Archive for the ‘Opinion’ Category

Another National LTE Network Coming?

Sunday, March 28th, 2010
RSS Feed Subscribe to EtherNEWS Bookmark and Share
Another National LTE Network Coming?
There’s quite a lot of fourth-generation mobile network construction happening, and about to happen, in the U.S. market, but the wild card now is that an entirely-new Long Term Evolution network might be built using spectrum originally allocated for satellite networks.
Harbinger Capital, which recently merged with SkyTerra, proposes to build a fully integrated satellite-terrestrial network to serve North American mobile users, featuring national LTE facilities that would operate on a wholesale-only basis.
The Federal Communications Commission apparently has required that, as part of the Harbinger purchase of SkyTerra, the firm operate as a wholsaler, and also that AT&T and Verizon traffic cannot account for more than 25 percent of total traffic carried on the Harbinger network.
The planned network would launch before the third quarter of 2011 and cover nine million people, with trials set initially for Denver and Phoenix. The next milestone is that 100 million people have to be covered by the end of 2012, 145 million by the end of 2013 and at least 260 million people in the United States by the end of 2015. Harbinger told the FCC that all major markets will be installed by the end of the second quarter of 2013.
Before any of that could happen, though, Harbinger would have to find additional investors willing to provide $5 billion worth of investment capital.
Analyst Chris King at Stifel Nicolaus estimates that Verizon’s LTE network will cost about $5 billion to deploy. Clearwire has also spent billions on its network, with analyst estimates ranging from $3 billion to about $6 billion. There is no particular reason to think the ubiquitous terrestrial network Harbinger expects to build would cost less.
http://tmfassociates.com/blog/2010/02/18/what-does-harbinger-do-next/

There’s quite a lot of fourth-generation mobile network construction happening, and about to happen, in the U.S. market, but the wild card now is that an entirely-new Long Term Evolution network might be built using spectrum originally allocated for satellite networks.

Harbinger Capital, which recently merged with SkyTerra, proposes to build a fully integrated satellite-terrestrial network to serve North American mobile users, featuring national LTE facilities that would operate on a wholesale-only basis.

The Federal Communications Commission apparently has required that, as part of the Harbinger purchase of SkyTerra, the firm operate as a wholsaler, and also that AT&T and Verizon traffic cannot account for more than 25 percent of total traffic carried on the Harbinger network.

The planned network would launch before the third quarter of 2011 and cover nine million people, with trials set initially for Denver and Phoenix. The next milestone is that 100 million people have to be covered by the end of 2012, 145 million by the end of 2013 and at least 260 million people in the United States by the end of 2015. Harbinger told the FCC that all major markets will be installed by the end of the second quarter of 2013.

Before any of that could happen, though, Harbinger would have to find additional investors willing to provide $5 billion worth of investment capital.

Analyst Chris King at Stifel Nicolaus estimates that Verizon’s LTE network will cost about $5 billion to deploy. Clearwire has also spent billions on its network, with analyst estimates ranging from $3 billion to about $6 billion. There is no particular reason to think the ubiquitous terrestrial network Harbinger expects to build would cost less.

