高级检索

    基于受限行为约束策略的桌面计算系统交互性能测量方法

    A Measurement Method of Interactive Performance for Desktop Computing System Using Restricted Behavior Policy

    • 摘要: 测量桌面计算系统的交互性能是分析和改进系统性能的重要基础.方便地获得准确测量结果的难点是如何确定输入事件和其直接触发的输出事件之间的对应关系.为此,提出一种采用受限行为约束策略的交互性能测量方法,首先分析了确定输入、输出事件的对应关系需要满足的约束条件,通过会话录制辅助机制合理转化难以满足约束条件的事件,通过用户行为检验机制进一步确认所有事件已经满足约束条件,从而保证可以获得更准确的测量结果.基于该方法,实现了一个支持Windows和Linux等主流桌面计算系统的测量工具VNC-IPA.实验以“慢动作”基准测量方法获得的测量结果为真值,结果表明,相比其他交互性能测量工具VNCplay,使用VNC-IPA获得的测量结果的相对误差平均降低了35.4%.

       

      Abstract: Measuring interactive performance is the important basis to analyze and improve desktop computing system. The difficulty of measuring interactive performance is how to get exact measurement result conveniently. Exact measurement requires establishing the corresponding relationship between output events and input events. The necessary restrictions for finding the output events of input events are analyzed in this paper. Then we propose an interactive performance measurement method using restricted behavior policy. The method consists of session recorder assistant and user behavior validator. Session recorder assistant is used to record user events and transform unqualified events simultaneously, and user behavior validator is used to check whether all recorded events are valid. Through these two steps, all events are ascertained to get exact measurement results. We implement an interactive performance measurement tool named VNC-IPA using the method. We conduct experiment to compare the accuracy of VNC-IPA and VNCplay. The experiments treat slow-motions result as true value, and it shows that the average of relative errors measuring with VNC-IPA is decreased by 35.4% with respect to VNCplay. Experiments results show that our method can get exact measurement results for interactive events conveniently. Besides, our tool is independent of operating system and can be used in Windows and Linux.

       

    /

    返回文章
    返回