Comprehensive end-to-end reliability assessment of optical network transports

A hybrid simulation and analytical approach has been developed to evaluate the end-to-end reliability of the network and a new metric, expected loss of lambdas, has been computed to provide more reliability details for the network. Applying the approach to the example network, the study shows that: 1. The 1+1 and 1+0 cases provide the two extreme cases for the end-to-end network reliability. Any other protections, IN and shared mesh, fall somewhere in between these two cases. 2. For no protection, the main failure contributors are the link elements, the downtimes for the demand pairs are very high, and have a wide variation due to various lengths and number of edge elements used. 3. For 1+1 protection, since the end-to-end links are fully protected, the main failure contributions come mainly from the node equipments. The downtimes are at least two orders of magnitudes less than those from the 1+1 case and have a relatively narrow spread. 4. For 1:N and shared mesh protections, two major sources contribute to the unavailability of the demand pairs, these are: (i) node element failures; (ii) the unavailability of the alternate optical path when required, due to prior failures of other working paths that it is supposed to protect. As a result, network with short or medium length of transports, the average unavailability tends to be closer to the results for 1+1, while the spread of the profile resembles that of the case with no protection. 5. The unavailability and the expected loss of load (lambdas) of the network, hence potential loss in revenue, decrease from a 1+0, to 1:N to shared mesh to 1+1; whereas the cost of implementing each protection scheme increase in the same order. An optimal balance between reliability cost and worth can be determined. 6. For long haul transport network, such as System B, redundancy and protection, such as 1+1 protection scheme, must be provided in order to provide acceptable end-to-end reliability. 7. At the network level, there are various design alternatives that can be implemented by network designers to achieve the desired end-to-end network reliability. Hence increasing component reliability beyond what it is today may not be necessary.