大规模网络中基于FFT的呼叫阻塞率的快速算法

Fast Algorithm for Calculating Call-Blocking Probability Based on FFT in Large-Scale Networks

在大规模网络条件下,Kaufman提出的一维递推快速呼叫阻塞率(CBP)算法由于其呼叫阻塞率的计算将导致系统计算溢出,改进的Kaufman方案虽然消除了计算溢出,但是其计算时间随网络规模的变大呈指数增长.有鉴于此,文中提出了一种基于快速傅立叶变换(FFT)方法和计算溢出避免预处理机制的呼叫阻塞率快速计算方法.仿真结果表明,该算法消除了计算溢出且降低了计算复杂度,具有计算简单、无误差、速度快的优点.

In large-scale networks, the 1 D fast recursion CBP （Call-Blocking Probability） algorithm proposed by Kaufman is unavailable due to the data overflow caused by the calculation of CBP. The modified Kaufman method overcomes the trouble of data overflow, but the time consumption exponentially increases with network scale. To get out of this trouble, a fast algorithm to calculate CBP based on the FFT（ Fast Fourier Transform） and the pre-treatment method is proposed. Simulation results show that the fast algorithm not only avoids data overflow but also decreases computation complexity dramatically, and is of the advantages of simplicity, rapid computation and high accuracy.