Friday, December 26, 2008

Upgrading TCP for the Heavy User

Bob Briscoe from BT wrote an article that appeared in the December 2008 IEEE Spectrum that discusses proposed changes to TCP that would help the Internet cope better with so-called bandwidth hogs. It basically does this by providing a mechanism that allows network routers to discard packets from the heavy users before discarding packets from casual users.

Heavy users will declare their packets to be eligible for discard and a feedback mechanism will be created that will inform core routers of congestion that is occurring in the network so that they can start discarding packets.

This approach should provide better performance to the casual user than they experience today without significantly degrading performance for the heavy user.

You can get more details from Briscoe's web site.

It looks to me that this is a much better approach than the limits being proposed by carriers today. It also seems to resolve the Net Neutrality argument by redefining what fairness means and equalizing the performance of the network to a broad set of users rather than equalizing access to bandwidth.

I question how much these techniques will help in an IP network dominated by TV traffic. TV traffic wants to use a lot of bandwidth and have a high QoS at the same time. An HDTV viewer will not like it if discarding packets from his or her video stream causes visual or audio defects.

Altair 4G OFDM Chips

I recently spoke with Altair, which is an Israeli company that is building OFDM chips to support both WiMAX and LTE. It is focusing on 4G technologies by building handset chips. The company currently has chips for WiMAX and for XGP, which is an OFDM technology deployed by Willcom in Japan.

Altair plans to introduce an LTE chip in the middle of 2009 and a multimode chip by the end of 2009 that will support WiMAX, XGP, and LTE. This will enable support of all three technologies in a single hand held device. The company is also working with a 2G/3G supplier to offer the ability to add WiMAX, XGP, or LTE to a 2G/3G device using the Altair chip as a coprocessor.

This is interesting because it shows that progress is being made on chips that will facilitate the integration of 2G, 3G, and 4G WiMAX and LTE networks. This will help with the evolution of existing networks to WiMAX or LTE. It may also facilitate roaming between WiMAX and GSM, WCDMA, and LTE networks, which may keep WiMAX from becoming the dead end that CDMA has become.

Sunday, December 21, 2008

FTTH Green Postitive in U.S. in Six Years

The FTTH Council has published a report by PricewaterhouseCoopers that shows that the environmental impact of the deployment of a typical FTTH network in the United States will be positive within less than 6 years in average considering only benefits associated to telecommuting.

FTTH is much greener than DSL or VDSL in the U.S. Both of these technologies require powered remotes, which significantly raises their energy requirements. This is a much more significant problem for VDSL than DSL.

Thursday, December 18, 2008

Creating a Competitive Fiber Market

Last week I posted an article about how Swisscom will support competitors in its FTTH deployment. Swisscom will deploy four fibers to each home and reserve one for itself and make the other three available to competitors. It pointed out that the cost of providing four fibers is only slightly higher than providing a single fiber.

This significantly helps Swisscom's business case because it will generate revenue from every FTTH home that it deploys. This will provide it with a return on the investment that it is making on deploying these fibers. Swisscom will also accept investment or trade fiber connections with its competitors that deploy their own fiber networks.

A similar facility sharing approach has been adopted in France and is likely to be adopted broadly in Europe.

Japan is taking another approach. NTT is deploying all of the fiber; however NTT East and West (the two incumbents in Japan) can offer only the fiber connection. The subscriber gets the Internet service from an ISP. NTT operates two of these ISPs itself (OCN and Plala), but has to treat other ISPs on an equal basis. KDI and Softbank are major ISPs that use NTT fibers along with a number of smaller ISPs.

The U.S. FCC has given up on creating a competitive environment with FCC. It traded exclusive use of the fiber for the commitment to deploy fiber services. I think this was a bad bargain. Both Europe and Japan have shown that better approaches are available.

Wednesday, December 10, 2008

ATT's 2020 Strategies

AT&T gave a presentation this week where it made some interesting comments about how it plans to address important issues for getting ready for 2020.

The company talked about how network usage will change. It expects that 70 percent of its voice minutes will come from mobile services by 2013 and that data usage will grow by 50 percent per year between now and 2013. Its charts indicate that at least half of its voice traffic came from mobile services in 2007. It expects that voice will account for less than 5 percent of the traffic on its networks by 2013. Its charts indicate that it expects video to account for significantly more traffic than data. Over all its expects its backbone traffic to grow 50 percent annually and access traffic to grow 30 percent annually.

The company also discussed its consolidation into what it calls "one company". It expects to end up with one business services organization, one consumer services organization, and one network and services organization. It is doing this to increase its efficiency and to deliver new integrated/converged services. It stated that its U-verse IPTV service is the template for what it plans to achieve.

ATT plans to focus several areas for new services - femto cells, VoIP, VPNs, along with hosting and CDN services.

ATT is moving along the directions that we forecast in our report Telecom 2020: Transformation Strategies. Mobile is becoming the primary delivery vehicle for many services. This presentation confirms our predictions of the major effects that will occur in the organization of carriers.

Swisscom to Deploy FTTH

Swisscom is building a fiber network for residential customers and SMEs. Work has already started in Zurich, Basel and Geneva, with the aim of connecting 100,000 households with fiber optic by the end of 2009. The plan is then to further extend the network to include residential premises in the cities of St. Gallen, Berne, Fribourg and Lausanne.

The first offerings for residential customers and SMEs will be launched in the first half of 2009. Over the next six years Swisscom is planning to invest $US2.3 billion in its fiber expansion.

