9.6 Spectral Compatibility of the DSL Systems

   


This section describes the spectral compatibility of the echo canceled (EC) transmission DSL systems relative to other DSLs deployed in the loop plant. The transmission DSLs included are ISDN, HDSL, and SDSL. For all spectral compatibility studies done in this chapter, only 50-pair cables with 26-gauge wire are assumed in each case.

9.6.1 ISDN

The transmit spectrum of ISDN is shown in Figure 9.5. Because ISDN is an echo-canceled symmetric transport system, we need to consider the effects of SNEXT. Because the SNEXT spectrum completely overlaps the ISDN transmit spectra, we expect this disturber to dominate over other disturbers whose spectra only partially overlap. Although the HDSL spectrum may fully overlap that of ISDN, the PSDlevel of HDSL will be lower than that of ISDN because the two systems have the same total transmit power.

In the evaluation of ISDN transceiver performance, echo canceler performance is considered for the ISDN transceiver. Seventy dB of echo cancellation have been achieved in practical ISDN transceivers. If there is no crosstalk in the cable, then the performance of the ISDN transceiver is limited by the performance of the echo canceler. Specifically, consider the scenario where we have a 50-pair cable of 26-gauge wire. If this cable has a single ISDN transmission system deployed and the remaining 49 wire pairs are not used, then the reach of an ISDN transceiver operating at a bit error rate (BER) of 10 -7 with 6 dB of margin is 20.5 kft on 26-gauge wire. When we add one additional ISDN signal into the cable, the added single SNEXT disturber reduces the reach to 20 kft. With ten SNEXT disturbers, the reach is 19.1 kft and with twenty-five disturbers, the reach is 18.6 kft. Finally, if the whole 50-pair cable is filled only with ISDN systems, the maximum achievable reach of an ISDN system would be 18 kft, limited by SNEXT. Figure 9.11 shows a summary of the the ISDN reach as a function of SNEXT level.

Figure 9.11. ISDN Reach as a function of SNEXT.

graphics/09fig11.jpg

Figure 9.12 shows the spectral plots of the transmit and receive signals of an ISDN system operating on an 18 kft 26 gauge loop. Also shown in the figure are the insertion loss of the 18 kft loop and the 49-SNEXT plus the amount of echo that the echo canceler does not eliminate. The area between the received signal and crosstalk curves defines the received SNR.

Figure 9.12. Spectral plots of ISDN reach with 49-SNEXT.

graphics/09fig12.jpg

We now consider the case when the cable includes a mixture of other DSL services. Other DSLs considered are HDSL, SDSL, and CAP RADSL upstream and downstream. For each case we consider the worst case scenario measuring the reach of ISDN in the presence of forty-nine disturbers from the "other" DSL in question. Figure 9.13 shows a comparison of the ISDN reach on 26-gauge wire in the presence of forty-nine disturbers from each of the "other" DSLs. Because of the total spectral overlap, SNEXT is a worse disturber to ISDN than any of the "other" DSLs, because SNEXT only has a partial overlap of its spectra with that of ISDN. Although the HDSL spectrum fully overlaps the ISDN spectrum, the PSD level of HDSL is lower than that of ISDN because the two systems have the same transmit power. The lower PSD level of HDSL therefore introduces less crosstalk into the ISDN band than does SNEXT.

Figure 9.13. ISDN Reach as a function of other NEXT disturbers.

graphics/09fig13.jpg

In summary, SNEXT is the worst case disturber to ISDN basic rate. Deploying "other" services in the same cable with ISDN will have less impact on the performance of ISDN than if only ISDN was deployed in the cable.

9.6.2 HDSL

The transmit spectrum of HDSL is shown in Figure 9.6. As with ISDN, HDSL is also a symmetric echo canceled system. In the evaluation of HDSL transceiver performance, echo canceler performance is considered for the HDSL transceiver. Seventy dB of echo cancellation have been achieved in practical HDSL transceivers.

If there is no crosstalk in the cable, then the performance of the HDSL transceiver is limited by the performance of the echo canceler. Specifically, consider the scenario of a 50-pair cable containing 26-gauge wire. If there is only a single HDSL transmission system deployed and the remaining forty-nine wire pairs are not used, then the reach of an HDSL transceiver operating at a BER of 10 -7 with 6 dB of margin is 13.7 kft on 26-gauge wire (Figure 9.14). When we add one additional HDSL system into the cable, the added single SNEXT disturber reduces the reach by 1.7 kft to 12 kft. With ten SNEXT disturbers, the reach is 10.6 kft, and with twenty-five disturbers, the reach is 10.1 kft. Finally, if the whole 50-pair cable is filled only with HDSL systems, the maximum achievable reach of an HDSL system would be 9.5 kft, limited by SNEXT.

Figure 9.14. HDSL reach as a function of SNEXT.

graphics/09fig14.jpg

Consider now the cases where HDSL is deployed in the cable with a mixture of other DSL services. Relative to HDSL, the "other" DSLs considered are ISDN, SDSL, and CAP RADSL upstream and downstream. For each case we consider the worst case scenario measuring the reach of HDSL in the presence of forty-nine disturbers from the "other" DSL in question. Figure 9.15 shows a comparison of the HDSL reach on 26-gauge wire in the presence of forty-nine disturbers from each of the "other" DSLs. Because of the total spectral overlap, SNEXT is the worst disturber to HDSL than any of the "other" DSLs. The other DSL spectra only have a partial overlap with that of HDSL.

Figure 9.15. HDSL reach as a function of NEXT from other services.

graphics/09fig15.jpg

In summary, SNEXT is the worst case disturber into HDSL. Because the other spectra in the cable have near-end crosstalk spectra that do not fully overlap with the HDSL transmit spectrum, the overall interference will be less than the NEXT from other HDSL signals.

9.6.3 SDSL

SDSL systems are echo-canceled systems and, therefore, are disturbed the most by SNEXT. In fact, SNEXT is the dominating disturber to SDSL. The wider the signal bandwidth, the greater the level of SNEXT. The bit rate of the signal is directly proportional to the signal bandwidth, so the reach of SDSL systems decreases with increasing bandwidth (hence, increasing bit rate). Figure 9.16 shows the reach of SDSL systems relative to SNEXT. Note the decrease in reach with increasing bit rate.

Figure 9.16. SDSL reach versus 49-SNEXT.

graphics/09fig16.jpg

Because SDSL performance is limited by SNEXT, NEXTs from other DSLs do have as much impact on SDSL reach. However, depending on the signal bandwidth, SDSL systems may impact the performance of other DSL systems such as ADSL or CAP RADSL. The spectral compatibility of SDSL into other systems is discussed in the subsequent sections.

In summary, SNEXT is the dominant disturber into SDSL.

9.6.4 ADSL

In this study, we consider the spectral compatibility of FDM-based ADSL. Figure 9.9 shows the transmit and near-end crosstalk spectral plots of the upstream and downstream DMT ADSL channels. The spectral compatibility of DMT ADSL and CAP RADSL are similar in that neither system has SNEXT to deal with because the FDD transmission scheme places the transmitted energy in a frequency band separate from the receive band for the same end of the line. They both have SFEXT, and they must deal with NEXT from other DSL services in the same cable.

As with CAP RADSL, DMT ADSL is a variable bit rate system, and the actual bandwidths of the upstream and downstream channels may vary depending on the bit rate and crosstalk. Shown in Figure 9.9 is the maximum possible useful bandwidth for the upstream and downstream channels.

To evaluate the spectral compatibility of the DMT upstream channel with other services, we compute the reach of 272 kb/s DMT upstream channel in the presence of crosstalk from other DSL services. Figure 9.17 shows a comparison of the reach of a 272 kb/s upstream DMT system in the presence of NEXT from HDSL, T1 AMI, ISDN, 784 kb/s SDSL, and SFEXT. Clearly, SFEXT is the best noise environment providing the least amount of interference. T1 AMI also provides low interference into the upstream channel because the AMI signal energy is very low in the DMT upstream channel band. The dominant disturbers into the upstream channel are HDSL and SDSL because the NEXT from these services provides full bandwidth overlap with the DMT upstream channel. The ISDN spectrum has partial overlap with the DMT upstream channel, and therefore has less impact on upstream channel reach than does HDSL and SDSL.

Figure 9.17. Upstream DMT spectral compatibility with other DSLs.

graphics/09fig17.jpg

To evaluate the spectral compatibility of the DMT downstream channel with other services, we compute the reach of a 680 kb/s DMT downstream channel in the presence of crosstalk from other DSL services. Figure 9.18 shows a comparison of the reach of a 680 kb/s downstream DMT system in the presence of NEXT from HDSL, T1 AMI, ISDN, 784 kb/s 64-CAP/QAM SDSL, and SFEXT.

Figure 9.18. Downstream DMT spectral compatibility with other DSLs.

graphics/09fig18.jpg

As with the upstream channel, SFEXT is the best noise environment providing the least amount of interference; however, its reach is lower that the upstream channel because the loop has higher loss in the frequencies of the downstream channel. Contrary to the upstream, T1 AMI provides the dominant level of interference into the downstream channel because the AMI signal energy is highest in the DMT downstream channel band. Because of the significant bandwidth overlap with the downstream channel, HDSL is the next dominant disturber in line. ISDN and SDSL have the least impact of NEXT into the DMT downstream channel. The degradation in reach from T1 AMI versus the best case of SFEXT is approximately 6000 ft; the corresponding in reach from HDSL is approximately 5000 ft.

