Frank, thanks again.
harmonicinc.com
Return Path
Traditionally, multiple return paths are simply combined such that the upstream signal from thousands of subscribers is fed to a single CMTS or HDT. This results in very high noise levels, and limits the number of subscribers who can access the service. As service penetration increases, operators must be able to segment the upstream to serve much smaller numbers of subscribers. Several methods exist for segmenting the return path, thus providing dedicated upstream bandwidth to customers. Figure 4 illustrates the pure DWDM option, which is basically the mirror image of the DWDM downstream. ITU return path transmitters in the mini-node transmit back to the OSN, where the signal is DWDM muxed with the signals from the other mini-nodes served by the OSN. Since the signals are 5-40 MHz analog, it is necessary to amplify with an EDFA before transmitting back to the head end in order to maintain acceptable performance. At the head end, the signals are demultiplexed and fed to individual return path receivers. The DWDM upstream option provides excellent segmentation. But it is not scalable, since the required ITU lasers in every mini-node, and the cabinet EDFA's, combine to make the initial system deployment relatively expensive. The system does provide excellent return path bandwidth of up to approximately 100 Mbps, assuming 16-QAM modulation. This corresponds to 1 Mbps peak rate per subscriber, which may be more than necessary in the early stages of deployment.
A more scalable and less expensive return path option is to combine digital transmission with DWDM. As shown in Figure 5, the 5-40 MHz upstream signal is transmitted by a 1310 nm laser from the mini-node to the OSN. The laser could be either a relatively low-inexpensive uncooled distributed feedback (DFB) laser or a very low-cost Fabry-Perot (FP) laser. The choice between the two depends on how much combining the operator plans to do. FP lasers are more noisy, particularly when no signal is driving them. DFB lasers therefore may be necessary when combining many return path segments, and when high priority services like telephony are offered.
At the OSN, the signal is received and combined with three other upstream signals. The combined 5-40 MHz signals are then digitized by a 10 bit sampling A/D converter. This results in a baseband digital signal of approximately 1 Gbps. This signal is then time division multiplexed (TDM) with the digitized signal from four other combined receivers and transmitted back to the cabinet via a 2.5Gbps ITU transmitter. At the cabinet, the signals are DWDM muxed with other return path wavelengths and transmitted to the head end. Because the signals are in digital format, an EDFA is not necessary. At the head end, the signals are demuxed, converted back to analog format, and fed to the appropriate CMTS, HDT or VOD controller. This system uses fewer ITU lasers than the pure DWDM return option, and no EDFA's, so it is more cost-effective. However, in the initial deployment shown, it only permits segmentation to the 400 home level, which corresponds to a peak upstream rate of 250 Kbps per sub. In order to provide the same segmentation and bandwidth of the pure DWDM option, every return path segment must be digitized separately without combining. Either additional digital 2.5 Gbps ITU transmitters will be necessary, or the eight digitized return segments must all be multiplexed together before transmission. However, this would require a 10 Gbps transmitter, which must be externally modulated, and is probably cost prohibitive compared to a common, directly modulated 2.5Gbps transmitter.
An final return path option is to employ a more efficient form of digitization of the QPSK and 16-QAM signals in the 5-40 MHz return band. Digital sampling of the return path signal is somewhat effective, but very inefficient. The 5-40 MHz waveform is digitized to produce a 1 Gbps signal, despite the fact the maximum useful information carried by the signal, assuming 16-QAM modulation, is only 100 Mbps. A possible solution to the digital upstream efficiency problem under development is to remotely demodulate the DOCSIS QPSK or 16-QAM upstream signal by moving some of the functionality of the CMTS from the head end to the OSN or mini-node. Utilizing such a technique makes the return path more scalable. A further possible step is to locate an entire reduced-functionality CMTS in the mini-node. The device consists of the PHY portion of a regular CMTS (QAM modulator, upconverter, QPSK demodulator) and a rudimentary MAC layer. A two-way ethernet switching fabric transmits the baseband digital signal from the head end to the mini-node.
SUMMARY
Utilizing a Dense Wavelength Division Multiplexing (DWDM) deep fiber architecture overcomes many of the drawbacks associated with traditional HFC architectures. The system is capable of providing enormous amounts of dedicated bandwidth. The architecture is also completely scalable, and cost-effective when compared with traditional dual-hop HFC architectures. The return path uses a combination of digitization and DWDM to provide segmentation and scalability. Next generation technologies may distribute demodulation presently associated with the CMTS out into the node.
|
