Full Duplex DOCSIS
By Brady S. Volpe
Putting the symmetry in docsis and enabling increased bandwidth and speed
Why does symmetry matter you ask, especially in broadband? Well you have symmetry in fiber. Big deal, right? Yes, now FDX allows for coax to do something that it could never do before, have symmetry just like fiber. So FDX continues the alphabet soup of our industry but through symmetry allows for increased broadband and speed in coax. So let’s take another bite of the alphabet soup together. In my last article I covered remote PHY (R-PHY) and touched on how this was an enabling technology for full duplex (FDX) Data-Over-Cable Service Interface Specifications (DOCSIS). I have covered R-PHY a lot this last year on my blog and now it’s time to focus on FDX. So what is FDX DOCSIS? FDX supports the potential to scale to symmetrical throughput, whether one uses R-PHY or R-MAC/PHY. So let’s explore how FDX DOCSIS works and the benefits it has to offer.
FDX DOCSIS is an extension to the DOCSIS 3.1 specification and is included as Annex F in the DOCSIS 3.1 PHY specification. Annex F was introduced in the I12 update on October 2017. So FDX DOCSIS is not a new specification. You can download the DOCSIS 3.1 PHY specification using this short link bit.ly/d31_specs. The basic concept for FDX DOCSIS is to achieve 10 Gbps symmetrical speed on the coax network — symmetrical being the keyword as all previous versions of DOCSIS supported limited upstream bandwidth and speeds.
There are some fundamental concepts to understand with FDX DOCSIS:
- There are no more diplex filters — which implies no actives in the plant after the remote PHY node or shelf
- The cable modem either transmits data or receives data, but does not do both at the same time
- The upstream and downstream share the same frequency band — again, no diplex filters
Spectrum Allocation
Let’s examine how these fundamental concepts differ from any previous version of DOCSIS to date. Section F.72 of the DOCSIS 3.1 PHY specification defines the FDX operational frequency band to be 108 MHz to 684 MHz. This ensures that the start of the band is above any existing mid-split systems, such as 85 MHz systems currently in existence, which assumes that there are legacy cable modems with 85/108 MHz diplex filters inside the cable modem. Remember that DOCSIS must always remain backwards compatible. The upper end of the frequency band being 684 MHz ensures that the band can support three 192 MHz orthogonal frequency-division multiplexing (OFDM) channels.
Let’s verify the math:
3 x 192 = 576 MHz (total bandwidth for three 192 MHz OFDM channels)
Next add the offset for starting at 108 MHz:
108 MHz + 576 MHz = 684 MHz (yes, stop frequency makes sense)
When the cable modem termination system (CMTS) communicates with the cable modem (CM) it must do so using up to four OFDM signals and the CM must be capable of receiving all four OFDM signals. Optionally, the specification states that the CMTS and modem should support up to five OFDM signals in the downstream, yielding a passband that goes up to 879 MHz.
For upstream communications, the cable modem must support transmission from 5 MHz to 684 MHz. Starting at 5 MHz ensures support for legacy modems. The cable modem must support up to seven orthogonal frequency-division multiple access (OFDMA) channels at 96 MHz bandwidth each and a minimum of four advanced-time division multiple access (A-TDMA) channels with an option of eight A-TDMA channels.
It is very important to consider that today’s DOCSIS plants allow the cable modem to receive downstream data and transmit upstream data simultaneously. This is because the downstream and upstream are separated by diplex filters. If you are not familiar with diplex filters, check out this YouTube video:
With FDX DOCSIS, there is not a separate upstream and downstream. It is one channel where the upstream and downstream co-exist. This means that a modem can receive data on the downstream and then transmit data on the upstream as scheduled by the CMTS, but it can’t do both at the same time. By intuition it may at first appear that this will slow down the modem communications, but in fact it is increased.
In previous versions of DOCSIS the cable modem was limited in the downstream, at best having only one or two OFDM channels and in the upstream by having at best two OFDMA channels. While the modem could transmit simultaneously, data was shared with many other modems. With FDX DOCSIS, the modem has a minimum of four downstream OFDM channels and seven upstream OFDMA channels. Further, the service groups for FDX DOCSIS will, in most cases, almost always be smaller because we must eliminate active devices (i.e., amplifiers) after the node. The net result is much more bandwidth in the upstream and downstream to each modem. The time to switch between upstream and downstream transmitting is extremely short and managed by the CMTS, resulting in very high speeds as seen by the subscriber. Figure 1 from the CableLabs site provides a visual example of the change from asymmetrical legacy DOCSIS to FDX DOCSIS.
Modem Interference
While no single modem will transmit and receive at the same time, there will often be two or more modems near the same location which could transmit at the same time another modem is trying to receive. This will cause a co-channel interference (CCI) issue and adjacent channel interference (ACI) between modems.
A function called sounding is used where the cable modems transmit special signals to each other and also receive these signals. Sounding helps the CMTS know which cable modems will experience CCI and/or ACI and then assigns these modems into interference groups. You can view sounding as an augmentation to station maintenance since it occurs periodically and the CMTS is involved to provide specific information to the modems to help reduce CCI and ACI.
