In Part 1, we discussed the user experience MSOs will be expected to provide in the future. CONTENT will remain important, but the other two winners in this future environment are CONNECTIVITY and CLOUD. Jointly these will reduce the complexity that exists in today´s networks and devices and unlock unfulfilled commercialization potential.
Connectivity and cloud
MSOs need to account for a wireless enabled world where there will be at least 10 devices per user:
- Video cameras
- A multitude of sensors
Many of these devices will source not only data but video content. We estimate that each such device will require about 10x the bandwidth of current devices, so we are talking about the need for 100x more access capacity than currently offered by wireless networks (Figure 1).
This presents a huge challenge – but also opportunity – for providers of wireless infrastructure and the associated wireline networks used to connect them. But delivering this capacity with the right economics for sustainable growth is key.
MSOs can address this challenge with optimal economics by coupling CONNECTIVITY and CLOUD and changing their network architecture to a distributed "edge cloud" architecture. In essence, instead of backhauling all the traffic to a centralized cloud destination across the core network, traffic can be more efficiently served from distributed locations in the MSO edge network, creating a new distributed cloud within the MSO network.
Moreover, this edge cloud is not only ‘bandwidth efficient’, but also minimizes the latency of control and signalling traffic, allowing virtualization of key MSO network functions such as:
- Layer 2 and Layer 3 MAC and IP functions
- Wi-Fi® control functions
- Subscriber management and policy management functions
- CDN functionality and associated video processing functions
- Advanced IP Communications (such as IMS) service control functions
This can be thought of as a “virtualized hub node” comprised of virtualized CMTS/CCAP and video and voice delivery functions.
To understand why this change is happening now, it is important to recognize that there have been recent advances in information technology (IT) and networking/communications technologies that will drive this change. Specifically, IT and data center systems have pioneered virtualization technologies, which are increasingly being used by service providers to help economically scale their network functions (so-called network functions virtualization -- NFV). At the same time, IP/MPLS networking techniques are appearing in the data center in order to automate the connectivity of these virtual software instances using software-defined networking (SDN).
This combination of NFV and SDN will play a big part in the future of MSO networks. Current networks are built on closed systems with tightly integrated software and hardware, with long development, test, and integration times. Consequently, equipment remains in the network for a decade or more.
As shown in Figure 2, NFV effectively allows massive improvement in the performance and cost of networks for many functions. It does this by:
- separating the hardware and software deployment cycles
- moving to a uniform way of deploying and managing the software
- deploying it on a standard computing architecture with high performance
- running multiple software instances on a single server (by using virtualization)
This technique is so powerful that all the software functions of the MSO network in the future will be virtualized. In fact, the only functions that will not be virtualized are:
- those that perform advanced packet processing at very high speeds (100G and above, today) as these typically require specialized packet processors to achieve the optimum economics
- those that terminate the physical layer (Layer 1), as they typically require advanced signal processing which is more efficiently performed in digital signal processors (DSPs)
But we need a way to connect these virtualized software instances together to create the desired service, and also to connect this control software to the underlying physical hardware. And this is the role of SDN. Simply put, it does 3 things:
- It automates the networking of different software components
- It connects the software to the network hardware
- It exposes a programmable interface that allows different applications and services to optimize their deployment in and on the network
So, SDN and NFV effectively work hand-in-hand to create a new flexible, scalable, programmable and affordable network fabric. But how does this improve the physical layer (Layer 1) access bandwidth which is the lifeblood of the MSOs? To understand this, we have to think about physics.
Physical limits of today’s networks
For any transmission medium, physics tells us that there are only 3 parameters (Figure 3) that we have to consider in transmitting data or video:
- Spectrum -- how much spectrum is available
- Spectral efficiency -- how efficiently you use that spectrum from a given location
- Spatial efficiency -- how much you re-use the spectrum at different locations
It’s just simple physics of propagation of electromagnetic radiation (in this case, RF signals) down a transmission medium (cable) with a desired signal-to-noise ratio (signal quality). This is true for coax cable, or copper twisted pairs, or transmission through the air or over optical fiber. All of these are just examples of applying Maxwell’s equations combined with Shannon theory of communications.
Now thinking about coax cable specifically, we have a constrained amount of spectrum for DOCSIS® (due to the presence of digital video on the majority of the spectrum available), and we are already using advanced modulation with maximum spectral efficiency (QAM 256 and beyond). So the only choice is to increase the spatial efficiency by splitting fiber nodes to serve smaller and smaller service groups.
And this brings us back to the edge cloud concept described above, as the optimal architecture to economically achieve high capacity and dynamic scalability is a small node architecture with the IP and MAC (Layer 2/3) and subscriber management functionality virtualized and run in the cable edge cloud. This leaves only the L1 (physical layer) functionality in the node. This is the so-called ‘remote PHY’ architecture being contemplated in the industry, or the ‘Remote MAC/PHY’ architecture, where only the L3 and above functionality is virtualized.
This cable edge cloud + distributed remote node architecture can also be used to efficiently backhaul, manage and control the millions of (distributed) Wi-Fi access points that will also form a key part of the MSO future: to offer an anywhere, anytime, high capacity connectivity experience to users.
And, to top it off, this architecture also offers the ability to host 3rd party or Enterprise cloud services, by offering high performance (high bandwidth, low latency) cloud hosting services, and as such can provide an important source of new revenue for MSOs as well.
So it really is a "win-win-win" architecture for MSOs, users, and application providers.
Read part 3 of this series for my conclusions on how the MSO network must evolve to profitably satisfy future user demand.
To contact the author or request additional information, please send an email to firstname.lastname@example.org.
Subscribe now for monthly updates straight to your inbox