An enhanced lightweight crypto-hash function technique using new mersenne number transform for internet of things security

Abstract

Internet of Things is a concept that describes the idea of connecting everyday physical objects to the internet. So no longer objects are just related to their user, but now it is connected to surrounding objects and database. The main challenge in designing IoT applications is in the field of security. IoT contains resource-constrained devices such as sensors, actuators, and Radio Frequency Identification (RFID) in the edge layer. In order to implement the security mechanism in these types of devices, lightweight cryptographic techniques are the obvious solution. There are several lightweight hashing techniques available today. Examples are PHOTON, QUARK, SPONGENT, GLUON, etc. These all are fixed length block sized and key sized lightweight hashing techniques. The existing lightweight hash family uses Maximum Distance Separable (MDS), Mixed column transformation or by using some registers for the desired diffusion. All transformation methods available only support fixed block size and key and require high hardware requirements. This thesis proposes a Novel New Mersenne Number Transform (NMNT) based Lightweight hash function for IoT applications. This proposed Lightweight hash uses New Mersenne Number Transform, which provides good diffusion property and employs a fast algorithm to compute the transform. Further, the hash function’s New Mersenne Number Transform supports the powerful property of variable transform length (powers of two). These properties make New Mersenne Number Transform suitable for the design of new Lightweight hashing technique. The proposed lightweight hash function is named lightweight New Mersenne number transform hash function (LNMNT) and it is evaluated in terms randomness, confusion and diffusion, distribution of hash function and different attacks. The randomness analysis testing is done using standardized NIST test suit. The hash function was evaluated by means of COOJA simulator and numerical models and was benchmarked against lightweight hash function PHOTON and the proposed system shows about 65 percentage of improvement in time of execution and 25 percentage improvement in randomness property. And also did some comparisons on other todays available lightweight hash function QUARK, SPONGENT, GLUON in a testbed implemented using Contiki OS platform running on Zolertia Z1 motes in terms of time of execution, cycles per byte and memory usage. The analysis result showed that our new lightweight hash function has good random property and is highly sensitive to a slight change in the input message and it consumed very low resources where the time of execution is only 1.3 seconds while the power consumption is 6.7 µW. The proposed LNMNT uses 54042 cycles per byte for the hash length of 128 bits making it compete very well in comparison with other standardized industry-adopted lightweight hash functions, in terms of cycles per byte and execution speed. Furthermore, other LWT hash functions are not adaptable to different hash digest lengths. However, with LNMNT, transform length can be changed and create variable-length hash digests without increasing the number of rounds.