Sammendrag
Optical transmission has led to a dramatic increase of the transmission capacity in telecommunication networks. Optical fibre is the predominant transmission means in telecommunication networks today, however, optical technology has so far been used to provide large end-to-end capacity circuits that may ultimately be provisioned on demand. Within the European research project STOLAS that has just been completed, we have studied and further developed a novel concept for the realisation of optical packet/burst switched (OPS/OBS) networks. OBS is a network where a series of IP packets is transported as one optical unit by a packet-switched optical transport layer. The network provides QoS differentiated end-to-end transport services based on statistical quality guarantees. The main motivation for introducing packet switching in the optical layer is that it combines the flexibility and granularity of packet based switching with the high-throughput that characterise optical technologies. The novelty of the specific technical concept addressed here, lies with the implementation of the so-called (optical) orthogonal labelling of the optical packets so that these resemble in praxis a conventional packet, the way we know it from IP. This allows for the reuse of well-established network concepts in this new paradigm.
We have carried out a comprehensive study of such networks that has led to architectures and solutions with considerable improvements of the overall performance, e.g. reduction of contention at the optical switch and of the end-to-end delay through the optical network. We have studied design trade-offs and provided improved designs both for the node and for the network as a whole. As an example, our analysis has shown that longer bursts result in lower loss rates. In addition we have proposed QoS schemes that are appropriate for this type of network and studied the resulting performance. Based on existing standards for conventional packet based networks and services, we have derived a set of requirements that should be posed upon an OPS/ OBS networks, given that the latter shall lead to a considerable increase of the performance that is attained by the IP network. We have carried out a thorough study, using both a theoretical analysis and network simulations, to quantify the performance gains that can be attained by OPS/OBS networks. Finally, we have carried out a direct techno-economic comparison with competing commercially available technologies and based on the results concluded on the viability and applicability of the technology as well as estimated the timeframe when this may become a commercially viable alternative.
According to our findings, IP as we know it is a great solution for low traffic density networks, however, as capacity increases it will simply not do without some help from optical technologies. In the short-term optical circuit switching will be implemented to increase switching efficiency, as this is already manifested in the network today. In the longer term, optical packet based switching will be implemented in some form in order to increase throughput, optimise the use of the available resources and at the same time allow efficient QoS differentiation and traffic engineering processes. In addition, we can anticipate the possibility of combined optical circuit and optical packet switched networks. We expect that a paradigm shift to OPS/OBS will find place as the technology matures because the actual technology cross-over point where OPS/OBS routers outperform classical IP routers lies at a node throughput of only a couple optical channels – i.e. corresponding to an order of magnitude of 10Gbit/s.
We will present an overall evaluation of the viability and potential applicability of optical packet/burst switching.
Vis fullstendig beskrivelse
