Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_EPOC] Study Group Questions



Hal,

 

If I understood you right, Homeworx used in QAM-32 for OFDM while we have targeted QAM-4096 that requests much higher SNR – this is one more difference and one of the factors dictating the rules for signal power level. Another factor is a peak to average.

I agree with your remark about an aggregate signal. However, we should validate the differences in the peak to average ratio behavior between QAM and OFDM. Once I simulated 160 channels 6MHz QAM256 signal and concluded that after the aggregation of N channels (~16, as I remember) the PAPR will be close to 15dB and will grow very slowly for a higher channels’ number. I believe, the same kind of simulation is required for OFDM in order to understand the limitations for different deployment scenarios. It is very probably, PAPR reduction algorithm(s) will be required for some cases.

 

Thanks,

Boris

   

 

From: Hal Roberts [mailto:Hal.Roberts@xxxxxxxxx]
Sent: Thursday, January 05, 2012 11:03 AM
To: Boris Brun; STDS-802-3-EPOC@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_EPOC] Study Group Questions

 

Boris,

 

I agree. I probably should have worded the statement, "OFDM was proven to coexist with HFC QAM and analog channels" to "OFDM has been proven capable of coexisting with HFC QAM and analog channels".

 

With respect to the higher QAM, I would not be worried about the upstream because the Homeworx system used the same waveform in both directions and coexistence with adjacent channel downstream QAM-256 was shown. QAM-1024 downstream may be a concern, however.

 

It is true the number of carriers in the EPoC will be higher, so that is one difference. However the spectral power density of the OFDM should be the same as that of the QAM signals. I believe it can be shown (as I stated before) that in the current multichannel QAM-256/analog environment we have already reached the point that the aggregate signal may be modeled as Gaussian noise. The fact that we use the NPR (noise power ratio) test which injects Gaussian noise to load the system is tacit admission of that. So the higher number of carriers and the impact on the aggregate signal peak to average will be negligible.

 

Fundamentally I think that the DOCSIS 3.0 signal requirements that we worked so hard on in the DRFI will turn out to be adequate for EPoC. Certainly that is where we should start. I would suggest that we look to the DRFI and see what requirements, if any, might need modification to accommodate OFDM.

 

Best regards,

 

Hal

 

-----Original Message-----
From: Boris Brun [mailto:boris.brun@xxxxxxxxxxxxxxx]
Sent: Thursday, January 05, 2012 8:22 AM
To: Hal Roberts; STDS-802-3-EPOC@xxxxxxxxxxxxxxxxx
Subject: RE: [802.3_EPOC] Study Group Questions

 

Hal,

 

It's good to know that once OFDM worked for HFC in both directions. Another example can be DVB-C2 protocol that defined OFDM for broadcasting, but probably was not deployed yet.

The difference between Homeworx and EPOC, from QAM/OFDM coexistence point of view, is a number of OFDM carriers and a variety of scenarios for QAM, Analog and OFDM carriers allocation over the spectrum.

Another difference can be a higher level of QAM constellation. To my knowledge, 10 years ago QAM-64 wasn't defined for US, but today there are many talks about QAM-256 for US and QAM-1024 for DS.

 

I absolutely agree with you that the properly designed OFDM signal can co-exist with Analog and QAM on HFC. We just have to define this signal and build a guidelines for all possible deployment scenarios.

 

Thanks,

Boris

 

-----Original Message-----

Sent: Tuesday, January 03, 2012 10:21 PM

Subject: RE: [802.3_EPOC] Study Group Questions

 

Boris,

 

While it is not well known within the cable industry, OFDM was proven to coexist with HFC QAM and analog channels 15 years ago (at least OFDM using 6MHz wide channels) in both upstream and downstream HFC.  ADC telecommunications, where I worked developing this technology, created an HFC telephony system called "Homeworx", that used QAM-32 OFDM with 8kHz wide subcarriers in FDD mode.  The power density of the OFDM channels were set to the same level as the QAM digital channels, whether upstream or downstream. The OFDM subcarriers occupied approximately 5MHz of the bandwidth. No additional guard band was used for both upstream and downstream.

 

This technology was initially deployed during '97 by Optus in Australia.  Later it was deployed in the US by MediaOne, Adelphia, ATT (and inherited and operated by Comcast after that), and a number of smaller MSOs. Over 500,000 lines were eventually deployed in the US. I know for sure that ATT in Los Angelis operated 256-QAM channels directly above and below the Homeworx OFDM channel. Extensive testing was done internally by ADC and by ATT to determine that this was safe.

 

In short, a properly designed OFDM channel can be treated like a QAM channel. Yes the OFDM peak to average ratio is higher but that doesn't matter once you have large numbers of 256-QAM channels on a plant as the peak to average curve starts to look like Gaussian noise.

 

It has been a long time but I could probably dig up relevant specifications we used for this. I am not saying some new requirements should not be crafted. However I think we can proceed confidently knowing that if this technology could coexist with adjacent QAM-256 over 10 years ago that new SoC technology applied to OFDM will easily be able to achieve this.

 

Best regards,

 

Hal Roberts

System Architect

Calix Networks

 

-----Original Message-----

Sent: Thursday, December 15, 2011 9:32 PM

Subject: Re: [802.3_EPOC] Study Group Questions

 

Hi Mark,

 

I think that coexistence issue can't be solved by manufacturer only. From my perspective, the lower level of coexistence is a coexistence between PHYs, or ability to deliver different kind of signals over the same network. It is not obvious that traditional HFC signals, analog video and QAM, will coexist with OFDM or other kind of modulation. Like it wasn't obvious that QAM can coexist with analog video signal and Cablelabs issued the DOCSIS DRFI document specifying the requirements to QAM manufacturers for coexistence with analog signals. Also, different cable network topology will cause a different signal distortion. As result, the Study Group should learn the influence of EPOC's PHY on existing signals for each network topology.

 

Regards,

Boris

 

 

 

-----Original Message-----

Sent: Thursday, December 15, 2011 9:30 PM

Subject: Re: [802.3_EPOC] Study Group Questions

 

Hi All,

 

I'd like to encourage, as per the prior "cry", that contributions are definitely in order and that merits can be discussed when presented.

 

Personally, I think coexistence issues can be solved by manufacturers and this will probably serve in boosting both the home gateway and perhaps "campus" gateway where other legacy systems exist.  EPON/EPoC will just be the service provider "WAN port" on the such gateways.

 

As cable operator contribution in the CFI has already laid out 3 cable network topologies that we have to study as well as the MxU topologies, my suspicion, and Howard can speak to this, is that the Study Group will be encouraged to focus work towards a single PHY standard; i.e. one that can work over all the topologies.  However, that focus is highly susceptible to cable operator input...<g>

 

Mark

 

________________________________________________________________________

 

To unsubscribe from the STDS-802-3-EPOC list, click the following link:

 

________________________________________________________________________

 

To unsubscribe from the STDS-802-3-EPOC list, click the following link: