No abstract

Modern microprocessors are loaded with a lot of performance optimization features. Single Instruction Multiple Data (SIMD) instruction set feature specially designed for improving the performance of multimedia applications is one among them. But most of the encryption algorithms do not use these features to its fullest. This paper discusses various optimization principles to be followed by encryption algorithm designers to exploit the features of underlying processor to the maximum. It also analyses the performance of four eSTREAM finalist stream ciphers – HC-128, Rabbit, Salsa 20/12 and Sosemanuk – under various methods of implementation. Scope of implementing these stream ciphers using SIMD instructions is examined and improvement in performance achieved by this implementation has been measured. Modifications in the algorithm which provide further improvement in performance of these ciphers are also studied

The stream ciphers RCR-64 and RCR-32 designed by Sekar et al. are the most recent additions to the Py-family of stream ciphers, originally designed by Biham et al. The ciphers are among the fastest stream ciphers on software. To the best of our knowledge, the only reported attacks on the ciphers are due to Ding et al., published in the Journal of Universal Computer Science. In this paper, we review these alleged attacks on the RCR ciphers and show that they are based on non-existent keystream biases stemming from flawed probability calculations.

SPECK is a family of lightweight block ciphers developed by Beaulieu et al. of the US National Security Agency (NSA) for the Internet of Things (IoT). It is an ARX-based design with a Feistel-like structure which supports keys of size ranging from 64 bits to 256 bits. SPECK has been standardised by ISO/IEC for radio frequency identification (RFID) devices. It has drawn the attention of many cryptanalysts and several cryptanalysis results have been published. In this paper, carry flag attacks on the full SPECK ciphers are presented. Depending on the key size and block size, the complexities of our attacks, to nearly ensure success, vary from 2 59 time and 2 14 data to 2 227 time and 2 62 data.