Modified Generalized Feistel Network Block Cipher for the Internet of Things

Abstract

With the advent of the Internet-of-Things (IoT) technologies, millions of low-resource devices are constantly used at the network’s edge. As a result, the large amount of private and sensitive data generated by these devices must be securely transported, stored, and processed, posing a challenge because these resource-constrained IoT devices cannot meet the criteria of conventional encryption ciphers. Due to this limitation on IoT-enabled devices, lightweight cryptography has emerged as a new area of study. Lightweight block ciphers, a subfield of lightweight cryptography, include the substitution–permutation network (SPN) and Feistel-based networks. Feistel networks are further divided into two types: classical Feistel networks and generalized Feistel networks (GFN). While classical Feistel ciphers divide a message into two sub-blocks, GFN divides a message into k sub-blocks for some k > 2 called the partition number. One popular form of GFN is the so-called Type-II. Unfortunately, this type of Feistel structure needs a large number of rounds to obtain a full diffusion property. A full diffusion means all output sub-blocks are affected by all input sub-blocks. Therefore, this paper proposed a new lightweight block cipher by modifying the GFN structure, focusing on providing optimal security to the cipher with a small number of rounds. The algorithm was subjected to a series of statistical and cryptographic randomization analyses in order to investigate the avalanche effect on the ciphertext and the algorithm’s random properties, such as confusion, diffusion, and independence. The avalanche criterion and output randomness results show that this algorithm meets the fundamental security requirement for a lightweight block cipher.

Keywords

• lightweight block cipher
• modified Feistel
• randomness test
• avalanche test