Wi-Fi Public Access from the Ground Up

Planning, installing, and operating a Wireless Internet Service Provider (WISP) business.

By Dr. Francis “Butch” Anton
CTO, Gemtek Systems

Whether you are a location owner with a steady stream of potential customers or a CLEC interested in amortizing bandwidth, the business of for-pay Wi-Fi public access can be represented as a value chain. Each link represents a distinct element.

What’s involved in planning, installing, and operating a WISP business? Let’s study the pros and cons of “build” versus “buy” in public access networking, and concentrate on what it takes to be successful at each particular component.

Who should read this article?
Techies with an entrepreneurial spirit. Wi-Fi enthusiasts.
What you need prior to reading this article:
You'll need some money, time and energy to take this knowledge to the street.
Where can I get the stuff you talk about in this article?
Check with manufacturers.
Where can I go to get help if I need it?
Join the lists and ask a fellow member.
Where can I go to give comments on this article?
In the forums.

What’s your Business Plan?

First question: should you try to go it alone, building out the network, acquiring and supporting customers by yourself, or is it better to establish partnerships? Perhaps the single biggest thing to take away from this article is that you don’t have to do everything yourself. Examine the value chain, concentrate on the areas where you can add the most value, and pick strong partners for the rest. A broker can strengthen your value proposition, resulting in a successful for-pay wireless business for you and greater value for your customers.

I. Network build out and maintenance

Building a network is the most capital-intensive part of the WISP business. The two main components are fixed capital costs of network equipment and recurring costs of bandwidth and network maintenance.
If you build a network that requires authentication and billing, you’ll need to purchase either high-end access points that integrate those functions or network equipment to perform these tasks. On the other hand, if you find a partner to perform the essential AAA (authentication, authorization and accounting) functions, you can focus on building the network inexpensively.

Partnering companies can offer services ranging from selection of bandwidth and hardware to purchasing to installation and maintenance. This is the right option if you don’t have a networking background, or if you intend to focus on customer acquisition and support.
If you are going deploy your own network, here are the steps you’ll need to follow to get your network up and running.

I.1- Bandwidth selection
A WISP sells bandwidth, so consider these factors:
- The amount of bandwidth required
- Bandwidth performance
- Redundant circuits and IP provisioning
- At least one static IP address

You can determine how much bandwidth you’ll need for your for-pay service based on the number of simultaneous users you expect and how much bandwidth each of those users might reasonably consume. Remember that Internet access is typically “bursty”: it is highly unlikely that users will all need lots of bandwidth simultaneously. Moreover, the traffic will be largely downstream traffic, meaning that data will be flowing to the customer, rather than from the customer. A good rule of thumb: you can handle up to about 10 simultaneous users with a DSL line (384K downstream, 128K upstream). If you anticipate more users, most providers have tiered DSL offerings, so you can increase the bandwidth provided both upstream and downstream. Typically, you can get a 1.5M/384K DSL line fairly inexpensively. This type of DSL line is usually a best-effort line, meaning that there are no guarantees on the bandwidth available at any time. If you want to offer carrier-grade services where best effort is not an option, consider purchasing a business-grade line, which usually comes with guarantees on bandwidth availability. Higher bandwidth circuits, like T1 lines, are also dropping in price, so if you require greater bandwidth this could be a cost-effective way to address the need.

You might also consider a partner who will deliver bandwidth to your door, and provide some of the other network elements, like equipment or installation. This will reduce the number of companies with which you have to work, making the process run more smoothly.

I.2- Site survey
When deploying a wireless network, it is imperative to ensure adequate RF coverage in all the areas where your customers will use your service. For single AP installations, this is a no-brainer. However, in more complex, multi-AP installations you must determine the optimal number and placement of APs for adequate coverage, either through actual measurement or through modeling.

Performing a site survey involves determining the number and position of access points, antenna selection/placement to maximize coverage, channel selection for the access points to prevent intra-channel interference, and should take into account any negative conditions that might exist, such as excessive path loss or multi-path distortion.

