Taking the Internet WirelessSaturday, September 15, 2007
As I was reviewing what is being written about the 700-MHz actions―who will bid, who will not, how it will all shake out and how much spectrum is enough to build a new network―I began thinking about the differences between connecting to the Internet via DSL or cable modems and connecting using wide-area wireless broadband data services. There are significant differences between the connections themselves and how capacity is added.
Let’s start with a connection to the Internet over DSL or cable that is always on and always connected. At home, this connection is yours alone and regardless of the bandwidth you signed up for; you are the only one using that bandwidth to reach the network hub or the Internet. Once connected to the Internet, your packets may be routed differently, but all of them end up at the same location because of the way IP packets are designed. In this mode, we are in a one-person, one specific bandwidth scenario, at least until we reach the Internet.
When you connect via a Wi-Fi access point, say at Starbucks, and that access point is connected to the Internet via a T1 connection with a total data speed of 1.54 Mbps, you usually have access to most of that bandwidth. However, if there are ten other people on the same access point, that bandwidth is shared among the group. Generally, this works pretty well. Your overall speed might be reduced to half or less of what it was before, but you stay connected and you are able to do your work. This is because we are sending and receiving packets. Your packets might be co-mingled with other customers’ packets, but it all sorts itself out on the Internet and everything reaches its destination.
One reason your session does not come to a grinding halt with ten customers sharing the same access point is because you are usually transmitting and receiving at different times and not on a continuous basis. There are two important points to keep in mind here. The first is that when you are using an access point, you are using wireless bandwidth that is shared with everyone using the same access point to send and receive data. Further, the total available bandwidth is not based on the bandwidth of the wireless connection but rather on the limitations of the wire (or fiber) that carries the signals to the Internet.
The second point is that as more customers join you on the access point, your data speed does not fall off in a logarithmic fashion. If you are getting 1 Mbps when you are using the access point all by yourself, and a second user joins you, your data speed won’t necessarily drop to one half of that because in most cases you won’t be transmitting and receiving at the same time―that is, unless you are streaming data, such as video. If you are streaming data, you are in an almost constant receive mode and you will have a much bigger impact on those sharing the access point than if you were surfing the Internet or checking your email.
Cellular (wireless) systems were designed to provide for frequency reuse, and therefore better spectrum efficiency. They were also based on the premise that not every customer would be using their phone at the same time. While capacity was designed to handle high-call volumes within reason, as calls and data usage increase, capacity needs to be increased by adding new cell sites closer together. This is a three-year process in most areas so capacity cannot be increased quickly.
Wired voice networks used to jam up on Mother’s Day and during disasters because the networks were based on a normal usage model. Wireless systems were also built for a normal usage model. Over time, the cost of wireless voice and data has come down, we use our wireless phones more than ever before, we talk longer and we have created the need for additional capacity. Wireless networks, depending on the technology, support data that must contend with voice calls for bandwidth or provide separate channels for data services. The mixture of voice and data requires fine tuning of the system because voice cannot be delayed or it will be unintelligible, so data usually has secondary access in mixed systems.
That is why there are systems with voice-only and data-only capabilities. Still, in times of heavy demand such as during emergencies, heavy voice traffic can cause calls to be dropped or blocked by the system. We have not yet experienced this with data services on our wireless broadband networks, but this does not mean we won’t. The design premise of our existing networks is based on today’s level of expected calls or data. If the model changes, we can end up not having enough capacity even in non-peak hours.
Well, the model is changing and Internet companies will impact the model even more. We are now uploading and downloading massive amounts of data and adding streaming audio and video services that require our wireless sessions to stay on the air for longer periods of time to accommodate these new demands. This new usage model has the potential to create havoc with current network design models, no matter what the technology.
Staying connected to a wireless broadband data connection 24x7 is not an issue, most systems regulate themselves and only send out small amounts of data to keep the connection alive or they go into sleep mode until we want to send or receive data. What is a problem is that wireless broadband data is shared bandwidth on a cell sector-by-sector basis. To understand the concept, think of a cell sector as being a Wi-Fi access point. The difference is that a Wi-Fi access point has a range of about 300 feet and a cell sector can cover a several-mile radius. This means the chances are higher that more customers share the same wireless broadband bandwidth as you and, to make things worse, the further you are from the cell center, the slower your data speed.
