THESE still are early days for the Internet, globally speaking. One billion people online; five billion to go.


The next billion to be connected are living in homes that are physically close to an Internet gateway. They await a solution to the famous “last mile” problem: extending affordable broadband service to each person’s doorstep.


Here in the United States, 27 percent of the population lacks access to the Internet, according to a study completed last year by the Pew Internet and American Life Project. Among those who do have access, about 30 percent still rely on slow dial-up connections. The last mile for households with no or slow connections may be provided by radio signals sent out by transmitters perched atop street lights, as hundreds of cities have rolled out municipal Wi-Fi networks, or are in the process of doing so.


The impulse behind these projects is noble. It’s a shame, however, that lots of street lamps and lots of dollars — a typical deployment in an urban setting will run $75,000 to $125,000 a square mile, just to install the equipment — do not really solve the last-mile problem.
If you’re sitting with your laptop at an outside cafe, you’ll be happy with the service. But if you happen to be at home, you realize that service to the doorstep is not enough: you still need to buy equipment to bolster the signal and solve the “last mile plus 10 more yards” problem — that is, getting coverage indoors.


Wi-Fi signals do not bend, and you usually can’t get much of a useful bounce from them, either. Because Wi-Fi uses unlicensed bands of the radio spectrum, by law it must rely on low-power transmitters, which reduce its ability to penetrate walls. Travel-round-the-world shortwave, this ain’t.


Trying to cover a broad area with Wi-Fi radio transmitters set atop street lights brings to mind a fad of the 1880s: attempts to light an entire town with a handful of arc lights on high towers. But overeager city boosters around the country soon discovered that shadows obscured large portions of their cities, and the lighting was not as useful as had been expected. Municipal Wi-Fi on streetlamps, another experiment with top-down delivery, may run a similarly short-lived — and needlessly expensive — course.


WiMax, which will be a high-power version of the tower approach, comes in two flavors: mobile, which has not yet been certified, and fixed, which is theoretically well suited for residential deployment. Unfortunately, it’s pricey. Peter Bell, a research analyst at TeleGeography Research in Washington, said fixed WiMax would not be able to compete against cable and DSL service: “It makes more economic sense in semirural areas that have no broadband coverage.”
An intriguingly inexpensive alternative has appeared: a Wi-Fi network that is not top-down but rather ground-level, peer-to-peer. It relies not on $3,500 radio transmitters perched on street lamps by professional installers but instead on $50 boxes that serve, depending upon population density, more than one household and can be installed by anyone with the ease of plugging in a toaster.


Meraki Networks, a 15-employee start-up in Mountain View, Calif., has been field-testing Wi-Fi boxes that offer the prospect of providing an extremely inexpensive solution to the “last 10 yards” problem. It does so with a radical inversion: rather than starting from outside the house and trying to send signals in, Meraki starts from the inside and sends signals out, to the neighbors.


Some of those neighbors will also have Meraki boxes that serve as repeaters, relaying the signal still farther to more neighbors. The company equips its boxes with software that maintains a “mesh network,” which dynamically reroutes signals as boxes are added or unplugged, and as environmental conditions that affect network performance fluctuate moment to moment.
At this time last year, two of Meraki’s co-founders — Sanjit Biswas and John Bicket — were still Ph.D. students at M.I.T., pursuing academic research on wireless mesh networks in the course of building Roofnet, an experimental network that covered about one-third of Cambridge, Mass., and offered residents free service.


Last year, Google invited Mr. Biswas to give a presentation about his experience providing wireless Internet service to low-income communities. At the time, Google was testing its first municipal Wi-Fi network in its hometown, Mountain View, Calif., using transmitters attached to street lamps.


After Mr. Biswas’s talk, a Google engineer told him that people using Google’s network said they could get online at home only by holding their laptops against a window. Mr. Biswas said he was not surprised. Using municipal Wi-Fi for residential coverage, he said, was “the equivalent of expecting street lamps to light everyone’s homes.”


Mr. Biswas and Mr. Bicket realized that their mesh-network gear designed for residential use could avoid that problem, and hasten the extension of Internet access worldwide. They founded Meraki, took a leave of absence from M.I.T. and, along with a third co-founder, Hans Robertson, moved to Silicon Valley. In short order, Google and then Sequoia Capital, one of Google’s original venture capital backers, invested in Meraki.


Moore’s Law, with its regular doubling of transistors on a single silicon chip, makes possible the miracle of a Meraki “mini,” as the company calls its basic product for the home. It contains a Wi-Fi router-on-a-chip, combined with the same microprocessor and same memory that formed the heart of a Silicon Graphics workstation 10 years ago. These components are now cheap enough today to be included in a box that sells for $49.


The fact that 200 million Wi-Fi chips will be manufactured this year leads to economies of scale that will drive down the price of extremely intelligent network equipment. Meraki’s products are still being tested, but word-of-mouth has attracted 15,000 users in 25 countries.
One early adopter was Michael Burmeister-Brown, a director of NetEquality, a nonprofit in Portland, Ore., that provides free Internet access to low-income neighborhoods. He had not been impressed by Portland’s municipal Wi-Fi service. Because the Wi-Fi transmitter has to be both close and within unobstructed view, the limitations brought to Mr. Burmeister-Brown’s mind the sign on the back of 18-wheel trucks: “If you can’t see my mirror, I can’t see you.”
In Portland, the access points were installed only at every other intersection in residential areas — creating an “I can’t see you” problem. MetroFi, the service provider, advises residents who are not close to a transmitter to buy additional equipment to pull in the signal, with a starting price of $119 — and that is without the “professional installation” option.


For NetEquality, Mr. Burmeister-Brown decided to try out the Meraki equipment in several neighborhoods. In the largest, consisting of about 400 apartments, five DSL lines were used to feed 100 Meraki boxes, which cover the complex with a ratio of one box to every four apartments. Each box both receives the signal and passes it along, albeit at diminished strength. For an initial investment of about $5,000, or $13 a household, the complex can offer Internet access whose operating costs work out to about $1 a household a month.


The bandwidth can match DSL service, but here it is throttled down a bit to deter bandwidth-hogging downloads. Nonetheless, Mr. Burmeister-Brown says everyone is able to enjoy Web browsing with what he describes as “really snappy response.” The sharing of signals among neighbors does not compromise privacy if standard Wi-Fi security protocols are switched on.
Meraki’s products are not yet for sale, and its networks have not been tested with extensive deployment across a large city. Nonetheless, the intrinsic advantages of its grass-roots approach, with next-to-nothing expenditures for both equipment and operations, are impossible to ignore.


MR. BISWAS says there are about 800 million personal computers in the world, but only 280 million are connected. The rest are “stuck in the 1980s” — close to being connected, but not quite.


Meraki does not wish to go into the Internet service provider business itself, but it aspires to equip any interested nontechnical person to become a “micro” service provider for his or her local community. If the provider wishes to use advertising to cover costs rather than charge an access fee, little would be needed in order to cover the minimal outlays for equipment and operations.


This low-cost network model offers the prospect of broadband service reaching inside many more households. One billion and one. One billion and two. One billion and three ... .