Path loss is the loss of power between the transmitter and receiver (the AP and the station). A number of factors contribute to path loss, such as physical distance between the two and intervening obstacles, like walls and objects. As path loss increases, data rates decrease. Minimizing path loss is essential in establishing and maintaining high data rates.

Multi-path distortion occurs when radio waves bounce off objects between the transmitter and the receiver. This can adversely affect a Wi-Fi network. Equipment vendors fight multi-path distortion by using diversity antennas on an access point and logic that measures the signals received by each and compares them, correcting for any distortion. If multi-path distortion is a factor in your environment, you should use APs with diversity antennas to measure and compare received signals and/or relocating the access points.

Wi-Fi (802.11b DSSS) operates in an unlicensed area of the 2.4GHz frequency band. This band contains 80MHz of available spectrum. Each channel usually occupies about 22MHz of spectrum, and in order to minimize interference needs to be spaced 25MHz from adjacent channels. Doing the math, you can see that there is only room for three non-overlapping channels, typically channels one, six, and 11 in the US. That means that in a given coverage area, you can have at most three non-interfering access points. If each access point uses a different network name, or Service Set Identifier (SSID), this type of configuration provides three distinct wireless networks, each having a maximum bandwidth of 11Mbps. However, if you wish to increase the total effective bandwidth beyond 11Mbps in a given area, you can deploy up to three access points in that area sharing the same SSID but configured on non-interfering channels, thereby increasing your total available bandwidth in that area to 33Mbps.

There is much more to performing an accurate site survey. If you are planning a multi-AP installation, either hire someone who is familiar with performing site surveys and has the necessary tools, or partner with a company with a proven track record of performing site surveys.

I.3- Equipment selection – Four Examples

Let’s examine four possible network configurations. The first two types assume you are deploying a closed network, i.e. one in which you own, deploy, and maintain the entire network as well as acquiring, supporting, and billing the entire customer base.

The second two show open network configurations, i.e. one in which you own your part of the network and your part of the customer base, but which you have opened up to access from visiting customers through the use of an AAA and billing broker.

I.3-a. Closed network, single access point

A single access point is usually sufficient to provide complete for-pay public wireless service for a location like a coffee shop or a bookstore. The architecture is fairly simple and requires a single access point with AAA functionality, a subscriber database, and some way to do billing.

The access point should provide DHCP functionality to assign IP addresses to your customers, and NAT functionality if you are sharing a single IP address.

Access control is typically performed by browser redirection to a login page. The user enters his username and password and that information is sent to the AAA server. Once the user has been authenticated and authorized, he is usually redirected to another page telling him that he can now access the Internet. This mechanism requires user credentials to be stored locally for each user authorized to use the network. One way to do this is to use a RADIUS server, interfaced to a relational database.
Finally, some sort of billing system is required, unless you are giving the service away for free or are using a pre-paid only system. With a pre-paid system, chances are you can get all the functionality you need from your AAA system to support debit-based accounting.

I.3-b. Closed network, multiple access points

For larger installations where the location size or construction makes it impossible to provide sufficient RF coverage with one access point, you must deploy multiple access points. This can increase the infrastructure complexity a little or a lot. Although the concept of a basic multiple access point installation is similar to a single access point installation, redundancy is much more important in a large installation than in a small one. This stems from a number of factors, including physical size of the installation, difficulty in getting to equipment at all hours, and the expectation of the customer concerning network operation. For instance, if the network is installed in an airport or train station, customers will access the network 24 hours a day and will expect it to perform reliably all the time. On the other hand, if the network is installed in a coffee shop, customers generally expect service to be available only during business hours. Implementing the redundancy required to meet these expectations and to offer a quality service can become quite expensive, especially when it’s not only networking equipment that requires redundancy.

For a multiple AP installation, access control must be used. It can be provided by a box in the network or by each access point. Providing access control at a central location makes the user experience cleaner, because, in general, all access points managed by that access control device share authentications. This means that as a user moves from access point to access point, he won’t have to re-authenticate. However, access control on the access point eliminates an additional box and thereby reduces the cost. Furthermore, some AP vendors are starting to include the ability for one access point to act as a master access control device, and the others in the network defer the access control to the master. Choose the technology that is appropriate for the user experience you wish to offer as well as the cost point you wish to meet.