This is a very rudimentary description of some of the issues, but it appears as though those who believe wireless broadband systems will have unlimited bandwidth do not understand the basics and the laws of physics, nor do they understand Shannon’s Law, which defines the maximum bits per Hertz of bandwidth. You can bend the laws of physics but you can’t break them.
Let’s go back to the Internet model for a minute. If you are eBay and the amount of traffic to your site starts causing delays for your customers, you add more server capacity, increase storage and perhaps caching capabilities. You might also add another fiber connection to the Internet and/or connect to the Internet at a different point. In a short period of time, you have reduced the delays and increased your capacity, all for a minimal expense.
If you are a wireless network operator and your customers begin to experience data delays or blocked or dropped calls, to increase capacity, you make sure your system is working properly, look at the amount of backhaul from the affected site to the network and possibly make some antenna changes. If you are lucky and have additional spectrum available, you can put it into service at that cell site. But if you do all that and still have a capacity problem, you have one, and only one, option―build additional cell sites.
The premise of cellular communications was to start with cell sites, say, ten miles apart, and as traffic increased you would split the cells and add more so they were five miles apart, then two and one-half miles apart, each time adding capacity to the system, increasing the cost of the system and increasing operating expenses. In congested metro areas, your cell sites might be very close together and you might even fill in with indoor coverage using Wi-Fi, bi-directional amplifiers or femto cells.
Unlike eBay, which can accomplish system-wide capacity expansion in a matter of a few days or weeks, wireless networks must expand their capacity one cell site at a time, and the timeframe is usually about three years. Here is a shortened version of the process. First you determine that you need more capacity and then you develop search rings, identifying areas where you could put a new cell site to take traffic off your existing site. After that, you identify several possible locations for the tower or cell site (buildings, etc.) and find out if your competitors have a site in one of these areas that you might be able to share. You then run computer simulations for each of the prospective sites and review them based on your previous experience with the local government agencies.
Once you have found the most suitable sites, you approach landowners and try to convince them to permit you to build a site on their property, paying them a monthly fee for the right to do so. Landowners usually have one of two reactions: They don’t want a site on their property because they are concerned about health issues, or they have heard stories about how much they can make with a site on their property and demand a lot of money.
During this time, if you are smart, you are working with the local governments and preparing them for the fact that you intend to file for permits to build the site. In today’s world, the chances that you will be able to erect a tower and a building and put antennas in the air are slim to none. In most cases, you have to view the area and find a way to mask the antennas to blend in with the surroundings. A mono-pine? Modifying a building edifice? Building your antennas into a sign?
Once you have a design that satisfies local government agencies, you submit your application for a permit, attend numerous hearings where you listen to the public plead with the local officials not to let this cell site be built because it will harm their health, it will be an eyesore and/or it will reduce home values. You take attorneys to these hearings who try to explain what the FCC permits local governments to do and not do. One thing they cannot do is deny the permit based on health issues, but they can deny it for a variety of reasons or postpone a decision for a long time.
Finally you get your permit, order the equipment and build the site. Meanwhile, some of your customers have been getting upset because of the number of blocked or dropped calls they are experiencing or slow data speeds. When the site is finally built, your customers start receiving better service and data speeds return to normal. This process has taken three years and you are repeating it in city after city for sites all over the country.
During this same period of time, eBay has added even more servers and even more fiber connections to the Internet without once filing for a permit or having to go through hearings and delays.
Wide-area wireless is about providing access to information when we are away from our homes and offices, but it is not about giving us the same access we have using DSL or cable. We have to share the available bandwidth with our neighbors and network operators have to add capacity one cell at a time, which is a tedious process.
Those who believe in open access have never stood in front of a city council with a room full of citizens complaining that the proposed cell site will ruin their lives, knowing full well that these same citizens are demanding better coverage! The old Indian expression, “until you have walked a mile in my shoes,” certainly applies here and I invite Google, Apple and anyone else to spend a few evenings at a city council or board of supervisors meeting when a new cell site is on the agenda. Email me and I will take you along to the next one I attend. Perhaps then you will understand that wireless is different from the Internet.
Andrew M. Seybold