Template Tools
21st Century Triple Networks: Ubiquitous 4G. WiFi, & Wires
Written by Dave Burstein   
Wednesday, 26 January 2011 11:34
(Version 0.1. Improvements welcome) The best engineers on theCerberus by William Blake planet are coming to the same conclusion: a hybrid 4G/WiFi/landline network is the way to meet mobile  demand. Folks like John Donovan of AT&T and Masayoshi Son of Softbank in Japan had this vision around 2007-2008. As the iPhone/iPad/Android made the coming demand clear, networks planners around the world evolved similar strategies.  

  • 4G gives wide coverage but is limited in capacity.
  • WiFi actually provides far more capacity, because the range of perhaps 100 meters means the spectrum can be reused thousands of times in a major city. (China Mobile is putting 20,000 WiFi hotspots across Beijing.)  A network builder tells me "WiFi is a solution to off load 'portable' traffic where possible and rely on 3G/4G for 'mobile' traffic." Femtos and perhaps small cells will play a significant part.
  • Landlines effectively have 10x the capacity of a similar wireless network and are already ubiquitous from both telco and cable. A top engineer tells me "The general rule is the quicker you can get the byte of information onto a hard facility (copper, fiber) the cheaper it is to operate the network." Randall Stephenson of AT&T explains “You're always going to have to have a fixed line capability to offload this traffic.”
   

LTE networks are proving out remarkably well and the engineers building them are excited. Verizon is consistently getting the 5-12 megabits expected with low latency. Telia customers in Sweden are often seeing 15 and even 20 meg. The cost per bit is 60-90% lower than 3G and the capacity easily 10 times the older generation. (3.5G HSPA+ and similar are also doing well.) For now, there's enough LTE capacity Verizon is launching their network without caps and Clearwire Wimax is also uncapped. Millions of light and rural users will save money using LTE to replace DSL and cable.


    Verizon will begin switching to LTE advanced in 2013, but at some point and in some places will face limits. The first choke point is crowds: we're about to have the first Superbowl with 10,000 iPhones with video cameras in stands. AT&T put WiFi in Times Square before New Year's Eve and the results were good.


     So cell tower 3G/4G ideally is supplemented with local WiFi/femto. Cell towers cover large areas, allowing comprehensive coverage except for a few dead spots. They offer limited bandwidth over that entire area, with a network like Verizon's LTE offering perhaps 35 megabits to share. WiFi is much lower power, limiting range to a typical 100 meters or so, less with obstructions. Within that range, the capacity is high; 3x3 MIMO 802.11N can carry 100's of megabits in a small area. Locally, 802.11 uses spectrum more efficiently, incorporated a limited set of "spread-spectrum" type features.

      WiFi was in few phones two years ago because it ran down batteries too quickly and cost too much. Moore's Law now enables low power, low cost WiFi. The latest chips from RALINK/Trendchip, for example, cost less than $5. Off mode power consumption is 0.012 mw,  transmit power is 19dBm, and the chips are 5 to 7 mm square. Easily 3/4ths of the phones sold by a carrier like Verizon will soon have WiFi as do just about all tablets. As Qualcomm, Broadcom and others include WiFi on their primary cellphones chips it will become ubiquitous.

    Whether cell site or home WiFi, the data moves as quickly as possible to landlines. A neighborhood with 200 homes has typically a gigabit worth of copper for DSL. Cable is shared, but almost always has shared capacity. Once you get to the local exchange, there's fiber to the core with essentially unlimited capacity, usually inexpensive. Incumbents like AT&T recognize the landline network gives them a crucial edge for the profitable wireless business. AT&T five years ago declared "we are a wireless company," and held back on fiber investments. But as mobile data demand developed, even "a wireless company" discovered how crucial the landlines are.
    Deep thinkers, like MIT's David Reed, are looking at spread spectrum and other network architectures that overcome congestion problems up to far beyond the likely demand for decades. Reed's seminal article, with Dave Weinberger, The Myth of Interference, shows the way to think about even more advanced architectures. Well worth considering for future networks.
     Carriers are choosing different strategies to get from where they are today to triple networks. Vodafone, Europe's largest wireless company, is adding millions of DSL customers through unbundling and giving them femto+WiFi gateways. Sky in Britain is buying a WiFi network named "The Cloud." Free.fr enables WiFi on their millions of DSL connections and bought a wireless license. AT&T is putting WiFi hotspots from Times Square NY to San Francisco with expansion plans.  China Mobile is adding 1,000,000 hotspots.

(more to come. Thoughts very welcome.)