Researchers at Alcatel-Lucent's (Euronext Paris and NYSE: ALU) Bell Labs say they have successfully demonstrated a 6x6 multiple-input/multiple-output (MIMO) digital signal processor (DSP) that can retrieve signal information from noisy space-division multiplexing (SDM) transmissions in real time. The achievement is a major step towards making SDM practical for field use.
SDM has attracted significant research interest as a way of boosting fiber capacity to the petabit level while circumventing the Shannon Limit, the theoretical ceiling of fiber capacity using current transmission techniques and modulation formats (see, for example, "MODE-GAP project makes progress in mode-division multiplexing"). As the term implies, SDM transmissions comprise multiple carriers in parallel that are physically separated. The separations can be small as different modes within a fiber core or as significant as the use of multiple fiber cores.
The more carriers or modes used, the greater the capacity. However, the complexity of receiving and extracting information from such transmissions also increases. For this reason, previous experiments with SDM have used off-line processing, where the DSP function resides in a compute engine outside of the transmission line, reports Peter Winzer, director, Optical Transmission Systems Research at Bell Labs.
The ability to house the DSP function in silicon as coherent transmission systems do now (using a 2x2 MIMO format) therefore represents a significant step towards making SDM a more realistic option for future fiber-optic networks, Winzer points out. The demonstration leveraged a Xilinx XC7V200T FPGA to house the DSP function, paired with twelve 5-GS/s 10-bit silicon germanium analog-to-digital converters (ADCs).
The incoming 30-Gbps signal comprised twelve 2.5-Gbps transmitter outputs run through three polarization-diversity lithium niobate in-phase-quadrature modulators and onto a 1550-nm optical carrier generated via a single external cavity laser (ECL) with 100-kHz linewidth. After amplification and launch power adjustments, the signals entered the three inputs of a custom photonic lantern that acted as an SDM multiplexer.
In between the transmitters and receivers ran 60 km of three-core coupled-core fiber. Winzer said the fiber was chosen as being most likely to create a mess sufficient to challenge the in-line DSP.
In a paper delivered at the IEEE Photonics Conference in October, Winzer and his colleagues reported that they encountered several "glitches" in the outputs of the ADCs, which they attributed to clock drift in the communication between the ADCs and the FPGAs. Nevertheless, they reported bit-error ratios ranging between 8x10-4 and 7x10-3 for the six modes and "exemplary constellations" for all six recovered signals.
Winzer says there's still work to do on the DSP technology before it's ready for commercialization; he theorizes that two or three generations of ASIC work might be necessary to reach production readiness.
Meanwhile, Winzer sees SDM as a viable option for future optical transmission needs. He acknowledges that carriers are reluctant to replace fielded single-mode fiber with the exotic fiber types normally associated with SDM, such a multi-core and few-mode fibers. SDM could be applied to existing infrastructure, using multiple single-mode fibers to replicate a multi-core fiber structure. However, Winzer points out that routes eventually will reach fiber exhaustion. At that point, if carriers have to deploy new fiber anyway, an SDM-friendly fiber would prove an economical option, since they're likely to require fewer splices than convention fiber runs, Winzer asserts.
For more information on high-speed transmission systems and suppliers, visit the Lightwave Buyer's Guide.
Sponsored Recommendations
Sponsored Recommendations