To enable other companies to expand their own fiber infrastructure after the construction work has started, Swisscom will be laying several fibers per household in all areas. One fiber will be used by Swisscom, while the others will be made available to other companies. The multi-fibre model will prevent the creation of a new network monopoly in Switzerland and also meet competitors' requirements for full access to the local loop (copper pairs) as stipulated by the Telecommunications Act.

Swisscom offers other companies interested in collaborating on the construction and operation of the fiber network four different cooperation models:

  • Construction partnership for other companies with their own ducts, such as electrical utilities or cable network providers. One of the partners takes on responsibility for building the fiber network in a region. Several fibres are laid, and when the network is completed each of the other partners is assigned one fibre. If all the partners network regions which are the same size and are to be shared, no compensatory payment is required.
  • Investment partnership for companies without their own cable ducts. Network expansion is jointly financed by all the partners. One partner builds the entire network and grants the investor usage rights to the fibers laid.
  • Rental of individual fibers for companies who do not wish to invest in network expansion but want to decide themselves on the preferred technical level for controlling the fiber network.
  • Leasing of transmission services similar to established practices.

Swisscom presented its reseller offerings to all Internet service providers at the beginning of November. In the initial phase these offerings will cover bandwidths of 30 to 50 Mbps for download and up to 10 Mbps for upload.

During the pilot phase, which starts at the beginning of March and will become a commercial service in autumn 2009, the offerings will focus on the areas in Zurich, Basel and Geneva which are already equipped with fiber cables. Internet service providers, including VTX, green, netstream and init7, are free to design their own end customer and reseller offerings.

This is an interesting model for handling competition over fiber. It will produce a lot more competition than the U.S. approach that gives the carriers exclusive use of their fiber plants.

More than 1,000 HSPA Devices Available

The Global mobile Suppliers Association (GSA) stated that there are now 1,095 HSPA devices available with 692 announced since October 2007. The following types of devices are available:
  • 451 mobile phones
  • 16 ultra mobile PCs
  • 218 notebooks
  • 154 USB modems
  • 143 PC data cards
  • 101 wireless routers/gateways
  • 8 Personal Media Players
  • 4 cameras
It is interesting that less than half of the HSPA devices are mobile phones. What is interesting is to see that a dozen consumer devices have become available. The number of HSPA (or 4G) enabled wireless devices will grow to become a significant segment if the operators create usage plans that are attractive for these devices.

Sunday, December 7, 2008

Is Google Paying its Fair Share?

The blog Internet Evolution published and article discussing comments by an analyst who says that Google is not paying its fair share of the Internet infrastructure.

This is an interesting issue. The core Internet network has a robust business model. As the ISPs and ASPs generate more traffic, for example, by providing more video and TV traffic, they need to increase the bandwidth that connects to them to the Internet backbone. They pay for every additional bit of bandwidth that they use. This means that the carriers revenues increase as the traffic increases, which is and has been a sustainable business model.

I think that the question that is raised in this article is if Google is using more bandwidth per dollar spent on backbone capacity than other companies. This may be assuming that Google distributes more videos than others.

The more interesting question is in the access network. The broadband access providers are stuck in the conundrum that customers expect more bandwidth over time but do not expect to pay more for it. This is most clearly demonstrated in countries in such as Japan and France where the price of a 100 Mbps FTTH connection is the same as a 10 Mbps DSL connection. Carriers in these countries are making huge investments to deploy fiber and are not getting significant revenue increases to support it.

I have wondered if Google and the other ASPs and ISPs should pay more for their backbone connections to support the deployment of FTTH. After all, these companies get tremendous benefit from the deployment of higher speed access networks. It only seems fair.

However, there are significant problems with this approach. One is that the combined profits of Google and the other ASPs and ISPs are not nearly enough to cover the cost of deploying fiber at a reasonable rate.

The second problem is that paying additional amounts to the backbone provider will not necessarily lead to a fair distribution of these funds to the access network providers. For example, if Google pays the premium to ATT, how does Verizon get support in its territory. Well, Google will probably by backbone services from Verizon also, so that there will probably be a reasonable allocation. But then, what about Surewest and the thousand or so small U.S. telcos that do not offer backbone services?

This could be handled by an extension of the Universal Services Fund, but then what about the broadband operators in other countries? Do we need to add in a UN Universal Fund charge as well?

None of this seems practical, which leaves the broadband providers without any way to monetize their investments in fiber. This will certainly discourage investment and delay the availability of fiber services That is too bad.

Is Wireless Important to Cable Companies?

A Light Reading article discusses comments by a financial analyst who believes that wireless is not an important element to add to cable bundles. His point is that bundles require discounts, which reduces revenues.

I think he is probably right about adding wireless to cable bundles. Wireless probably will have less effect on consumer decisions than the other elements of these bundles.

However, in the long run as carriers offer wireless and broadband wireline services based on a single, integrated IMS infrastructure to deliver VoIP services there will be tremendous opportunities to integrate wireless and wireline services into a single offering. The cable companies will have to be in a position to offer similar integrated wireless and wireline services to stay in the game.

The moves that the cable companies are making with Clearwire today are a good step to prepare for that day.

ATT's Cuts

Light Reading published and article with various opinions on ATT's announcement that it would cut 12,000 people and reduce its CapEx budget. The gist of it is that ATT will not make cuts in wireless or broadband but will cut its wireline operation. The article ends with a quote from one analyst who believes that cutting spending on its POTS network will be a major mistake for ATT.

What ATT is reportedly doing is consistent with how networks will evolve during the next decade. Wireless and broadband will become more important and POTS services will fade away. Personally, I think ATT is making the correct choices.