Tag Archives: DOCSIS 3.1

Can Cable Networks Deliver a Gigabit?

coax cablesTime Warner Cable recently promised the Los Angeles City Council that they could bring gigabit service to the city by 2016. This raises the question – can today’s cable networks deliver a gigabit?

The short answer is yes, they are soon going to be able to do that, but with a whole list of caveats. So let me look at the various issues involved:

  • DOCSIS 3.1: First, a cable company has to upgrade to DOCSIS 3.1. This is the latest technology from CableLabs that lets cable companies bond multiple channels together in a cable system to be able to deliver faster data speeds. This technology is just now hitting the market and so by next year cable companies are going to be able to have this implemented and tested.
  • Spare Channels: To get gigabit speeds, a cable system is going to need at least 20 empty channels on their network. Cable companies for years have been making digital upgrades in order to cram more channels into the existing channel slots. But they also have continued demands to carry more channels which then eats up channel slots. Further, they are looking at possibly having to carry some channels of 4K programming, which is a huge bandwidth eater. For networks without many spare channels it can be quite costly to free up this much empty space on the network. But many networks will have this many channels available now or in the near future.
  • New Cable Modems: DOCSIS 3.1 requires a new, and relatively expensive cable modem. Because of this a cable company is going to want to keep existing data customers where they are on the system and use the new swath of bandwidth selectively for the new gigabit customers.
  • Guaranteed versus Best Effort: If a cable company wants to guarantee gigabit speeds then they are not going to be able to have too many gigabit customers at a given node. This means that as the number of gigabit customers grows they will have to ‘split’ nodes, which often means building more fiber to feed the nodes plus an electronics upgrade. In systems with large nodes this might be the most expensive part of the upgrade to gigabit. The alternative to this is to have a best-effort product that only is capable of a gigabit at 3:00 in the morning when the network has no other traffic.
  • Bandwidth to the Nodes: Not all cable companies are going to have enough existing bandwidth between the headend and the nodes to incorporate an additional gigabit of data. That will mean an upgrade of the node transport electronics.

So the answer is that Time Warner will be capable of delivering a gigabit next year as long as they upgrade to DOCSIS 3.1, have enough spare channels, and as long as they don’t sell too many gigabit customers and end up needing massive node upgrades.

And that is probably the key point about cable networks and gigabit. Cable networks were designed to provide shared data among many homes at the same time. This is why cable networks have been infamous for slowing down at peak demand times when the number of homes using data is high. And that’s why they have always sold their speeds as ‘up to’ a listed number. It’s incredibly hard for them to guarantee a speed.

When you contrast this to fiber, it’s relatively easy for somebody like Google to guarantee a gigabit (or any other speed). Their fiber networks share data among a relatively small number of households and they are able to engineer to be able to meet the peak speeds.

Cable companies will certainly be able to deliver a gigabit speed. But I find it unlikely for a while that they are going to price it at $70 like Google or that they are going to try to push it to very many homes. There are very few, if any, cable networks that are ready to upgrade all or even most of their customers to gigabit speeds. There are too many chokepoints in their networks that can not handle that much bandwidth.

But as long as a cable network meets the base criteria I discussed they can sell some gigabit without too much strain. Expect them to price gigabit bandwidth high enough that they don’t get more than 5%, or some similar penetration of customers on the high bandwidth product. There are other network changes coming that will make this easier. I just talked last week about a new technology that will move the CMTS to the nodes, something that will make it easier to offer large bandwidth. This also gets easier as cable systems move closer to offering IPTV, or at least to finding ways to be more efficient with television bandwidth.

Finally, there is always the Comcast solution. Comcast today is selling a 2 gigabit connection that is delivered over fiber. It’s priced at $300 per month and is only available to customers who live very close to an existing Comcast fiber. Having this product allows Comcast to advertise as a gigabit company, even though this falls into the category of ‘press release’ product rather than something that very many homes will ever decide to buy. We’ll have to wait and see if Time Warner is going to make gigabit affordable and widely available. I’m sure that is what the Los Angeles City Council thinks they heard, but I seriously doubt that is what Time Warner meant.

The Gigasphere

cheetah-993774If you haven’t already heard it, you will soon be hearing the term ‘gigasphere’. This is the marketing term that the large cable companies are adopting to describe their upward path towards having faster data speeds on their cable systems. The phrase is obviously meant as a marketing counter to the commonly used term of gigabit fiber.

The gigasphere term is being promoted by the National Cable Television Association (NTCA) as the way to describe the new DOCSIS 3.1 technology. This is a technology that can theoretically support cable modem speeds up to 10 Gbps download and 1 Gbps upload.

The large cable companies are all starting to feel consumer pressure from fiber, even in markets where fiber is not readily available. Google and other fiber providers have excited the public with the idea of gigabit speeds and I am sure cable companies are being asked about this frequently.

Right now the term gigasphere is largely marketing hype. If you have fiber to your home or business, then with the right electronics you can get gigabit speeds. But cable systems have a long way to go before they can offer gigabit speeds over coaxial cable. There is already talk of cable companies offering gigabit products, such as the recent announcements from Comcast. But these speeds are not being achieved using coaxial cable and DOCSIS 3.1, they are using fiber – something Comcast doesn’t highlight in their marketing.

With enough upgrades and money, the cable systems can eventually achieve gigabit speeds on their coaxial networks. But for now their speeds are significantly less than that. A cable company faces a long and complicated path to be able to offer gigabit speeds over coaxial cable. Their biggest hurdle is that the bandwidth on their cable systems is mostly used by TV channels, and only empty channel spaces can be used for data. DOCSIS 3.1 allows a cable system to join together the spare channels on their network into one larger data pipe.

