Advertisement
Articles
Advertisement

Reverse Path Design for Docsis 3.0 and Beyond - Part 3: Link Budget and System Set-Up

Sun, 11/30/2008 - 7:00pm
Lamar West, Ph.D.

INTRODUCTION
This is the third in a series of articles about the set-up and operation of DOCSIS networks. The demands of DOCSIS 3.0 and 64-QAM (quadrature amplitude modulation) in the upstream make it necessary to revisit our network design approach. The network must be approached from a perspective that considers cable modem termination system (CMTS), HFC plant, and subscriber devices (modems and set-top boxes) as a part of an overall unified system design.

In the previous article we summarized the results of measurements made to quantify the available dynamic range of two representative types of reverse path optical links (DFB and Digital Reverse).  In this paper we propose a reverse path link budget that considers the system impairments typically encountered in HFC networks. The intent of this link budget is NOT to cover every possible HFC configuration.  Rather, we intend to present a scenario that covers the majority of HFC configurations that will be encountered in the foreseeable future. A set of system assumptions will be presented that clearly define the applicability of this budget.

In addition, we will present a set-up guideline that facilitates proper system balance and alignment. This guideline is not absolute. It may be modified to accommodate individual operator requirements. The intent is to present an alignment philosophy that is in harmony with the requirements imposed by the long loop ALC (see article one in this series).

REVERSE SYSTEM ASSUMPTIONS
The following basic assumptions apply to the link in this analysis:

  1. The highest modulation index of any upstream carrier is 64-QAM.
  2. There are no more than 7 RF carriers present on any given upstream optical link at any given time.
  3. There is no more than 4-way node combining in the hub or headend before upstream signals reach a CMTS blade.
  4. The CMTS upstream RF input amplitude window around the commanded receive level is set at no more than ±4 dB (i.e. the allowable range of amplitude variation of an upstream signal with respect to the commanded signal level before the CMTS will adjust the signal level at the cable modem transmitter in a subscriber’s home).
  5. The frequency response over the reverse band for the RF splitter/combiner network in the hub or headend is no greater than ±1 dB.
  6. There are no more than 25 System Amps and 35 Line Extenders connected to any given optical link.
  7. There are no multiple optical hops (i.e. optical detection and re-lasing).
  8. There is no upstream optical amplification.
  9. The ingress level is sufficiently low so that it does not contribute significantly to the overall optical modulation index (OMI).
  10. The RF power at the output of the optical receiver does not exceed the maximum power handling capability of that receiver.
  11. The optical power will not exceed the maximum limit into any optical receiver.

These basic assumptions cover the vast majority of reverse links that are presently in operation. Links that do not comply with these assumptions may be designed and analyzed by means of other methods.  Note that items number 4 and 5 result in a worst case carrier inequality at the optical link input of ±5 dB.

These assumptions represent a severe operating scenario for the reverse path in an HFC system. Most systems will not be burdened by all of these constraints. However, adequate operation will be achieved under less severe constraints if adequate operation can be demonstrated when these assumptions apply.

The other performance impairments considered in our link budget are:

  1. Optical link noise (laser RIN, optical RX noise, etc.)
  2. Optical crosstalk
  3. Optical fiber nonlinearity (primarily stimulated Raman scattering or SRS)
  4. RF amplifier noise
  5. Average ingress noise (averaged over optical links to be combined)

OPTICAL LINK NOMINAL OPERATING LEVELS
An error budget based on these constraints is presented in Appendix 1. This error budget can be used to determine nominal operating levels for the reverse path optical link.  The error budget is designed to allocate part of the available dynamic range of a reverse path optical link to the other impairments typically encountered the system. Based on this budget, the BER dynamic range limit as measured on the optical link alone should be increased by 13 dB on the low carrier power side (left hand side) and decreased by 3 dB on the high power side (right hand side) to accommodate these other impairments.

Consider the case of an analog DFB reverse optical link. BER dynamic range data for such a link is shown in figure 1. This data was taken with a 0 dB transmitter input attenuator.

7 dB link loss with Uneven Carrier Amplitudes 
Figure 1. 7 dB link loss with Uneven Carrier Amplitudes

Note that this data was taken with seven QAM carriers, six of which were at nominal amplitude +5 dB and one of which was at nominal amplitude -5 dB (see article two in this series). This is consistent with our system assumptions 4 and 5 described previously.

