Howard Rubin
Senior Product Marketing Engineer
Q. What are some of the issues associated with providing a broad range of services for small and medium businesses?
A. There are quite a few things to think about, but one of the key challenges is achieving the network performance needed to support services such as T1 over DOCSIS. This requires bit error rates as low as 1 x 10-8. This can be difficult to ensure over an entire network because some areas of the HFC plant have degraded or are otherwise "impaired." Another challenge is minimizing jitter and latency when transmitting T1 services over a DOCSIS channel that is not synchronized to a T1 clock.
Q. What can be done to ensure reliable T1 service across an entire network, including the areas of impairment?
A. It's important to appreciate that DOCSIS throws out an entire packet if any uncorrectable bit errors exists. So you can think of DOCSIS as amplifying bit errors, and PER (packet error rate) and BER (bit error rate) as being of similar magnitude.
We've invested significant effort in studying how to optimize RF signaling to ensure that a cable operator gets the most from the HFC plant. Much of our testing was done on actual networks. We observed many things during our field testing, but most significantly our analysis revealed that impulse noise has more effect on BER than any other type of noise. Contrary to the conventional wisdom up to that point in time, we saw bursts in channels above 20MHz that were equal to the DOCSIS channel and last for milliseconds, a relatively long period of time.
How can an MSO combat impulse noise? There are a number of ways, but we've determined that RF performance is noticeably enhanced by optimizing both modulation profile changes and receiver register settings in the CMTS. We also found that impulse noise performance with the default configurations did not attain FEC (forward error correction) limits, so we carefully analyzed burst acquisition settings and now achieve FEC-limited performance in our Cuda CMTS.
Q. Can you elaborate on how synchronization and latency are important considerations when deploying T1 services?
A. Since DOCSIS 2.0 transport is not synchronized to T1 clocks, a cable operator must allocate slightly more bandwidth in a UGS upstream flow, to mitigate jitter and packet loss, and buffer the packets to smooth data flow. Careful adjustment of the buffer size can result in acceptable latency with no dropped packets.
Q. How does DOCSIS 2.0 help a cable operator ensure reliable commercial services?
A. The DOCSIS 2.0 standard added a number of enhancements to improve performance and robustness, but it's worth remembering that DOCSIS only provides the tools for obtaining these benefits. Finely-tuned ingress filter DSP software, look-ahead channel hopping, FEC-limited performance and enhanced burst acquisition must be implemented correctly, with tradeoffs between technologies properly balanced, if network performance is to be maximized.
Q. What is the impact of technologies such as S-CDMA and A-TMDA?
A. We've compared A-TDMA (advanced time division multiple access), S-CDMA (synchronous code division multiple access) and legacy TDMA. A combination of extensive field testing and network modeling enabled us to identify which technology performs better in a range of environments.
We found instances where S-CDMA offered advantages over A-TDMA, but also saw examples of S-CDMA failing to match the performance of default TDMA, again under default configuration. We noted that all three nominally struggle to provide the BER required to support reliable T1 service but have identified enhancements that improve immunity to noise and more favorable error rates The experience we've gained has been invaluable in allowing us to optimize CMTS settings to ensure best HFC performance.