提出了一种基于实测的呼叫接入控制算法.该算法在为每个会话提供合适的服务质量等级的同时,确保切换呼叫的掉线率低于预定义的目标值,同时提高系统统计复用增益(系统允许接入呼叫连接数).该算法根据切换呼叫掉线数目和掉线概率来预测下一时刻可能的切换掉线数,据此动态调整预留门限值.同时采用Hopfield神经网络在所有已接入连接和允许接入的新呼叫间自适应和公平地分配带宽资源.仿真结果表明,该算法可保证掉线概率在目标值范围内,同时可降低新呼叫阻塞率。从而提高了系统的统计复用增益.
Presents a measurement-based call admission control (CAC) algorithm was presented. The scheme measures the state parameters of cellular system, i.e. , the number of handoff-dropping calls and the handoff-dropping probability at each interval, and forecasts the possible number of handoff dropping calls at next interval. Based on the forecasting value and the current handoff-dropping probability, the reservation threshold is updated. Also, a Hopfield neural network (HNN) is used to allocate bandwidth resources adaptively and fairly among the calls. The simulation results show that our CAC algorithm can guarantee the predefined handoffdropping probability while increasing the statistical multiplexing gain in providing acceptable service grades for every connection.