Design And Analysis On Several Primitive Components In Stream Ciphers

Having advantages of simple structure and hardware, fast encryption speed and limited error propagation, stream cipher is widely used as an encryption in high-speed data transmission. Hardware-based stream ciphers are suitable for constrained environments because of its much lower power consumption than the other ciphers. This dissertation focuses on the primitive components of stream ciphers: linear feedback shift registers, nonlinear feedback shift registers and Boolean functions. The contributions of this dissertation are listed as follows.(1) By analyzing the relations between the delayed sequence and the initial states, the paper lower bounds the numbers of flip-flops in the design of LFSR-based stream ciphers resisting the power analysis attack. Besides, a novel lightweight design to resist power analysis attack is proposed. With this method, the number of flip-flops required is decreased to five and the power consumption is significantly reduced.(2) For stream ciphers combined by an NLFSR and a Boolean function, a new Walsh spectrum analysis method of nonlinear components is presented. This is achieved by analyzing the Walsh spectrum of the output Boolean function and NLFSR function. Besides the security properties of balancedness, nonlinearity and correlation immunity of nonlinear components, this paper also focuses on the key values of Walsh spectrum, which may influence these security indicators. With this method, the weakness of the nonlinear component can be found much easier. Finally, the work points out the security weakness of Grain v1.(3) The paper presents a construction of generalized bent Boolean functions defined on 2nZ with values in Z4. The main usage is to derive generalized bent functions on odd number of variables. In addition, by using Boolean permutations, we provide a specific method to construct generalized bent functions on odd number of variables.(4) Based on the analysis of the value and the run characteristics of the Bent function, we present an algorithm for searching the true value table of the Bent function. Comparing with the existing algorithms, the proposed algorithm has a lower storage complexity, and is easy to compute in parallel.