the heading of design. after splitting the blocks we apply transposition to one block and substitution to the other and vice versa to the result of the first round this process goes on for ten rounds and finally we have our cipher text when the result of the two encrypted blocks is merged into one. The decryption process is the same only the keys decrypt the opposite way i.e. the last key decrypts first. The process of key management is very crucial, even though the user key is hardly used in encryption the application has to maintain the user key and the automatically generated keys intact until decryption successfully takes place. MATLOG is a highly flexible and dynamic algorithm it can be set to transposition in several ways, the keys are always different and they process the poly-alphabetically generated data, the keys are also of different size and value for every round within substitution function making the substitution half highly flexible too and best of all it can be commercially implemented both on hardware and the application layer. It can also be set to perform cryptography on storage devices and not only communications. The design of MATLOG is based on independent blocks eliminating the block dependencies like that of cipher block chaining (CBC) , this allows the algorithm to support parallel processing making it as efficient as it would be as any stream cipher algorithm. I have eliminated the Initialization vector in this algorithm and replaced it with the One-time pads making it stronger and leaving no patterns of encryption which was one of the flaws on initialization vector, it needed frequent management one of the reasons that caused the WEP 1 protocol to fail was the short size and mismanagement of the initialization vector, instead my idea of spreading blocks is highly random and the transposition and substitution functions reinforce the safety to the my idea of IV's replacement.