Cable networks, just like any other telecom network in the world, had to adapt to the rising demand for data. During the 90s and 00s, these requirements led to the rise of hybrid fiber-coaxial (HFC) networks: optical fibers travel from the cable company hub and terminate into optical nodes, while coaxial cable connects the last few hundreds of meters from the optical node to nearby houses. These connections mainly were asymmetric, with customers having several times more bandwidth to download data than to upload.
In the past decade, the way we use the Internet has changed. With the rise of social media, online gaming, video calls, and independent content creation such as video blogging, users need more upstream bandwidth than ever. These new requirements have led to quick progress in the DOCSIS standards that regulate data transmission over coaxial cables. For example, the latest DOCSIS 4.0 standards allow full-duplex transmission with symmetrical upstream and downstream channels. Meanwhile, fiber-to-the-home (FTTH) systems—with fiber arriving directly to the customer premises—are also becoming widespread and allowing Gigabit connections that are faster than HFC networks.
Despite these upgrades in the transport mediums and standards, cable networks have experienced surprisingly few upgrades in their architectures. They still rely on centralized architectures in which the network operator’s headend performs almost all functionalities of both the physical layer (PHY) and medium access control layer (MAC). This means that the headend must modulate and demodulate data, convert between analog and digital, perform error corrections, provide cable modem termination system (CMTS) services, and do some resource allocation and flow control.
However, as traffic demands grow, cable providers need to deliver more and more bandwidth to their optical nodes and customer premises. The headend equipment is getting more congested, consuming more power, and running out of ports to handle more fiber connections. This solution centralized on the headend is struggling to scale up with increased demand. As it often happens in the telecom sector, operators need to figure out ways to deliver more bandwidth to more customers without spending significantly more money.
The multiple benefits of distributed access architectures
These issues are the reason why cable providers are moving into distributed access architectures (DAA) that can spread functionalities across access network nodes and reduce the port congestion and equipment required at the headend. Remote PHY has become increasingly popular among providers because it separates the PHY layer from the traditional cable headend and pushes its functions (such as modulation or digital-analog conversion) into the optical fiber access nodes of the network.
This shift can enhance the performance, capacity, and reliability of fixed access networks by using more digital transmission. It also reduces the complexity and power consumption of the headend, which previously translated into higher costs due to the required cooling.
Furthermore, separating PHY and MAC layers makes it easier to virtualize headends and their network functions, which significantly cuts expenses due to the use of commercial-off-the-shelf (COTS) equipment compared to more specialized equipment. Virtualization also allows deploying new services and applications more quickly to users and migrating workloads to optimize power consumption and reduce energy costs. On top of that, Remote PHY achieves all of these benefits while keeping the existing HFC infrastructure!
Distributing digital-analog conversion
One of the most significant upgrades provided by Remote PHY networks is digital transmission deeper into the access network. In Remote PHY, data and video signals are kept in a digital format beyond the core headend, all the way into the upgraded optical node, where the signal is then converted into analog RF format. The fiber links between the headend and the access node that were previously analog will become digital fiber connections over Ethernet.
Since digital signals are more noise-tolerant than analog signals, the network benefits from this increased digital transmission length. Analog and radiofrequency signals now travel smaller distances to reach customer premises, so the signal accumulates less noise and boosts its signal-to-noise ratio. This improvement potentially allows the delivery of higher bandwidth signals to customers, including an increase in upstream bandwidth. Furthermore, the reliability of the link between the headend and the new optical node increases due to the greater robustness of digital links. These advances in reliability and performance make digital optics more affordable to buy and maintain than analog optics, reducing the costs for the network operators.
Let’s provide a very simplified example of how it all comes together. A network operator wants to increase their bandwidth and serve more customers, but their traditional centralized headend is already crowded with eight analog optical fiber links of 1Gbps each. There is no room to upgrade.
By installing Remote PHY technology in both the headend and the node, those analog links can be replaced by higher-capacity 10G digital links. The increased capacity at the headend allows for more optical node splits, while the new digital-to-analog conversion capability of the nodes allows them to care of more coaxial splits, all to serve new areas and customers.
Using DWDM in Remote PHY
The tremendous progress in electronic and photonic integration made Dense Wavelength Division Multiplex (DWDM) technology affordable and available to access networks, and this technology is quickly becoming a workhorse in this network domain. The availability of affordable DWDM transceivers made the deployment of Remote PHY even more powerful.
With Remote PHY improving the capacity of the headend, cable access networks had more bandwidth to serve more customers. However, some ways of using that bandwidth are more efficient than others. Operators can do extra node splits for customers by using their dark fibers and more grey transceivers, but that solution doesn’t scale in so cost-effectively due to the installation and maintenance of a new fiber link. Another option is time division multiplexing (TDM), which multiplexes the data of different node channels into specific time slots. This solution allows operators to carry different node channels in a single fiber but has speed, latency, and security trade-offs. A single time-multiplexed channel cannot transmit at the same speed and latency as a dedicated channel, and the data of all node channels are in the same multiplexed optical link, so the nodes and their customers can’t have fully secure channels to themselves.
DWDM solutions, on the other hand, can avoid the speed and security trade-offs by multiplexing extra channels into different wavelengths of light. Instead of several TDM channels “splitting” the 10G bandwidth among themselves, the DWDM channels can each transmit at 10G. And since each WDM channel has its own wavelength, the channels are transmitted independently from each other, allowing users to have secure channels.
Without sharing an optical link as in TDM, DWDM channels can also provide bidirectional communication (upstream and downstream) with less electronic processing than TDM channels. This feature is particularly beneficial for the modern Internet consumption patterns described earlier in the article.
Let’s go back to our previous example of the upgraded headend with 10G digital fiber links. Thanks to DWDM technology, a single 10G port on this headend can support additional optical nodes in the network more cost-effectively than ever. Let’s say a new apartment complex was built, and the network operator needs to deploy a new node to service this new building. In the past, this deployment would have required lighting up a dark fiber and setting up an extra fiber link or using TDM technology with lower data rates, latency, and security. With DWDM, the new node can simply be carried through a different wavelength channel in the already existing fiber link. And as we will describe in our next article, autotuneability in DWDM transceivers makes their setup and maintenance even more affordable.
Cable networks need to serve more customers than ever with more symmetric upstream and downstream capacity, and they need to achieve this without changing their existing fiber and coaxial infrastructure. These goals become possible with the onset of Remote PHY and more accessible DWDM transceivers. By separating the MAC and PHY layer, Remote PHY reduces the load on the cable headend and allows for more virtualization of network functions, making it easier and more affordable to upgrade and manage the network. Meanwhile, DWDM enables connections from the headend to the Remote PHY nodes that serve tens of customers with a single fiber.