在一对多传输模式中,广播/多播比单播能够提供更高的传输效率,单组数据重传次数的数学期望是可靠广播理论中的一个基本参数。迄今为止,相关文献基于离散概率分布函数给出了两类该参数的解:精确的级数解和用于分析复杂度的近似解。针对前者计算时间较多、后者精度较差的情况,构造对数函数幂级数部分和的近似表达式,将离散概率分布函数进行连续化处理,得到误差更低、物理意义更加明确的近似解。在此基础上,将级数解和概率连续近似解进行组合,导出高精度的近似解。数值模拟实验表明,其平均误差比已有近似解低2~3个量级。
In one-many transmission model, broadcast/multicast was more effective than unicast at delivery efficiency. The mathematical expectation of individual packet transmission times was a fundamental parameter in the theory of reli-able broadcast. Up to now, there were two categories of solutions for this parameter in relevant literatures:exact series solutions and approximate solutions used for analyzing complexities. The former may consume too much computing time, and the latter’s precision was very low in some regions. Combining one series solution with the solution based on the as-sumption of continuous probability distribution, derive a few simple and high-precision solutions whose physical mean-ings were more obvious. Simulation shows that the average deviation has been lower than the current solution 2~3 order of magnitude.