Based on our previously defined goal of a BER of 1E-7, the carriers on the analog DFB reverse link when considered alone must be at a level between –13 dBmV and +6 dBmV.  If we add in the limitations imposed by our link budget we must constrain the carriers to be between 0 dBmV and 3 dBmV.  In other words, when all of the impairments in our link budget add on a worst case basis we will still obtain adequate performance! These results suggest a nominal per carrier input amplitude of 1.5 dBmV (approximately +2 dBmV) for this reverse DFB optical link. 

Similarly we can consider a digital reverse link such as the Cisco bdr® product. BER dynamic range data for such a link is shown in figure 2. This data was taken with a 0 dB transmitter input attenuator.

BER for bdr® Link with Uneven Carrier Amplitudes 
Figure 2. BER for bdr® Link with Uneven Carrier Amplitudes

As with the DFB data shown previously, this data was taken with seven QAM carriers, six of which were at nominal amplitude +5 dB and one of which was at nominal amplitude -5 dB (see article two in this series).

Based on our previously defined goal of a BER of 1E-7, the carriers on the bdr® reverse link when considered alone must be at a level between –7 dBmV and +11 dBmV. If we add in the limitations imposed by our link budget we must constrain the carriers to be between +6 dBmV and +8 dBmV. In other words, when all of the impairments in our link budget are considered we still obtain adequate performance! These results suggest a nominal per carrier input amplitude of +7 dBmV for this bdr® reverse link.

REVERSE SYSTEM ALLIGNMENT PROCEDURE EXAMPLE
The simplified technique presented herein is intended to illustrate the alignment philosophy. Details may be added to suit the requirements of individual installation constraints. 

A simplified block diagram of the reverse path system is shown in figure 3.

Simplified Reverse Path Diagram 
Figure 3. Simplified Reverse Path Diagram

The first part of the alignment is to complete a paper design that defines the desired operating levels. The paper design should specify critical RF carrier operating levels. In this example we have chosen the following operating levels:

  1. The reverse RF carrier amplitude at the node port will be +17 dBmV.
  2. The reverse optical transmitter will use a DFB laser.  The desired reverse optical transmitter RF operating level will be +2 dBmV as described previously.
  3. The desired reverse RF carrier amplitude at the output of the optical receiver will be +25 dBmV.
  4. The reverse RF carrier commanded signal level will be 0 dBmV at the CMTS’s upstream port.

The most critical part of the reverse system set-up is alignment of the optical link. As described in the first article of this series, alignment of the optical link involves adjusting the gain/loss structure between the output of the reverse optical receiver and the upstream RF input of the CMTS.  This alignment is done open loop. By this we mean that a signal of known amplitude is injected into the system. This signal is of fixed amplitude (not adjusted by the CMTS). The signal may be a CW RF carrier, a sweep signal or any other convenient reference of known amplitude. This signal is injected into the node output test point. A diagram of this process of this process is shown in figure 4.

Transmitter Input Pad Selection 
Figure 4. Transmitter Input Pad Selection

The node in our example has a 20 dB forward output test point. A signal generator is connected to this test point in order to inject a reverse test signal. The reverse RF carrier design amplitude at the node port is +17 dBmV. A level of +37 dBmV is injected at the test point to match this design level amplitude (+17 dBmV + 20 dB = +37 dBmV).

The nominal operating point for the reverse transmitter with a 0 dB input pad is +2 dBmV. The node has miscellaneous reverse input losses of 2 dB. A reverse transmitter input attenuator of 13 dB is chosen to match the test carrier amplitude to the transmitter nominal level. This is verified by a measurement of –18 dBmV at the transmitter 20 dB test point (+2 dBmV - 20 dB = -18 dBmV). This procedure will give a reference carrier at the proper laser OMI for adjustment of the losses between the reverse optical receiver and the CMTS reverse input.

The next step is adjustment of the losses associated with the reverse optical receiver in the hub or headend. This is done with the test signal generator still connected to the node output test port. The signal level meter is moved to the output of the reverse optical receiver in the hub or headend. This is indicated in figure 5.

