Dynamic Channel Assignment in a Mobile Communication system Using Neural Networks

在通常的蜂窝系统的动态信道分配算法中，相同的信道被分配在具有恒定间隔的小区中。近年来，提出了一种新的信道分配方案，即当载波干扰比大于期望值时，可以以比通常更小的间隔分配同一信道，该信道分配问题在图论中是一个难处理的（NP完全）问题。我们建议应用Hopfield的神经网络的蜂窝系统中的信道分配问题。据说Hopfield的神经网络可以非常快速地解决优化问题。计算机模拟表明，神经网络对信道的有效利用是有用的。

In usual dynamic channel assignment algorithms for cellular systems, the same channels are assigned in cells with a constant separation. Recently, a new channel assignment scheme was proposed, in which the same channel can be assigned with less separation than usual if the CIR (carrier-interference ratio) is more than the desired value.The channel assignment problem mentioned above is known to be intractable (NP complete) problem in graph theory. We propose applying Hopfield's neural networks to the channel assignment problems in cellular systems. It is said that Hopfield's neural networks can solve optimization problems very quickly. The computer simulations show that the neural networks are useful for the effective utilization of channels.