Ad hoc网络的移动性、高度动态拓扑使得实时多媒体传输受到严重影响，为了解决多媒体通信中的传输不稳定和丢包问题，应用Markov过程提出了基于概率度的分簇算法（PD）。该算法首先在当前时刻预测下一时刻任意两点间链路的连通性，然后选取链路的连通概率较高的节点作为簇头。仿真结果表明，PD算法成簇数目少于最高连通度算法（HD），而且簇内节点无重叠，从而有效降低了分簇结构的维护开销。由于簇成员数量并不明显多于HD，因此说明PD算法效率较高，而且没因分簇数目减少给簇首增加负担。最后在此分簇基础上，结合UDP协议，给出了一种基于概率度分簇的ad hoc网络实时多媒体通信方案，不仅使得传输稳定，而且解决了数据包失序和丢包问题。

The transmission of real-time multimedia is seriously affected by the mobility and highly dynamic topological of ad hoc networks,a clustering algorithm based on probability degree(PD) is proposed by using Markov process for solving the problem of instability and data losing in multimedia communication.Firstly,the algorithm forecasts the next time connectivity of the link between two random nodes based on that of current time,and then the node of the highest link connectivity probability is selected as the Cluster Head.The simulation results show that the number of clusters of PD is less than that of highest connectivity degree algorithm(HD),without overlapping nodes so that maintenance cost is effectively decreased.The number of cluster members shows minor differences,so it illustrates that PD algorithm has higher efficiency and no increase burden of Cluster Head,although the clustering number decreases.Finally,on the basis of clustering and considering the UDP protocols,a scheme of real-time multimedia communication in Ad hoc networks based on Probability Degree clustering is given.The multimedia communication is stable in ad hoc networks with the scheme and the problem of losing and out-of-sequence of data packet is solved.