Reverse Optical Receiver Output Pad Selection 
Figure 5. Reverse Optical Receiver Output Pad Selection

The reverse optical link in this example has a link gain of 32 dB. Therefore it is necessary to select a 9 dB attenuator at the output of the optical receiver to achieve the design goal of +25 dBmV as measured on the signal level meter. In this example no optical attenuator was inserted before the optical receiver.

In this example we consider a splitter/combiner network that presents a total of 21 dB of loss as shown in figure 6.

CMTS Reverse Input Pad Selection 
Figure 6.  CMTS Reverse Input Pad Selection

The signal level meter is connected to the output of this network. In order to obtain the desired 0 dBmV level to match the CMTS command signal level a 4 dB pad is required. Note that the loss and gain between the optical receiver and the CMTS determine the operating levels of the optical transmitter (OMI).

Once this balance and alignment of the optical link and losses in the hub/headend is complete it should not be altered. These attenuators set the operating point of the reverse optical link. Once the long loop ALC is activated the CMTS will ensure that the optical link operates its nominal level. However, changes to the pads or gain stages in this part of the plant can adversely affect operating levels not only in the optical  link but also in the RF plant and can change the transmit levels of set-top boxes and cable modems!

At this point the signal generator may be moved out further into the RF plant in order to complete balance and alignment. Good engineering practice dictates that losses in the reverse band should be designed to force reverse path transmitters in modems and set-top boxes into the upper end of their transmit power range. Doing so will maximize carrier-to-ingress ratios.

CONCLUSIONS
These results indicate that the reverse optical links tested provide a substantial dynamic range even under full carrier loading and worst case carrier amplitude inequality. With full loading and worst case amplitude inequality the DFB link gave a 1E-7 BER dynamic range of 19 dB.  With full loading and worst case amplitude inequality the bdr link gave a 1E-7 BER dynamic range of 18 dB.

For short links (less than 7 dB link loss) the analog DFB and bdr® (digital reverse) link give approximately the same dynamic range. The DFB dynamic range starts to decrease as the link loss is increased beyond 7 dB. However, the bdr® link maintains its dynamic range at much higher link losses than 7 dB.

A reverse path link budget was presented that considered the vast majority impairments that affect reverse path link performance. Even when these impairments add on a worst case basis there is still adequate dynamic range for the link to give acceptable performance.

A balance and alignment procedure was presented that initially focused on set-up of the optical link and losses between the optical receiver and CMTS. Once these are properly adjusted the long loop ALC will ensure that the operating levels in the optical plant will remain at their nominal point.

APPENDIX 1:  LINK BUDGET DETAILS
We first consider the factors that limit the minimum carrier amplitude as referenced to the input of the optical transmitter for adequate performance in our system. Our performance goal for the link is an uncorrected BER of 10E-7. A composite end-of-line carrier-to-noise ratio of 27.4 dB is required in order to achieve a BER of 10E-7 with 64 QAM.[1]

Consider an ideal noiseless link with various impairments and noise sources summed in with the desired signal. A block diagram of this network is shown in figure A1.

Block Diagram for Noise Calculation 
Figure A1. Block Diagram for Noise Calculation

Impairments that affect the noise performance consist of:

  1. Optical link noise (laser RIN, optical RX noise, etc.)
  2. Optical crosstalk
  3. Optical fiber nonlinearity (primarily stimulated Raman scattering or SRS)
  4. RF amplifier noise
  5. Average ingress noise (averaged over optical links to be combined)

Working backwards through the network we first encounter the combiner/splitter network in the hub or headend. We have constrained this network to contain 4-way combining at maximum. Therefore, the noise funneling from this network can degrade the carrier-to-noise by as much as 6 dB. The required carrier-to-noise at the input to this network (optical receiver output) must be at least 33.4 dB.

Optical crosstalk occurs in multi-wavelength optical lengths as a consequence of the finite isolation of optical multiplexers (MUXs). As a consequence, a receiver intended for a specific wavelength may see light from another wavelength. Optical MUX isolation is typically specified at greater than 35 dB. After detection this results in a RF interference of 70 dB. We reduce this amount by 10 dB to accommodate our RF carrier inequality budget of +/-5 dB. The resulting crosstalk budget places noise at –60 dBc.