FDX DOCSIS becomes much more complicated when echo cancellation is introduced. Echo cancellation is required in FDX DOCSIS to cancel adjacent leakage interference (ALI) and adjacent channel interference (ACI) resulting from the cable modem’s own upstream transmissions. [1]
Figure 1. Existing DOCSIS (asymmetrical — left) vs. FDX DOCSIS (symmetrical — right)
The complexities of FDX DOCSIS become quite deep at this point. In fact, the hardware designers of FDX DOCSIS have been doing significant research, modeling and lab trails to ensure that FDX DOCSIS is even doable in a real-world environment. To make it happen, echo cancellation techniques are absolutely critical. It is important to understand that cable modems will be transmitting at very high levels in the upstream over very wide bandwidths. DOCSIS 2.0, 3.0 and even 3.1 modems transmit in very narrow bandwidths — maximum 200 MHz. But DOCSIS 3.1 modems operating in FDX mode will transmit in the upstream up to 684 MHz. This has never been done in a live cable plant. The upstream RF attenuation losses will be significant The RF bleed-over into adjacent subscribers homes will be significant. Techniques included in the specification, such as echo cancellation and sounding are provided to help overcome these types of challenges.
It is beyond the scope of this article to go into the details of signal echo cancellation, but it is important to mention that the technology is complex and central to FDX DOCSIS functionality.
FDX DOCSIS and R-PHY
As mentioned in last quarter’s article, R-PHY is a building block for FDX DOCSIS. This is best illustrated in Figure 115 of the DOCSIS 3.1 PHY specification, which is illustrated in Figure 2 of this article.
Figure 2 provides a high level plant model of FDX DOCSIS inclusive of the CMTS and FDX node. The FDX node consists of an R-PHY remote PHY device (RPD) plus a launch amplifier. Yes, that is your typical R-PHY node. Notice that there are no active devices such as trunks, bridgers or line extenders between the R-PHY and Tap at Interface E where the drop to the cable modem (Interface F) occurs. Again this is to reiterate that no diplex filters are present because FDX DOCSIS requires bi-directionality between 108 MHz and 684 MHz. A diplex filter would prevent bidirectional communication.
FDX Challenges
FX DOCSIS has many challenges and has not been deployed in a production environment. CableLabs and vendors are in the testing phases with a number of preliminary success stories, but we still have a long road ahead.
As previously mentioned, echo cancellation is critical. While it has shown great success in the lab, there are many real-world scenarios that it must be tested against. How well will it perform with impaired drop cable, bad in-home wiring, and poorly performing mainline taps?
We have never seen cable modems transmitting inside homes at the power levels and bandwidths FDX DOCSIS 3.1 cable modems are expected to operate. We do not yet know what type of unknown issues we may run into.
A number of operators are already looking to migrate the CMTS into the R-PHY. This is called media access control (MAC) PHY. There are a number of benefits to the MAC-PHY node over the R-PHY node as well as disadvantages.
FDX and PNM
We have already seen how proactive network maintenance (PNM) has helped with DOCSIS 2.0 and 3.0. Its power in DOCSIS 3.1 is already being unleashed with much more capabilities. Recently, in the CableLabs InGeNeOs PNM working group we have started working on PNM for FDX DOCSIS. While the specific conversations are under NDA, it is clear that PNM will add a lot of value to FDX DOCSIS, but there is a lot of work yet to be done in order to add the necessary PNM hooks for this aspect of the specification.
Summary
FDX DOCSIS is quite alluring in that being able to provide subscribers with symmetrical 10 Gbps service enables operators to be aggressive and competitive against threats such as AT&T, Verizon and even Google. Further, this can be done with incremental upgrades using existing deployed R-PHY devices and by serving subscribers within the passive access range of the R-PHY. It is quite exciting to see how we can continue to push the capabilities of the HFC plant.
There are some hurdles which we will need to overcome. The first is demonstrating that FDX DOCSIS hardware is able to meet the real-world challenges. Not all DOCSIS 3.1 hardware today supports the full DOCSIS 3.1 specification so it becomes even more important that the hardware is verified before FDX can be fully realized. Next we need to understand where and how we can deploy FDX DOCSIS in the presence of legacy systems and how to gracefully upgrade systems to maximize the return on investment (ROI). Finally, we will need effective tools, both hardware and software such as PNM, to ensure that we can monitor and proactively resolve issues in the FDX environment before they become evident to the end user.
The future looks bright and symmetrical!
—
References: [1] CM-SP-PHYv3.1-113-171220
Brady S. Volpe
Brady Volpe is Founder of The Volpe Firm, Inc and Nimble This LLC. He has nearly 25 years of broadband cable and telecommunications industry experience specializing in RF, DOCSIS, PNM, and Internet Protocol. Mr. Volpe has been providing a wide range of troubleshooting, design solutions, services and seminars for cable operators and broadband companies specializing in DOCSIS, System Design, PNM, and Troubleshooting. He is a highly respected published speaker, both domestically and internationally. Mr. Volpe holds a BSEE and a MSEE.
Shutterstock: Leonardo da Vinci, Public Domain
Pingback: Full Duplex DOCSIS | Enabling increased bandwidth and speed