Tuesday, July 1, 2014

Backhaul Increasingly is a Strategic Matter for ISPs

Backhaul sometimes is a strategic advantage or key impediment for one or more service providers, as well as an important driver of operating cost. Backhaul often accounts for as much as 25 percent of total operating cost for a mobile service provider, for example.

As transmission networks become more dense, using small cell and carrier Wi-Fi architectures, for example, backhaul will be a major issue, mostly because the cost of backhaul has to scale to much-lower levels than has been the case for mobile and enterprise backhaul prices.

Where a traditional enterprise backhaul had substantial revenue generated by the link, a carrier Wi-Fi or small cell might have close to zero incremental revenue generated by the link.

The cost of backhaul also has been a key impediment for ISPs seeking to provide higher access speeds in regions distant from an Internet access point.

When the “Broadband Technology Opportunities Program” was launched in 2008 to promote high speed access advances in rural areas, you might have predicted that most of the money would be spent to create or augment access facilities.

Instead, middle mile backhaul facilities received significant funding. The reason was simple enough: in many rural areas, it is the backhaul to Internet points of presence that is the key impediment to faster end user Internet access.

You might argue that is the case in many parts of South Asia and Africa as well. There is little point in creating new access networks where backhaul is insufficient to support those access assets and potential customers.

Wi-Fi also now is an essential part of the backhaul strategy for most mobile service providers, allowing carriers to offload half or more of total Internet access traffic from the mobile network to the fixed network.

In some cases, up to 80 percent of mobile traffic is offloaded to Wi-Fi networks.

In similar fashion, deployment of mobile cell capacity likewise drives growth of demand for backhaul. And though much attention has been focused on the impact of new small cells and carrier Wi-Fi, standard macrocell deployments can be important as well.

“Over the past several years of experience, a fairly strong correlation between domestic carriers, aggregate CapEx and our level of organic growth in American Tower” can be seen, said Jim Taiclet, American Tower Corp. CEO. “For example, from 2010 to 2012, we saw aggregate spend on wireless CapEx of about $25 billion to $30 billion supporting our organic core growth rates in the range of seven percent to eight percent during those years.”

In 2013, when U.S. mobile service provider capital investment grew to nearly $35 billion a year, American Tower has seen revenue growth of nine percent.

Backhaul bandwidth demand scales in other ways beyond the number of tower sites and radios, though.

Between 2012 and 2013, average daily U.S. smartphone data consumption grew by almost 40 percent, while connected tablet usage increased by over 50 percent. And then there is mobile video, consumption of which might grow an order of magnitude between 2013 and 2018, according to Cisco projections.

Small cell deployments will have an impact, but American Tower presently generates 95 percent of its revenue from macrocell sites.

And one big question is how much incremental demand Sprint might drive, as it activates new 2.5 GHz capacity.

As an example, said Taiclet, Sprint would have to add 30,000 to 40,000 transmission locations to have 2.5 GHz coverage match the existing 1.9 GHz network footprint. That could possibly double the number of tower locations operated by Sprint.

All of those examples--Wi-Fi offload, small cell backhaul, existence of backhaul facilities in emerging markets, additional 2.5-GHz cell sites and BTOP funding--illustrate the roles backhaul often plays as a strategic matter for mobile service providers, not merely a tactical necessity.

No comments:

Costs of Creating Machine Learning Models is Up Sharply

With the caveat that we must be careful about making linear extrapolations into the future, training costs of state-of-the-art AI models hav...