ML based Intelligent Wavelength Assignments in all Optical Networks

For the past several years, Internet traffic has been increased tremendously and this growth demands fora high-speed communication network such as optical network. The RWA is one of the key tasks for the optical network, due to its information transparency and wavelength reuse characteristics. If multiple required wavelengths are available on the entire route between a source-destination pair, a wavelength selection mechanism is used to select the best wavelength. The wavelength selection may be performed either after a path has been determined, or in parallel during the path selection. It is also important that wavelengths should be selected in such way that the performance of the network should be optimal. In this direction, a number of heuristics have been proposed. Among them, some significant heuristics such as First-Fit, Least Used, Most Used, and Random WA. In the First-fit (FF) scheme, a list of available and used indexed wavelengths is maintained. This scheme always chooses the lowest indexed wavelength from the list of available wavelengths and assigns it to the connection request. When the call is completed, the wavelength is added back to the available wavelength list. To implement this scheme, any global information of the network is not needed. FF performs better in terms of BP and fairness because of its low computational overhead and complexity. Least used (LU) scheme assigns the connection request to the available wavelength which is already used on the minimum number of other fiber links in the network. If several available wavelengths share the same number of usages, FF scheme is used to select the wavelength. In the Most used (MU) scheme, the available wavelength that is already used on the maximum number of other fiber links in the network is assigned to the connection request. Similar as LU, if several available wavelengths share the same maximum usage, FF scheme is used to break the tie. In the Random WA scheme, a list of available wavelengths is maintained. Whenever a connection request arrives in the network, this scheme randomly selects a wavelength from the set of available wavelengths and assigns it to the connection request. Then, the list of available wavelengths is updated by removing the used wavelength from the free list. When the call is completed, the wavelength is again added to the list of free wavelengths. The wavelength assignment plays an important role in further enhancing the network performance; two schemes mainly ARWA and FT-ARWA are developed. The dynamic wavelength assignment is carried out in ARWA algorithm on immediate and intermediate assignments. The blocking probability, throughput and the bandwidth utilization is improved when more flows are combined together for ARWA algorithm.