Considering a WISP

23 May, 2020

I love the Atlanta area. The city has great food, great people, a fantastic dance scene, some of the best bars and live music of anywhere I’ve ever been, and any number of other endearing qualities.

Unfortunately, one thing that the Atlanta area fails at is providing good internet service. In most of the city and its suburbs the only options are AT&T, Comcast, or Spectrum (what Charter rebranded Time Warner in the hopes that you wouldn’t notice that it still has possibly the worst customer satisfaction of any internet provider) all of which provide terrible service at a very steep price.

I’ve fantasizing about starting a neighborhood non-profit or cooperative (co-op) wireless internet service provider (WISP) for ages. This post is a collection of thoughts on how I might go about doing that, and what a neighborhood wireless network might look like.

Getting Started

The hardest part of starting a WISP is getting the WISP started. By that I mean: where do we get internet service, and how do we setup those first connections that will allow us to expand the network in the area? To do this, I envision sharing my home internet connection with a few neighbors who may just consume internet and pay me every month, or who may become the initial board members of the future WISP. Once the idea is proven with a few houses with line-of-sight roof lines and no need to worry about Fresnel zones, we can start trying to expand to more complicated installs.

Expanding to other houses would require some general ground rules to keep costs low (though any of these can be bent for a particular install if the benefit is deemed great enough):

Avoid renting roof space at a cost (try to provide free Wi-Fi to apartment complex common areas, bars, coffee shops, or lobbies in exchange for roof space)
Avoid any spectrum that requires an FCC permit (stick with 5Ghz unless there’s good reason to use something else)
Avoid renting telephone poles, and don’t rely on donated pole space (use it, but make sure there’s another way to do the install in case the telephone company changes its mind later)

Doing Business

Regardless of the type of tax entity or business vehicle we use, the point of the WISP is to provide better internet at a lower price than is currently available in my neighborhood. Because of this public service mission I envision the WISP being started as a co-op or not-for-profit if it ever expands beyond the “neighbors sharing a wifi connection” stage.

If it were started as a not-for-profit, internet service would have an initial startup fee to help cover the cost of equipment. This fee could be subsidized or entirely paid for by donations from neighbors who want to see the network expand. With enough donations or grants a fund could be endowed to buy the rooftop equipment and the initial setup fee could be lowered or waived. Afterwards internet service would be pay-what-you-can with a suggested monthly donation of $20.

Households that agree to share their own internet connection could donate bandwidth to the network and we would pay part of their internet costs. In exchange they would get reliability (if their connection goes down they are routed through the network to another bridge point), a slightly higher QoS than other households, the ability to use our public access points and stay connected as they move around their neighborhood (more on this later), and we would subsidize part of their internet cost.

In the co-op model I envision members paying for their own equipment out of their one-time membership dues, either by paying their dues up front or by paying slightly more per month over the course of a year.

Once dues are paid the members would have voting rights and discounted service. Any money made by the co-op (after any network expansion projects or equipment payments) could be given back to members and employees in the form of a dividend based on the level of service purchased or the amount of bandwidth donated.

Both the not-for-profit model and the consumer owned co-op model have significant tax benefits over a traditional business, and both would give us access to grants, shared capital funds, and good loan terms that would otherwise be unavailable to us. The main difference is that the co-op model would have the additional mission of generating community wealth, while the not-for-profit model would be purely about expanding the network but wouldn’t necessarily be owned by the members or (eventually) employees.

Networks

Each access point would advertise up to three networks:

the Public Network,
the Community Network, and
the Private Network.

The public network is only advertised in coffee shops, parks, apartment complex common areas, and anywhere else where someone wants to provide their customers with free access to the internet. It has no password and anyone can join the network at a public access point. We would offer it for free to businesses, cities, or strategic individual houses that allow us to put an antenna on their roof: thus helping us expand our network to other households in the area. It would have the lowest Quality of Service (QoS) of the three networks when the link becomes saturated. All devices on the public network are isolated from each other.

The community network has a higher QoS than the public network and is always advertised by every access point in the system. Any paying member can access it using WPA2 and the same login and password they use to manage their account on the website. One of the benefits of using our service would be staying connected to the internet throughout your entire apartment complex, or while walking around your neighborhood if enough members live in your area. All devices on the community network are isolated from each other

The private network isn’t actually a single network. It is a user-configurable network that is unique to each user’s account and is only advertised on their particular roof top access point. Like the community network, private networks also use WPA2 and the user’s account username and password, but they do not isolate connected devices from one another. This is the network that you can use in your own home, or that businesses would use for their point of sales systems and other private infrastructure. User’s private networks have the highest QoS of the three network types so that your home internet is always the speed you paid for even if someone is rudely torrenting without any throttling on the public network in a coffee shop down the street.

Cost

I’ve sourced some equipment and done some back-of-the-envelope calculations for what getting setup with a few initial houses would cost. Surprisingly, the most expensive part is a non-penetrating pitched roof mount. Costs could be brought down significantly if a cheaper alternative could be found (the best price I found online was $160), or if the roof mount could be built out of off the shelf materials. These calculations also are missing miscellaneous expenses like large spools of Cat 5e cable, or indoor access points which may be required for multi-story houses, so take these numbers with a grain of salt.

The total cost for consumer premise equipment comes out to around $238.98 / household and $288.98 for households hosting an internet connection. Unfortunately, this puts it well out of the size where I could expect consumers to just pay for their rooftop equipment up front. The cost of a gig fiber connection in my area is ~$95/mo. If we split this evenly between 8 people and tack on $20/mo to help pay for the equipment, this puts the cost of internet right around $30/mo to break even.

Splitting a fiber connection 8 ways may or may not over provision the link, and once it were more than a few neighbors sharing internet we would want to investigate alternative ways to getting online that hopefully don’t involve reusing expensive consumer or business class internet connections.

SamWhited|blog

Considering a WISP

Getting Started

Doing Business

Networks

Cost

Further Reading