I often get asked the question: “I’m having my HFC network upgraded for DOCSIS 3.1, what KPI’s (Key Performance Indicators) should I put in my contract?”

My first reply typically is: “Who pays the bill?” I do know it's the operator who will be paying the bill, but as always in technology, the right answer depends on many factors. In this blog I will clarify how some of the factors play an important role.

First, I want to start with a common misconception that I’m still confronted with many times: Many people believe it's a must to upgrade an HFC network to be able to use DOCSIS 3.1. This is not correct. DOCSIS 3.1 can be perfectly used on any existing network, and even better, it will even provide benefits (if used in the right way).

Back to the HFC KPIs.

The goal of upgrading the network will be to deliver more speed and/or capacity. It's important to realize that speed and capacity although linked to each other are not the same. From an HFC perspective, the speed you can deliver to your customers is based on the number of customers on the segment (and their usage) and the capacity of the segment. Of course the speed is also determined by the amount of channels the cable modem and CMTS can support, but this is not an HFC parameter. The capacity of the segment is determined by the amount of channels (and their modulation) in the segment.

Excluding the optional frequency expansion up to 1 GHz, there has been no improvement in the technology that improves the capacity of the network until DOCSIS 3.1. An HFC network full of 1.0 channels of 256QAM has the same network capacity as a network full of 3.0 256QAM channels. If our Wi-Fi would still be using the same technology as in the end of 90’s, nobody would even remember what Wi-Fi is. It’s time for a change. DOCSIS 3.1 brings that change by introducing a new modulation and techniques to improve the network capacity, that of course at the same time improves the speeds you can get.

How much capacity improvement can you get? The exact mathematical number for the capacity improvement depends on a whole bunch of factors of which the two most important ones are modulation that will be used, and the usage of more frequency spectrum.

In DOCSIS 3.1 the highest order modulation that is mandatorily supported by cable modems and cable modem termination systems in the downstream is 4096QAM, while in pre-DOCSIS 3.1 it is only 256QAM. 4096QAM provides 12 bits/symbol, while 256QAM only provides 8 bits/symbol. So by using the higher order modulation the capacity could be increased by about 50%. Sounds great, doesn’t it? There is one drawback though: to be able to run 4096QAM the signal arriving at the modem needs to be of pretty good quality. Of course one could build a network where the signal quality is great for every single modem, but at what cost? To be able to demodulate 4096QAM about 41dB CNR is needed compared to only 27 dB for 256QAM, that’s an improvement in signal quality of about 14 dB. Since the signal quality at the cable modem is largely determined by the drop cable, it is pretty expensive to get a 14 dB improvement in signal quality to every single cable modem. So what number should you specify? Well it’s not a single number: it’s a statistical distribution. In DOCSIS 3.1 the CMTS will address the cable modem with the modulation that the CMTS decides is most suitable, meaning that if your network today already has modems that have excellent signal quality, they could already be fit for 4096QAM.

So what statistical distribution should you specify? The cost of your network improvements versus other mechanisms to improve network capacity are the key criteria.

One important factor to take into account is what I personally call, the law of diminishing benefit. Let's take a look at the table below:

Modulation Bit/s symbol Estimated Required CNR [dB]  Benefit over previous row
QPSK 2 9  
16QAM 4 15 +100%
64QAM 6 21 +50%
256QAM 8 27 +33%
1024QAM 10 34 +25%
4096QAM 12 41 +20%

To achieve 2 bit/symbol more modulation efficiency, you always need about 6dB improvement in CNR. But 2 bits/symbol at 10 bits/symbol is only a 20% gain in capacity, while you need 7dB better CNR.

So how does the cost of improving your network with a 20% gain in capacity by network improvement weigh against the cost of a node split?

While performing HFC upgrades it is recommended to also consider frequency expansion. While the decision to perform the frequency expansion might be easy, the required quality in the new spectrum is a different issue. In the end you don’t have to look at the frequency range of your network, but at the capacity of the network. A well-designed and maintained network up to 1 GHz might have more capacity than a poorly designed network up to 1.2 GHz. In the end the capacity of the network, certainly in a DOCSIS 3.1 environment, is not defined by the number of channels you can use, but also by the quality you have in each of these channels, and since DOCSIS 3.1 uses OFDM with an independent modulation per subcarrier (see OFDM(A) in DOCSIS 3.1: multiplexing, modulation and multiple access?) you have to look at the quality of each part of your spectrum. Since the quality of the channel is unique per connected modem and the CMTS addresses the modem with a well-selected modulation, the required KPIs for your HFC network need to be well-chosen taking into account the different cost-influencing factors.

In this blog, we discussed which factors you have to take into account when defining the Key-Performance Indicator for your new HFC network.


Add new comment