中文摘要：

针对现有的算法只能生成单分形或只能生成重分形序列的问题，文中提出了一种新的分形序列生成方法——调整方差随机二分法，通过调整该方法中的参数值σ，可生成单分形或重分形序列，而生成单分形序列的自相似度和重分形序列的勒让德谱取决于σ．仿真实验结果证明该方法可以快速地生成较精确的自相似序列．采用文中提出的方法能生成具有不同程度局部突发的分形序列，可方便地模拟不同的网络流量．

英文摘要：

As the existing generation algorithms of fractal sequence can only generate monofractal or muhifractal sequences, this paper proposes a new generation method of fraetal sequence named variance-adjusting random dichotomy （VARD）. By using the proposed method, a monofractal or multifraetal sequence can be generated by adjusting the input argument σ, and the self-similarity of a monofractal sequence and the Legendre spectrum of a multifractal sequence depend on the input argument. Simulated results indicate that the proposed method helps to rapidly generate self-similar sequences with high accuracy and fractal sequences with various local bursts, and is convenient to the simulation of different network traffics.