Research On Statistics-based Credible Algorithm Evaluation Framework

The information security is the important issues which relates with national security and sovereignty, social stability, national culture inherits and develops. Today, with the extensive attention in the information security, the rational application of a variety of credible algorithms which safeguards the information security has become increasingly important. Therefore, how to analyze and evaluate the credible algorithm's security level and performance has become increasingly urgent.On the bases of the algorithm evaluation, the statistics as well as the computer simulation technology, this paper try to put forward a statistics-based credible algorithm evaluation framework(SBCAEF). The framework mainly composed of three parts (algorithm modeling library, virtual evaluate environment, environment simulating library), the algorithm modeling library mainly contains three kinds of credible algorithms: the data check-up, the message digest algorithm and the encryption algorithm, which are all called by dynamic link library . Virtual evaluate environment is using computer simulation technology to build a virtual evaluate environmental platform, which including the corresponding evaluate programs that designed for different algorithms as well as the improvement of model. While environment simulation library includes the input data environment simulation, natural/man-made attacks environment simulation, and the principle of the input data environmental simulation is based on time-triggered random number generate mechanism to simulate the input data of evaluation in reality, and natural/man-made attacks environment simulation is programmed to simulate natural or man-made attacks environment. Finally, the SBCAME model will be respectively used to evaluate message digest algorithm, the data check-up, and encryption algorithm. The large evaluate amounts of data will use statistical methods to analyze, that is make each evaluate data seems as a series of evenly distributed sample points, using the "ergodicity" of stationary stochastic process, calculating the expectation and variance of the sample to evaluate the security level and performance of algorithm.The statistics-based credible algorithm evaluation model which this paper established may utilize in the evaluation of the future more security algorithms. The result achieved from this paper unified the characteristic of each algorithm, and had pointed out their respective suitable application situation. It has the important reference value regarding the reasonable use existing credible algorithm and design and study the new safer algorithm.