Now that we’ve looked at typical closed network configurations, we’ll examine what is needed to build open networks in which customers of other networks or customer owners can pay for visiting your network without having to sign up as a customer of your network.

I.3-c. Open network, single or multiple access points

As in the example of the closed network, the vast majority of open network installations will require only a single access point. However, even if the installation involves multiple access points, the complexity of installing, owning, and operating an open network would be far less because you can outsource the customer acquisition, customer support, authentication, billing, and settlement functions to concentrate solely on owning the network.

In this case the AAA and billing functions of the closed network are replaced by a partnership arrangement with a broker/settlement/clearinghouse company. Outsourcing these functions allows you to focus on the network component. This reduces the cost of the network, showing much quicker return on investment (ROI). The customer base you draw from is greatly increased, since customers for other networks who have relationships with the broker can visit and pay for use on your network.

The same principle applies in a multiple access point installation, however the benefits can be greater depending on network architecture. Where redundant equipment is required to build out a reliable closed network, outsourcing entirely eliminates those functions. This greatly increases the ROI of such a network, since the infrastructure costs are limited solely to the APs and related networking equipment.

II- Equipment installation and configuration

If your AP requires physical access to configure it, make sure you do this (and test it) before you install the access point in a location where it will not be readily available!

Installing access points can be either very easy or very hard. In some cases, installation consists of nothing more than placing the access point on top of some existing object, like a bookshelf, near the center of the area to be covered. If the installer is lucky, the bandwidth will be nearby and the network cable can be hidden quite effectively. Unfortunately, things don’t always work out this cleanly. Location owners have gone to great trouble to craft an environment and their plans might not have included the access points being installed. This can lead to more challenging installations, such as placing access points inside false ceilings. This type of installation has two benefits. First, with proper antenna positioning, the area below the access point can be covered very easily while not losing RF signal to areas above the access point (i.e. in the roof) where there are no customers. Second, it is very easy to hide network cables in false ceilings. Disadvantages in this type of installation include accessibility and lack of power at the installation location. If it is in a false ceiling or other hidden location, the AP is not readily accessible when there is a problem, either for troubleshooting (i.e. looking at the status lights) or physically accessing the device for service or replacement. Furthermore, there are rarely power outlets in false ceilings, so arrangements should be made to safely power the APs. An excellent choice is to use Power over Ethernet (PoE, or IEEE 802.3af), which provides power over the existing network cable. With the right management software, PoE also allows you to remotely power cycle your APs.

Configuring access points is usually very straightforward. Although AP manufacturers have not settled on a standard for configuring these devices, you can use one or more of the following methods: serial access, telnet, web browser, or SNMP. Some of these require physical access to the device, so be aware of that when performing your installation.

Depending on how your network is architected, different devices might provide different services. For example, if you’ve chosen to use an external device to provide DHCP and NAT functions, you’ll need to assign a static IP to this address and NAT addresses to the local network devices, perhaps including the access point(s). If this is the case, make sure you chose an AP that will obtain an address via DHCP.

Your network provider, especially if you have obtained bandwidth via DSL, may use PPP over Ethernet (PPPoE) to provision the connection. PPPoE is typically used in a single computer environment where a PPP session is established between the client computer and a server (usually a DSLAM). This does not work very well in a public access environment, because you can have only one PPP tunnel active at a time. If you find yourself in this situation, make sure that you purchase an AP or other network equipment that can terminate the PPPoE tunnel at the edge of the network and provide NAT and DHCP to the local LAN.

Some sort of access control must be provided to prevent people from using your network for free. Conceptually, the type of access control required doesn’t vary from the single to multiple access point installations, although there are practical considerations. It’s better if the access control in a multi AP installation were not done on a per-access-point basis, since this would require the user to re-authenticate each time he associated with a new access point.

Choosing a descriptive SSID for your network is also important. The network name allows your customers to know that they’re using your network and provides an important branding opportunity for you, the network owner. Choose a name that readily identifies your network and gives brand reinforcement. In multiple AP installations, using the same SSID across access points is essential for the user to be able to roam freely between them.