In order to get to gigabit speeds a cable company has to convert all of the channels on its network to digital, something most of them have already done. But further, they are going to need to treat them the same as TV on the web – transmitting them as raw data instead of as individual channels. Cable systems today use a broadcast technology, meaning they send all of the channels to customers at the same time. But if they convert to IPTV they can send each home just the channels they want to watch, which would massively condense the system bandwidth needed for television.

But this conversion is going to be costly and the equipment to do it is not yet readily available. CableLabs is working on this technology and it ought to be on the market in a few years. But that change alone is not the whole price of conversion. An IPTV system will require all new settop boxes, and in many systems a major reworking of the power taps and other components of the outside cable network. I don’t see many cable companies rushing towards this expensive conversion unless they are in a market where it is competitively necessary.

So for now, the gigasphere is mostly a marketing phrase. But it’s one that you are going to start hearing all of the time in relation to cable system data capabilities. This will obviously confuse the public who will assume that gigasphere means that they will be able to buy gigabit speeds from their cable companies, when they almost certainly cannot.

It’s not like cable companies don’t have fairly fast data capabilities. Most urban systems today are already capable of speeds in excess of 200 Mbps download. And there are systems working to get to 500 Mbps, which is probably about as fast as you can go without converting to IPTV. But it seems the marketing folks in the industry are counting on the fact that customers won’t know the difference between the various flavors of fast and will be happy with their gigasphere products. And they are probably right. Where’s my 500 Mbps cable modem?

Path to Gigabit Cable Systems

UHDTV_resolution_chartI saw an article yesterday where Pat Essex, the president of Cox talked about offering gigabit cable modem speeds to customers later this year. This conflicts with what the Cox CTO told Fierce Cable, and who said that the company has a five-year plan to free up enough bandwidth on their systems to be able to fully implement such a conversion. I think the CTO has it right. Perhaps later this year Cox might find a way to offer a gigabit to a handful of customers as a press release opportunity.

The article estimated that it would take ‘hundreds of millions’ for Cox to make this upgrade. What sorts of changes will Cox have to make on their networks to have speeds this fast? They will have to do most or all of the following network upgrades:
• Convert everything to digital. One of the biggest upgrades that benefit a cable network is to get rid of all analog channels. Analog channels can transmit only one TV channel in a 6 MHz slot, but once converted to digital you can fit as many as twelve channels in the same slot. Cox is probably mostly digital now, but anywhere they are not would need to finish this conversion.
• Convert to all MPEG4. MPEG4 is a compression scheme and it is the most efficient way to squeeze a cable channel into the least amount of bandwidth. A lot of channels are still delivered in an older compression technology called MPEG2. The big cost of making this kind of conversion is that any settop boxes that still can’t receive MPEG4 would have to be replaced.
• Split Nodes. A node is the number of customers in a geographic area that share the network. This is important when talking about data speeds, because whatever speeds is delivered to the node is then shared with all of the customers on that node. Passive fiber-to-the-home networks have nodes that are no greater than 32 homes. Active Ethernet fiber networks effectively have a node size of one home. But cable systems normally have nodes around 200 – 250 homes. Cable companies have to build more fiber to get closer to homes to get the node sizes smaller and closer in size to what fiber can deliver.
• Upgrade to DOCSIS 3.1. DOCSIS is the acronym for the technology that operates the cable modem network. This new standard was just released in October 2013. The standard lays forth a technology that will support capacities of 10 Gigabits downstream and 1 Gigabit upstream using 4096 QAM, if all of the capacity is used for data. The DOCSIS 3.1 specification does away with 6 MHz and 8 MHz wide channel spacing and instead use smaller (20 kHz to 50 kHz wide) orthogonal frequency division multiplexing (OFDM) subcarriers; these smaller pieces of spectrum can then be bonded together to create one large data path that could end using multiple 200 MHz wide frequency paths.

The new standard is expected to be fully deployable by 2016, so I don’t quite understand the timing mentioned by Cox. The biggest issue with deploying DOCSIS 3.1 is that unless you change every cable modem to be DOCSIS 3.1 compatible you will need to have enough system bandwidth to operate both the new and old cable modems side-by-side using two sets of network bandwidth. There is no sensible logistical path to flash-cut from one to the other and so cable networks will have to operate two large swaths of data simultaneously.
• Upgrade the whole network. In order for a network to have enough bandwidth to make the conversion it will have to be at least 1,000 MHz, and possibly larger. Today many cable networks are smaller than this and there is a lot of work needed to make such an upgrade that includes such things as changing taps and repeaters throughout the network, and even in some cases replacing the coaxial cable and many of the drops.

These changes can all be made. Each of these are major upgrades and require a lot of changes in the network. There are parts of the cable plant that will need to be rebuilt. Fiber must be built. Settop boxes and cable modems all have to be swapped. There are major electronics upgrades needed.

But one thing that this upgrade path doesn’t consider is that there is still an ever-increasing need to add more channels to a cable system. Cable customers want more and more HD channels which are much larger than standard channels. And now we are looking at the threat of having to offer super HD channels using 4K which are many times larger than today’s HD channels. At the end of the day there is only so much bandwidth in a cable network. Cable network engineers spend all of their time making tradeoffs between programming and cable modem bandwidth and it’s a challenging job. Do I think Cox can eventually offer gigabit service to everybody and not just to a small number of people for the press releases? Yes, but they have a whole lot of work to do and a whole lot of money to spend first.