相对时钟倾斜的相干检测算法

时钟倾斜引起单向传输延迟测量结果的漂移,并直接影响多媒体交互应用的服务质量。精确地测量时钟倾斜有利于更好补偿其影响。提出了一种通过多个节点同时互发报文以提高时钟倾斜测量精度的算法。该算法首先估算各节点到某给定节点的相对时钟倾斜,并由此计算任意两点间的相对时钟倾斜。再通过对估算值和测量值的对比,调整估算值。即,将时钟倾斜的测量问题转化为估算值与测量值间距离函数的最小化问题。证明了在所选目标函数下,求解结果是各节点时钟倾斜的无偏估计,并且其误差小于点对点直接测量的误差。在Internet上的实际应用也表明,该算法的结果具有较好的精度和稳定性,并且能够满足实时处理的需要。

Estimating clock skew by measuring multiple nodes simultaneously

Clock skew shifts one-way delay measure result.It leads to poor QoS of multimedia collaboration.An accurate skew estimate will help to compensate the effect of skew.In this paper,a less error skew estimate algorithm is introduced which is based on simultaneous measure to multiple nodes.Firstly,a random estimate,which could be used to calculate relative skew between each node,is given to each node.Finally,the estimations are optimized by comparing calculated relative skew to measured one.In one word,measuring of skew is transformed to optimizing problem on distance function of calculated and measured skew.It was proved that the result of optimum problem is unbiased;error is less than error of point to point styled measure.The applications on real Internet show that this algorithm is accurate and stable,furthermore it is fast enough for real-time process.