Exploitation of Dynamic routing for Optical Packet-Switched Network Reliability

Optical Packet-Switched Networks can be considered a flexible solution for high-speed core network implementation thanks to their ability of exploiting resources based on statistical multiplexing. One of the main requirements for an OPS Network is its reliability, that is the capability to recover from network device or link failure. In this work dynamic management of packet routing is studied and applied with the aim of keeping packet loss as low as possible in the presence of single link failures. Dynamic management consists in using available network resources by exploiting Wavelength Division Multiplexing (WDM) with full wavelength conversion in relation to possible relevant congestion arising in the Fibre Delay Line (FDL) buffer, which is used to solve contentions between optical packets. In the presence of a single link failure a key parameter is the delay time to detect the failure itself and call the recovery procedure that must be chosen to optimize resource utilization. Dynamic recovery is done by using shared alternative paths classified by their cost expressed in number of hops to reach the destination. Two different techniques are considered, namely Link Protection (LP) and Dynamic Routing (DR), and compared. A simple analytical model for packet loss probability on the failed link as a function of recovery delay is presented and evaluated by comparison to simulation results. Moreover packet loss, throughput and load fluctuation consequent to the fault are obtained by simulation. An example of network design is finally described to cope with loss probability constraints in the presence of failure.