基于自回归技术的自适应服务质量测量方法

SLA管理已经成为服务商和用户关心的重要内容。在进行服务质量管理时，服务质量的测量是其他管理功能的基础。本文从采样的角度分析了服务质量测量和传统网络测量的异同点，给出了服务质量采样频率的理论下限。在讨论已有方法不足的基础上，提出了基于预测的自适应采样算法，该算法在采样区间内随机确定采样的具体时间点，并结合自回归预测方法对即将采样的结果进行预测，在此基础上以一定的概率接受预测结果，放弃对准确预测时间点的采样，从而减少了采样次数，降低了对服务的影响。最后的实验结果说明了算法的有效性。

An Adaptive Method for Service Level Monitoring Using Auto-Regressive Technique

SLA (Service Level Agreement) and its management have become more and more important for both service providers and customers, within which measurement of service level is fundamental for other managements. This paper discusses difference between service measurement and traditional network measurement from sampling perspective, and deduces a theoretical lower-limit on sampling frequency. To reduce the sampling and keep high accuracy of service level data, the paper presents an adaptive algorithm based on estimation method, which determines time to sample using random function during a small even interval, then constructs an auto-regressive model according to historical data to estimate the current data of service at the sampled time, accepts the result by an adaptive probability method, and gives in sampling at this interval. The algorithm decreases sampling's influence on measured service through reducing sampling times. Experiments show its good efficiency.