In summary, HDSL and SDSL are dominant disturbers into the upstream channel of ADSL. T1 AMI is the dominant disturber into the ADSL downstream channel. The best case for deployment of FDM ADSL services is to fill the cable completely with ADSL and eliminate all NEXT. If the cable contains a mixture of DSLs, then NEXT from HDSL and SDSL are the dominant disturbers into the upstream channel, and T1 AMI and HDSL are the dominant disturbers into the downstream channel.

9.6.4 T1 AMI

Figure 9.19 shows the system model for determining the spectral compatibility of the downstream ADSL channel into T1 AMI. Because T1 AMI is a repeatered link, there are numerous points to consider observing the crosstalk, and they are labeled crosstalk points #1 and #2 in the figure.

Figure 9.19. Crosstalk scenarios for DSLs into T1 AMI.

graphics/09fig19.jpg

In the conventional provisioning of T1 links, the first repeater is placed at a maximum of 3 kft from an end-point and a maximum of 6 kft between repeaters. Note, however that T1 lines were originally designed as trunk lines to interconnect COs and the wire gauges were 22 AWG or 19 AWG. Because the distribution plant usually uses 26 AWG wire (thinner than 22 or 19 AWG) directly out of the CO, the provisioning rules used in the distribution plant are not ubiquitously known. However, in this study, we will assume a worst case scenario of 26-gauge wire using the same repeater spacing rules for the trunk plant.

The ADSL downstream transmit signal is strongest at the CO transmitter output. So at the CO (crosstalk point #1 in Figure 9.19) the ADSL downstream signal introduces the strongest level of crosstalk into the T1 AMI receiver. The T1 AMI signals have maximum energy at the transmitter output of both the end units and repeaters. In the first repeater span, the loop segment length is 3 kft and so the received AMI signal would not be attenuated as much as it would be in a midspan repeater spacing of 6 kft. At crosstalk point #2, the downstream ADSL signal is attenuated by 3 kft of cable, and so the crosstalk level into the first repeater would be attenuated by that amount.

To estimate the performance of the AMI signal, we compute the signal-to-noise ratio (SNR) at the AMI receiver input and crosstalk points #1 and #2. In each case, the number of downstream ADSL disturbers assumed is twenty-four. The SNR is measured in two ways: (1) at the T1 AMI center frequency of 772 kHz, and (2) averaged through out the entire T1 AMI band to the first null, that is, 0 to 1.544 MHz. In the AMI receiver, we assume that the receiver provides automatic gain adjustment, no equalization, and ideal time sampling. To achieve a BER of 10 -7 , we assume a 17.5 dB SNR is required for the three-level signal at the input to the AMI receiver. The margin achieved is the difference in SNR at the receiver input and the 17.5 dB reference value.

Table 9.3 below shows the spectral compatibility computation results with twenty-four ADSL downstream channel disturbers. The third column shows the SNR margin seen at the AMI receiver inputs measured at the AMI center frequency of 772 kHz. For both crosstalk points #1 and #2, the SNR margin is roughly 1 dB so the T1 AMI system should still provide service with better than 10 -7 BER. The last column shows input SNR averaged over the entire T1 AMI band. At crosstalk point #1, the averaged SNR is roughly the same as the SNR at the center frequency. For crosstalk point #2, the averaged SNR is greater than center frequency SNR because of the greater attenuation suffered by the AMI signal on the 6 kft loop.

In summary, an estimation of spectral compatibility of ADSL with T1 AMI was provided using very pessimistic assumptions. Specifically, the same provisioning rules of repeater spacings on 22- and 19-gauge wire was applied to 26-gauge wire and the losses seen on 26-gauge wire were significantly greater. In all cases of twenty-four ADSL disturbers into a T1 AMI receiver, the input SNR is at least 1 dB greater than that required for achieving 10 -7 BER performance.

If the repeater spacings are shorter than those assumed here, then the margins will improve. Based on this data, we expect ADSL to not degrade T1 AMI service in the distribution plant. As shown earlier, T1 AMI is the dominant disturber into the ADSL downstream signal.

Table 9.3. Spectral Compatibility Computation Results with 24 ADSL Disturbers

Crosstalk Point

Center Frequency (772 kHz)

Averaged SNR (0 to 1544 kHz)

SNR (dB)

Margin (dB)

SNR (dB)

Margin (dB)

#1

18.7

1.2

18.8

1.1

#2

18.8

1.3

25.5

8.0



   
Top


DSL Advances
DSL Advances
ISBN: 0130938106
EAN: 2147483647
Year: 2002
Pages: 154

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net