一個有線電視網路的用戶數通常在幾萬人到幾十萬人不等，對於有線電視業者而言，如何以最低成本提供有效的電視網路服務，關乎營運效率與服務品質。目前在小型網路可以透過最佳化演算法得到。然而，對於大型網路的規劃，由於傳統演算法計算的時間隨著網路大小成冪次增加，因此，要想找到大型網路規劃最佳解，在有限時間內是一個極端嚴苛的挑戰。在本文中，我們提出了一個多層次網路規劃演算法，透過將大型網路分解成數個小型網路，並針對各個小網路分別求最佳解，從而解決大型網路規劃的問題。根據計算實驗的結果，我們的多層次網路規劃演算法雖然在最佳網路成本上略高於單層規劃的結果，然而，隨著網路規模增加，計算時間上卻遠遠低於單層規劃演算法。對於大型有線電視網路而言，這提供了在有限計算時間下，一個可行的網路最佳化演算法。

In general, there are tens to hundreds thousands of subscribers in a CATV network operated in real world. For CATV network operators, the problem size for network planning is too large to solve. In this paper, we propose the multilayer algorithm for large CATV network planning problems. Since the problem size that single layer solution procedure can process is limited, we have to break down a large network into several smaller networks. In the computational experiments, we have compared the result of single and multi layers solution procedure. The solution quality is evaluated by two criteria, network costs and computing time. As shown in experiment results, the costs of network constructed by multilayer solution procedure are greater than those by single layer solution procedure. However, the computing time can be dramatically decreased. For those large networks that could not be solved by single layer solution procedure in limited time, the multilayer solution procedure still provides another feasible alternative.