From the Village Lane to the Highway
Optical Fiber Networks Reach High Speed Using Photonic Packet Switching
Ever since the first live telephone traffic was sent through fiber optics in April 1977, the evolution of optical networks can be characterized by two major trends: Packets replacing circuits and photonics replacing electronics. Together with these two trends, the ever-increasing bandwidth demand drives optical fiber networks to consider, develop, and add new solutions such as Photonic Packet Switching (PPS) into their family of next-generation dynamic optical network components.
Historically, economic efficiencies and technological performance improvement in backbone networks were often achieved by substituting photonics with electronics. This trend started by replacing regenerators, installed every 40-80km, with in-line optical amplifiers. The benefits materialized in decreased power consumption, reduced number of components, and facilitated and accelerated network upgradeability. The introduction of optical amplifiers solved many problems on interconnects (e.g. physical fiber). However, one challenge had not been resolved yet: a performance bottleneck at the point where switching and routing takes place; the network nodes.
The full performance potential of optical communication networks has yet to be unlocked. With respect to its easily-achievable high bandwidth capabilities of 40Gbps/channel (and beyond) the restriction arises from converting the optical signals to the electronic domain. Limitations of electronic processing performance of the routers are another factor. Manufacturers are well aware of the challenges to provide packet forwarding at wire speed with minimum packet size. From this point of view, electronics impose a bottleneck on the data rate which, in principle, would be achievable via fiber optics.
A large number of telecommunication operators are (still) operating IP/MPLS core/backbone networks with point-to-point DWDM links between the core nodes. At every node the optical signal may need to be converted to the electronic domain. These optical-electrical-optical (OEO) conversions and associated electronic processing yield in an additional cost at the service provider’s POP. It also requires more space to accommodate a larger number of racks and shelves and additional power and cooling would be necessary to operate the active electronic components.
Limitations in the world of electronics
When talking about speed and the evolution of optical fiber networks we predict a traffic jam or electronic bottleneck in the networks that evolve with the OEO approach. To visualize this better, one can think of traffic intersections (Figure 1). Performing OEO conversion at a node is similar to a 4-way stop intersection. In the telecommunication network, the data traffic is transported nearly at the speed of light in the optical fibers, gets converted at the node into the electronic domain (traffic light is on red) and is then converted back to the optical domain to eventually complete its journey through the node and to its destination. Compared to its initial propagation speed, the time which is necessary for converting and processing the data is like a stop-and-go principle.
Next page