III. Customer Acquisition is Different from Network Build Out

Attracting and retaining a customer base large enough to sustain the business is the most difficult element in the WISP value chain. Building out a large network footprint is capital and time intensive. If you partner with a company that provides sufficiently large network footprint, you can to achieve the “network critical mass” required to attract customers, and you can focus on offering those customers a superior product.

PART TWO

Customer Support

Having a for-pay public wireless network is no good if nobody uses (and pays for) it. Let’s consider what demographic to approach for your customer base, how to target those customers to attract them to your network, and how to take care of those customers once they’re your customers.

As is the case of network build out, partnering with a company that represents a customer base is an effective way of obtaining customers without doing the work yourself. In addition to not having to do all the work, you can leverage the relationship, which might include a very strong brand, that your partner has with their customers.

Target customers

While you would love for everybody who has a Wi-Fi-enabled device to visit and pay to use your network, realistically that won’t happen. So, it’s important to identify a market segment that is most likely to use your network then directly or indirectly target those customers to bring them to your network.

Do not overlook the obvious target customers. Are there people who already visit your location and use their laptops? If so, talk to them. Find out what they do for a living and what they do with their laptops while they are at your location. This helps identify a market segment for you to approach.

Another approach is to look for customers to whom wireless Internet access would be complimentary to the products or services you already offer.

Finally, never underestimate the value of market research. There are several ways to perform market research. One way is to pay for studies done by a well-respected research firm. This can be expensive and the value you obtain may not be easily quantified. The Internet is a low-cost source of market research However, perhaps the best set of market research lies with your competitors. What works for them? What doesn’t work for them? Knowing what not to do is sometimes as important as knowing what to do.

Targeting those customers

Once you have identified one or more customer segments to approach, how do you proceed? After all, word of mouth is only so effective at reaching your target audience. Advertising and promotions are two ways to reach your intended customers.

Advertising

Think of advertising as information about your service placed into the hands of those who are likely to use your service. Your advertising must clearly state the advantages of your service and place it in a context appropriate to your potential customers.

There are many different kinds of advertising, including print media, radio, television, and, increasingly, the Internet. Some of these may have marginal value, depending on your target audience, so try to choose one that matches well with them.

If you have decided to partner with a company to acquire customers, joint marketing may be available to you. This is a great way to reach a customer base that is already receptive to these types of services. Your partner company may have programs in place to co-market the service. Examples of this include complete sets of web templates to advertise the service on your website and collateral for your wireless network location(s). Take full advantage of these benefits.

Promotions

Promotions can inform potential customers about your service. For instance, you could offer a free day to potential customers or work with a Wi-Fi card vendor to include a free day of service in the box of each card sold. If your service is competitively priced and offers good utility, these activities will result in new customers at a fairly low acquisition cost.

Sign up

Once you have convinced customers to use your service, you’ll have to get them signed up. Your partner company may have an entire drop-in web architecture for this: otherwise you’ll need to do it yourself. Make sure that the mechanism you choose interfaces properly with your back-end AAA solution. Also make sure that you provide adequate security for obtaining and storing customer information.

The sign up page is probably the first interaction that a potential customer will have with your service. Make sure that the execution of the sign up page is clean and tight, meaning that there are no unnecessary elements in the design and process and that the entire process is thematically tied. Remember, you don’t have a second chance to make a first impression.

Support

Customers will have problems with your service. This is a reality. You cannot control what the customer has on their access device, nor do you have control over how they think. Therefore, you need to take advantage of those things you can control, like the availability of your service and how you resolve problems when they do occur.

Provide your own in-house support or outsource? Will it be 24x7 or only business hours? Do you provide e-mail only or telephone support? All of these are important questions and will determine how much customer support costs you. Also remember that if you partner for your customer acquisition, customer support may be included as part of the deal. This could result in significant cost savings, so seriously consider this during your decision making process.

Billing

Once you have a service and customers, all you have to do is get paid. There are two components to this: the type of payment plan you offer and the mechanism by which you actually get paid. In this section we’ll discuss pre-pay versus pay after use and then we’ll talk about actual payment instruments, like cash or credit cards.