http://tmfassociates.com/blog/2010/02/18/what-does-harbinger-do-next/


RSS Feed Subscribe to EtherNEWS Bookmark and Share

Sprint Accelerates 4G Build

Tuesday, March 23rd, 2010
RSS Feed Subscribe to EtherNEWS Bookmark and Share
Sprint has accelerated its previously-announced timetable for adding new 4G cities in 2010, adding Los Angeles and Miami to the list of cities it will activate this year.
Newly announced markets that will see 4G in 2010 are Cincinnati, Cleveland, Los Angeles, Miami, Pittsburgh, Salt Lake City and St. Louis.
Earlier this year Sprint announced that it planned to launch 4G in Boston, Denver, Kansas City, Houston, Minneapolis, New York, San Francisco and Washington, D.C., in 2010.
Sprint argues that its 4G WiMAX network operates up to 10 times faster than a 3G network, based on typical download speeds of 600 kbps for a 3G network compared to a 4G network’s 6 Mbps download speed.
Recent studies suggest typical performance of a 3G network ranges from  600 kbps to 1.7 Mbps while 4G average speeds run between 3 Mbps and 6 Mbps.
Separately, Verizon Wireless has accelerated its own timetable for 4G deployment, aiming to activate its Long Term Evolution in at least 25 cities, potentially reaching 100 million people, by the end of 2010.
That would be a stunningly quick deployment of a new nework, given the fact that Sprint has been building out its 4G network for a couple of years.
AT&T plans to start activating its 4G network in 2012.
By some estimates, the number of U.S. mobile tower sites will grow from about 228,000 (2008 base) to as many as 300,000 as a result of all the activity.
In 2008 the U.S. had 228,000 cell sites feeding a backhaul network. These sites were each served by between 5 Mbps and 10 Mbps of backhaul capacity on average. This added up to mobile operators in the U.S. spending $6.2 billion on backhaul services. Globally, service providers spent more than $24 billion in 2008 on backhaul services.
In a case study Yankee Group conducted with one of the tier one mobile operators in 2009,  more than 50 percent of data traffic was coming from desktop or laptop computers using aircards or dongles.
Approximately 24 percent was coming from screen phones and a little more than 20 percent was coming from smartphones (iPhones, BlackBerries). If the traffic trends continue, by 2012 more than 55 percent of the traffic will be coming from the smartphone “advanced OS” category, i.e., more than half of the mobile data traffic will be generated by truly mobile, instead of nomadic, users sending large amounts of data, voice or video traffic.
The backhaul implications are fairly simple, in terms of backhaul bandwidth requirements. A 2G voice network can get by with two T1 links. A 2.5 G EDGE network requires 4 T1s. A 3G HSPA network requires at least 10 T1s. Coming 4G networks might need as much as 10 times that much bandwidth.
By 2012 Yankee Group expects more than 300,000 cell sites in the United States, each supporting between 50 Mbps and 100 Mbps in backhaul capacity. In general, leasing a T1 costs about $300 per month, per mile. Seven T1s, adding up to a backhaul capacity of just more than 10 Mbps, will cost a U.S.-based provider about $2,100 per month per mile.
If we were to keep throwing T1s at the problem, this would result in a backhaul bill of $82 billion by 2012 and the monthly average cost per site would be about $23,000 compared to today’s average of $2,100 per mile.  That’s not an option.
Performance requirements also are growing. Network architects as asking for one-way jitter of 1 to 3 milliseconds and one-way latency of 3 to 5 ms.

Sprint has accelerated its previously-announced timetable for adding new 4G cities in 2010, adding Los Angeles and Miami to the list of cities it will activate this year.

Newly announced markets that will see 4G in 2010 are Cincinnati, Cleveland, Los Angeles, Miami, Pittsburgh, Salt Lake City and St. Louis.

Earlier this year Sprint announced that it planned to launch 4G in Boston, Denver, Kansas City, Houston, Minneapolis, New York, San Francisco and Washington, D.C., in 2010.

Sprint argues that its 4G WiMAX network operates up to 10 times faster than a 3G network, based on typical download speeds of 600 kbps for a 3G network compared to a 4G network’s 6 Mbps download speed.

Recent studies suggest typical performance of a 3G network ranges from  600 kbps to 1.7 Mbps while 4G average speeds run between 3 Mbps and 6 Mbps.

Separately, Verizon Wireless has accelerated its own timetable for 4G deployment, aiming to activate its Long Term Evolution in at least 25 cities, potentially reaching 100 million people, by the end of 2010.

That would be a stunningly quick deployment of a new nework, given the fact that Sprint has been building out its 4G network for a couple of years.

AT&T plans to start activating its 4G network in 2012.

By some estimates, the number of U.S. mobile tower sites will grow from about 228,000 (2008 base) to as many as 300,000 as a result of all the activity.

In 2008 the U.S. had 228,000 cell sites feeding a backhaul network. These sites were each served by between 5 Mbps and 10 Mbps of backhaul capacity on average. This added up to mobile operators in the U.S. spending $6.2 billion on backhaul services. Globally, service providers spent more than $24 billion in 2008 on backhaul services.

In a case study Yankee Group conducted with one of the tier one mobile operators in 2009,  more than 50 percent of data traffic was coming from desktop or laptop computers using aircards or dongles.

