Browse Prior Art Database

Self Learning Performance Analysis Method for Optical Transmission Networks Disclosure Number: IPCOM000200493D
Publication Date: 2010-Oct-15
Document File: 9 page(s) / 3M

Publishing Venue

The Prior Art Database

Related People

Prior Art Publishing GmbH: CONTACT

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 16% of the total text.

Page 01 of 9

Self Learning Performance Analysis Method for Optical

Transmission Networks


In future optical networks, dynamic routing will play an important role, especially in strongly meshed net- works. The traffic demands change from time to time and the network operator has to adapt the network according to this change. The customer can decide which path a wavelength channel takes in the network by switching the channel in the network in the wanted direction. Figure 1 shows an example, in which a wave- length channel λn is routed from network node 1 over network node 2 to node 3. After rerouting, the channel starts at node 1 and is transmitted over node 2 and 5 to the end node 6. This dynamic switchable network ar- chitecture can be prepared by developing optical cross- connects, which are able to route a channel in more than 8 directions.

The performance of a wavelength channel λn depends on many things, such as the given network architec- ture, the deployed components, the fibre input power, and the type of the channel's neighbors. If the network operator wants to switch a signal channel, a planning- tool has to decide whether this signal can be routed or not, or if a regenerator has to be placed. Furthermore, the change of the performance of the neighbor chan- nels by the new routed channel has to be taken into ac- count. If the new routed signal infLuences its neighbor channel too strong, the new route cannot be installed. Then a new channel assignment can be proposed.

Currently, the network configuration is calculated by using worst case rules. Worst case rules means for example statistically distributed performance data of some components, so that a defined confidence level has to be considered. Furthermore, planning is done manually, which is very time consuming, as every re- routed or new routed traffic demand has to be re-cal- culated and optimized. Hence, planning a switching process becomes difficult, ineffective and expensive, as a higher number of regenerators have to be placed. The operator has to make a try and error approach (in simulation and in the field), which is very time and cost consuming.

The basic problem is that the real change of the per- formance of the network cannot be calculated by general rules. More- over, high security margin has to be added, which reduces maximum reach. The exact rules depend on the whole network architecture and traffic demand. On the other hand, the planning tool has to be coupled with the network element manager software, which controls the switching process so that real- time switching can be assured.

Figure 2 shows the process of how a network is currently planned. After the network is planned and installed by the network planning tool, the param- eters, which have been optimized by the planning tool, are given to the NEMS (Network Element Manager Software). Each time when a new channel has to be installed or an existing route has to be r...