Digital subcarrier optical networks (DSONs)
Huang, Wanjun, 1978-
Tacca, Marco, 1973-
MetadataShow full item record
Energy efficient networks are increasingly becoming a desirable feature in today’s market. Both the number of users and the average amount of data traffic generated by each user continue to grow, requiring more powerful network routers and switches, which in turn dissipate large amount of electric power to operate. This problem is in part circumvented by deploying all-optical wavelength division multiplexing (WDM) solutions in the network, which eliminate any electronic processing of the in-transit data at the intermediate network nodes by dedicating a path of light (a wavelength) across the network to directly interconnect two edge nodes. However, the all-optical approach is only suitable when the average quantity of traffic to be exchanged by two edge nodes is sufficient large to warrant one entire (or many) dedicated wavelength(s). Considering that optical transmission rates are moving up from today’s 10 Gbps to 40, 100 and even 160 Gbps per wavelength, the fraction of edge nodes that exchange such amount of traffic is not (surprisingly) limited, as many of the edge node pairs would require only sub-wavelength connectivity. Sub-wavelength connectivity is today offered by either Optical Transport Network (OTN) or Multi Protocol Label Switching with Transport Profile (MPLS-TP). These solutions run on top of the WDM layer. Unfortunately, the amount of required electronic processing in these solutions is such that an order of magnitude higher power consumption results compared to all-optical networks. Part of this extra power consumption is due to the electronic buffering of the in-transit data at the intermediate nodes. This paper points to an alternative solution to achieving sub-wavelength bandwidth assignment to edge node pairs, which eliminates the need for data buffering at the intermediate nodes. Sub-wavelength channels or circuits are creating by using spectrally efficient orthogonal frequencies in each wavelength, with each frequency arrying a fraction of the wavelength bandwidth. By assigning one or more such frequencies to one edge node pair, an end-to-end sub-wavelength circuit is created. At the intermediate nodes, incoming frequencies are switched to outgoing frequencies via specially designed frequency selective switches or cross-connects. The power consumption required to switch frequencies in and out is estimated to be only a fraction of the power dissipated by current transport solutions, thus mitigating the energy consumption struggle when assigning subwavelength capacities to edge nodes.
"This research was supported in part by NSF Grant No. CNS-1111329."