Approximately 24 percent was coming from screen phones and a little more than 20 percent was coming from smartphones (iPhones, BlackBerries). If the traffic trends continue, by 2012 more than 55 percent of the traffic will be coming from the smartphone “advanced OS” category, i.e., more than half of the mobile data traffic will be generated by truly mobile, instead of nomadic, users sending large amounts of data, voice or video traffic.

The backhaul implications are fairly simple, in terms of backhaul bandwidth requirements. A 2G voice network can get by with two T1 links. A 2.5 G EDGE network requires 4 T1s. A 3G HSPA network requires at least 10 T1s. Coming 4G networks might need as much as 10 times that much bandwidth.

By 2012 Yankee Group expects more than 300,000 cell sites in the United States, each supporting between 50 Mbps and 100 Mbps in backhaul capacity. In general, leasing a T1 costs about $300 per month, per mile. Seven T1s, adding up to a backhaul capacity of just more than 10 Mbps, will cost a U.S.-based provider about $2,100 per month per mile.

If we were to keep throwing T1s at the problem, this would result in a backhaul bill of $82 billion by 2012 and the monthly average cost per site would be about $23,000 compared to today’s average of $2,100 per mile.  That’s not an option.

Performance requirements also are growing. Network architects as asking for one-way jitter of 1 to 3 milliseconds and one-way latency of 3 to 5 ms.


RSS Feed Subscribe to EtherNEWS Bookmark and Share

LTE: Cleaning Up the Cell Site

Thursday, March 18th, 2010
RSS Feed Subscribe to EtherNEWS Bookmark and Share

I’ve winced every time I’ve heard the time “convergence” over the past several years.  Convergence has always been a marketing word for “mess”, where multiple technologies co-exist and intermingle in ways that increase Tylenol consumption and slow down true telecom innovation.

Today’s wireless networks, including the current 3G deployments, still rely on this dirty word with “converged” cell site connections – duplicating provisioning of both TDM private lines for voice, timing and signaling and Ethernet for data.

There are many good reasons why.  Until recently, Ethernet hasn’t proven as reliable as required to carry conversations, and T1s are already in place at cell sites where sync is required to keep radios locked on a common frequency and phase for roaming hand-offs.  Necessary for now, but inefficient (and despised?) all the same.

LTE offers a chance to do some spring cleaning at the cell site, simplifying backhaul connectivity with a single, performance-assured Carrier Ethernet link.  Simplicity looks like it’s making its way back into telecom, right?

Cell Site Evolution

Unfortunately, we may be gaining capacity and working with less equipment, but the clutter has simply moved from physical equipment to the way it’s configured.  No one ever had their Mom tell them “clean up your virtual room”, but this is where the mess goes in LTE backhaul networks – into the provisioning, monitoring and performance assurance required to compensate for having all your data running through a single pipe.

Making a clean break to a fully packet-based architecture, voice calls will be VoIP, carried over the same all-IP infrastructure carrying the latest generation of multicast and on-demand web-based video, Internet, messaging and email traffic.  With each vying for available bandwidth, maintaining per-application Quality of Service (QoS) is critical – the best-effort, limited-bandwidth backhaul connections serving legacy data services will not suffice.

4G services require ultra-low latency, jitter, and packet loss with assured throughput and availability.  Latency can spell the end of conversations if signaling delays interrupt session continuity when roaming between cells.  Jitter and packet loss can make audio inaudible and video unwatchable.  Insufficient backhaul bandwidth leads to congestion, increasing latency, packet loss and packet retransmission resulting in degraded QoS.  Availability is the most basic of all –  if the network goes down, so do your customers – outages and lack of bandwidth are the primary drivers for customer churn.

So while Ethernet to the cell site is certainly the future (and looks clean from the perspective of slick, stylized network diagrams), it doesn’t come without its own baggage.  Best to be prepared for the surprises that are popping up in field trials – keep an eye on QoS, monitor it proactively or you may just discover the monsters in the closet.

CTIA next week will be a good place to explore these trends – check out the backhaul pavilion, get trained and attend the talks going on to learn all about what we’re facing.


RSS Feed Subscribe to EtherNEWS Bookmark and Share