波長分割多工在光纖網路上將頻譜分割成許多不同的頻道(波長)，不同的波長上傳輸不同速度的資料。良好的波長分配演算法讓波長作最有效的利用，才能避免資料在傳輸的過程中，因光波長衝突所造成波長轉換與調整延遲的問題，尤其網路流量動態增高時能有效解決傳輸效能瓶頸、提升整體網路的效能。然而，波長分配問題在之前的
研究被證明是NP-Complete，無法得到最佳解。本論文提出以決策樹演算法ID3 為主的動態波長分配機制應用於波長分割多工技術架構下的環狀接取網路，利用其具有快速分類和預測的學習法來產生波長分配的規則，決策出最適當的波長，使得傳輸過程中光交換器切換與波長轉換發生次數最小，期望減少調整延遲、增加連線機會，因而提升整體傳輸效能。本文利用乙太網路環境配合軟體程式來模擬環狀接取網路，不同的連接埠來模擬不同的波長頻道。實驗結果證明，只要波長在足夠可供使用下，運用決策樹演算法產生波長分配規則於不同的網路流量，以及不同存取節點數與波長數的比例(1:1、2:1 與4:1)下，能有效地減少光交換器切換與波長轉換的次數，因而降低調整延遲的時間、提升整體傳輸的效能。

Wavelength Division Multiplexing (WDM) is a technique can divide spectra into many different channels, and deliver the data in the different wavelengths with different transmission speed. To use the wavelength more effectively, we must apply a good routing and wavelength assignment (RWA) algorithm to avoid the problem of wavelength conflict, blocking, and to reduce the number of conversions and tuning latency. Previous studies showed that the RWA problem is NP-Complete. This paper proposes a heuristic RWA algorithm based on Decision Tree - ID3 algorithm. The simulation environment is emulated by Ethernet and software. The simulation proves that using decision tree algorithm to produce the wavelength assignment rule is able to reduce optical switch switching and wavelength conversion occurrence for different ratios of nodes and wavelengths (1:1, 2:1 and 4:1). Therefore, it minimizes the time of the tuning latency and can promote the system transmission performance.