Pre-pay vs. pay-as-you-go

Two payment options exist in a for-pay service: pay before you use (pre-pay) and pay after you use (pay-as-you-go). Pre-pay is much more convenient for the service provider (you), and pay-as-you-go is much more convenient for the customer.

Pre-pay can be implemented in a couple of different ways. One way is to have the customer buy a block of access in advance. Depending on how you charge for your service, this may result in the user purchasing a fixed number of minutes, hours, days, bytes, etc. Another way to implement pre-paid billing is to sell the customer a voucher, representing a single or multiple use of your network. Physical vouchers cost money to produce, so consider that cost when making your decision.

The block of access purchase is probably easier to implement in an on-line fashion via a website, and the voucher system is probably more suitable for point of sale transactions, though the line between the two blurs more and more over time.

Pay-as-you-go is suitable for long-term customers. Initially there can be more risk with pay-as-you-go customers, because they are consuming services for which they haven’t yet paid. Furthermore, they have no established credit history with you. However, for the customer, pay-as-you-go is certainly more convenient, since they are automatically billed at the end of the billing period for their actual usage. Additionally, they don’t have to be concerned with running out of pre-purchased time at a critical moment. However, pay-as-you-go is the least anonymous of any of the payment options. If you are a customer owner who wishes to retain customers over the long haul, you really need to offer a pay-as-you-go option. To minimize risk, you can do this in combination with automatic credit card billing or account debiting.

If you have chosen to partner with a billing provider, the agreement may include one or both of these payment options. Because a billing provider will aggregate customer owners, and hence customers for purchasing vouchers, the per-voucher cost could be substantially less when partnering. All of these factors should be considered when deciding to go it alone or use a partner.

Cash, check, credit card

What payment instrument should you choose? Credit cards are the obvious choice. If you are going to accept only one payment instrument, make it credit cards. However, credit cards offer no anonymity. If you feel that this is a differentiating factor for your service offering, consider coupling point of sale voucher cards. Checks present a bit of a difficulty. One possibility is to only offer point of sale vouchers payable by check. More difficult is to offer pre-pay block sale via check. Finally, offering pay-as-you-go via check is the most risky, since the service has already been consumed long before the check has cleared. You may want to consider limiting your exposure initially by forcing a customer who wishes to use a check to pre-pay for the first time or two until they have established a positive payment experience with you. Only after that should you consider taking checks as payment for pay-as-you go charges.

Credit cards allow for instant gratification for the customer. However, they also open you up to instant fraud. Without adequate credit card verification, you expose yourself to potentially devastating loses. I cannot state this strongly enough: adequately protect yourself against credit card fraud.

Partnering in customer billing is a very good option to mitigate many of these difficulties. Your billing partner can assume nearly all the risk and difficulty of collecting money for end users. Remember this when you make your decision of building versus buying.

Brokers

Brokers play a critical role in the wireless value chain because they bring together customer owners with network providers. This way it is possible to focus on building only one side of the value chain (i.e. either customer ownership or network ownership) and still have access to a successful counterpart in the value chain. As new networks are added or new customer owners sign up, both gain access to each other through the broker.

While customer owners and network owners can sign direct bilateral agreements, it is easier to sign a single agreement with a broker. The aggregated customer base represented by a broker is far greater than any single customer owner can hope to achieve. Because of this, the broker can negotiate very favorable rates with the network owners. For the network owner, having immediate access to a large customer base makes building out network footprint easier to justify.

If you are a network owner, build an open network and affiliate yourself with a broker. If you are a customer owner, you’ll gain access to the largest network footprint you can by signing up with a broker.

Summary

As you can see, a lot goes into building out a for-pay Wi-Fi public access network. While it may seem daunting, it doesn’t need to be. Examine the value chain and determine where you can best add value. Find partners to help with the other parts. Build a business model that supports these choices. Find your target customers and tell them about your service, focusing on ease of use and utility. Then listen to what they have to say after they use your service. Remember, you’re building a living network that must change with your customers’ needs.

Good luck!