The second source of optical link noise is fiber nonlinearity (mainly Stimulated Raman Scattering (SRS)). We budget RF crosstalk resulting from fiber nonlinearities at a worst case level of -42 dBc. This must be worsened by 10 dB in order to account for the potential amplitude inequality between RF carriers of +/-5 dB resulting in an equivalent RF crosstalk of -32 dBc. 

The effect of SRS in a WDM system varies from wavelength to wavelength. There is a single wavelength that will exhibit the worst case performance because of SRS. The other wavelengths will have substantially better performance. Therefore, the 6 dB penalty for combining in the hub or headend will not apply to this impairment. For the purposes of calculation we will simply increase the SRS limit by 6 dB to accommodate our method of calculation that assumes equal contributions from each of the four combined links. Therefore the effective SRS impairment averaged over four wavelengths will be approximately -38 dBc.

An analysis of RF plant noise requires an estimate of the maximum number of active devices associated with any node. For the purposes of this analysis we have assumed a worst case of 25 System Amps and 35 Line Extenders for any node.

The thermal noise nose floor for a 75 ohm system in a 5.12 MHz bandwidth (the equivalent noise bandwidth of a 64-QAM carrier in our analysis) is -58 dBmV. The noise figure of a Cisco System Amp is 12 dB.  Consequently, the equivalent noise floor of a system amp is

A1 (A1)

Combining the noise from 25 System Amps gives a composite equivalent noise floor of

A2  (A2)

Similarly, the noise figure of a Cisco Line Extender is 7.5 dB. The equivalent noise floor of a Line Extender is

A3  (A3)

Combining the noise from 35 Line Extenders gives a composite equivalent noise floor of

A4  (A4)

Combining the noise from the Line Extenders and the System Amps on a power basis we get

A5  (A5)

Based on a typical carrier input level of +17 dBmV at each RF active device, the resulting RF plant noise at a level with respect to the carrier will be

A6  (A6)

For the purposes of this analysis we have chosen a goal of average carrier-to-broadband ingress noise for 64-QAM carriers of 40 dB. This means that the averaged noise over any four reverse path links to be combined in the headend will be 40 dB below the carrier in the channel equivalent noise bandwidth (i.e. channel symbol rate in hertz). In reality, the noise will vary from link to link, but the average of 40 dB is suggested to permit the 6 dB penalty from signal combining referenced previously. Note that we are not considering narrowband interfering carriers, but rather broadband noise such as impulse noise.

We now begin with our goal of a carrier-to-noise at the output of the optical receiver of 33.4 dB. We can then back out the required optical link noise from the end goal.

A7  (A7)

Consequently, we require a carrier-to-optical link noise of 37.5 dB. This is a 10.1 dB increase over the required CNR of a link as measured in the lab (i.e. no ingress, no RF plant noise, single wavelength, etc.). An additional 3 dB is added to accommodate the variation of lab results form unit to unit and over temperature extremes. Therefore a total increase in carrier level of 13.1 (approximately 13 dB) over the minimum level required for a BER of 10E-7 as measured for a single optical link in the lab is required to accommodate real system impairments.

We now consider the limits on maximum carrier amplitude as referenced to the input of the optical transmitter for adequate performance in our system. In the previous analysis the impairments could be treated as behaving like Gaussian noise (i.e. thermal noise, RIN noise, Optical receiver noise, interfering QAM subcarriers, etc.). However, the sensitivity of QAM, particularly 64-QAM, to clipping noise is significantly higher than the sensitivity to Gaussian noise. Therefore clipping noise is the dominant impairment mechanism limiting the maximum carrier amplitude.

The clipping limits for maximum carrier amplitude may be taken directly off of the BER curve. Variations from unit to unit and over temperature will require a decrease of 3 dB from the maximum level required for a BER of 10E-7 as measured for a single optical link in the lab. It is assumed that RF amplifiers will not contribute significantly to the overall clipping distortion. 

In summary, this error budget analysis indicates that an increase of minimum carrier amplitude of 13 dB and a decrease of maximum carrier amplitude of 3 dB is the level required for a BER of 10E-7 as measured for a single optical link in the lab. This combined with the worst case carrier amplitude inequality should give the limits of performance in the corner case scenario based on the link assumptions described earlier

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[1] Ciciora, Farmer and Large, Modern Cable Television Technology, © 1999 Morgan Kaufmann Publishers, San Francisco, CA, pp. 187-188

Topics

Advertisement

Share This Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading