First-Hand:Wi-Fi's Early Days
Wi-Fi's Early Days
Submitted by Alex Hills
We hunched over our sandwiches in the crowded university cafeteria. It was November 1993. The room around us hummed with restless conversation. Students bolted down their food and compared homework solutions before rushing off to classes. And the two of us struggled with a problem – a problem we had made ourselves.
Marvin Sirbu and I, both professors at Carnegie Mellon University, had worked together to create a new institute within the university. But the fledgling institute needed our help. It didn’t quite fit the Carnegie Mellon mold, and its future might be in doubt. We were both a little worried.
It was called the Information Networking Institute. In the late 1980s and early 1990s, a period of accelerating innovation, computer and communication technologies were merging into one, and their importance was rising in the world of business. That’s why our institute was interdisciplinary. It offered an educational program to train a new kind of professional – one proficient in both technologies and also knowledgeable about business. The institute’s new master’s degree program was a direct response to the increasing demand for such graduates. We were educating the new breed of professionals. But Marvin and I also wanted to engage Carnegie Mellon professors in information networking research. We knew that a Carnegie Mellon institute could not be successful without a strong research program.
We needed to come up with a project encompassing all the parts of information networking and drawing on the talents of existing Carnegie Mellon professors. We were a small institute, and we couldn’t afford to recruit new faculty. Yet the project would need to attract outside money from government and private sources.
Talking over the clatter of plates, I asked “what about wireless?” Marvin stopped chewing. His eyes widened. “Wireless?” he said. His mind was racing. I said I wanted to create a new high-speed wireless network – a network that could be used by wireless researchers to do experiments. I thought that we might someday be able to expand the wireless network to cover the entire campus, allowing it to be used by all of our students, faculty and staff. As we talked, Marvin warmed to the idea.
Marvin and I had heard the buzz about wireless. Cellular telephones were gaining popularity in the United States and elsewhere. Mobile computers were on the horizon. Futurists talked about the day when a computer would be carried in a shirt pocket and remain connected to a computer network, wherever its owner went. It seemed like science fiction, but there was an obvious synergy between mobile computing and wireless communication. When computers became truly mobile, wireless networks might allow seamless, anytime-anywhere connection to the Internet.
Marvin and I struggled to hear each other’s voices over the clatter of dishes and silverware, but we agreed that a wireless project could build on the strengths of the new institute and on our own interests. Marvin was interested in computer networking, and I had worked with radio, the underlying technology that could support a wireless network. There were other professors at Carnegie Mellon working in areas related to wireless. They would be interested, too. Our enthusiasm rose.
We were on the verge of a decision that this was the right project for the institute when Marvin realized he was late for a meeting. He pushed back his chair, stood up and rushed off. I lingered at the table, beginning to make a plan. A few of our Carnegie Mellon colleagues were already working on mobile computing, and they would want to join the research initiative. They could use a wireless network to support their work. The combination of wireless and a set of interrelated mobile computing projects could be a powerful mix that would attract other researchers and lead to the new research that the institute needed.
The cafeteria was closing. I took my tray to the dishwashers. The room, now empty, had become quiet. I walked back to my office, thinking about wireless and wondering what problems lay ahead.
Nearly four years later, in October 1998, our wireless team members drifted into the room in ones and twos, cradling coffee cups. We clustered around the big pot of coffee on the conference table and began an intense session, poring over the building floor plans spread out before us. We had already done this many times, at least once for each building’s wireless design. I looked around the table. We had a great team working on the “Wireless Andrew” network. Andrew Carnegie and Andrew Mellon would have approved.
Our newest team member was Lisa Picone, the first woman to join us. She was born and raised near Pittsburgh, but, because of her travels, her Pittsburgh accent had faded. She had been an electronic technician in the U.S. Navy, making the rank of Petty Officer Second Class and spending most of her eight Navy years in Europe, some in Naples, Italy and the rest at SHAPE – Supreme Headquarters Allied Powers Europe – in Mons, Belgium. Lisa worked easily with the guys. She was, after all, a Navy veteran. She didn’t need to draw attention to herself. She was quiet, confident, and capable. She knew what she was talking about.
Lisa had joined Mark Campasano, also quiet and confident, as a technician on our field measurement team, working in buildings across the campus, doing coverage testing and choosing access point locations. Mark was the original technician working on the project. He had been with the team since 1994.
In the windowless room, sitting around the conference table, we, one by one, fell silent, descending into thought. We were following a familiar routine, using a wireless design process that was by then well developed.
A few years earlier we had sketched the beginning of a design method and, over time, improved it bit by bit. We aimed to place access points in a way that would provide complete coverage. Our design process assured, even with fluctuating radio signals, that a computer user anywhere in a building would have a reliable network connection. The key was correctly locating the access points.
Yet our conference room design sessions were possible only because Lisa and Mark had spent many hours making measurements. The two worked to find the best places to install access points, and their task was complex. The coverage area of one access point interlocked with the coverage areas of other nearby access points. All of the access points worked together. In each building the two technicians went through a step-by-step process. They chose a location for an access point, located a second access point, and then a third. They didn’t try to analyze the jumble of radio signals bouncing up and down the hallways – they just measured signal strengths and placed access points accordingly.
Sometimes I worked with Lisa and Mark, helping with the measurements. They followed what we sometimes called the “duct tape method.” It was a trial and error process. Mark set up an access point in a temporary location, using duct tape to hold it on the wall. Lisa measured its coverage by walking through the building and measuring signal strengths to find the edge of the space where a good signal could be received. Then Mark chose a location for a second access point, a location that would give complete coverage of a second part of the building, and Lisa made more measurements.
It was a sequential process. First one access point…then another…and then another. It sounded simple, but, with all its details, it wasn’t. It often didn’t make sense to place an access point either directly above or directly below one already installed on an adjacent floor because their coverage areas would have too much overlap, causing unwanted interference between the two units. So we staggered the access points, moving the second access point down the hall so that the coverage areas of the two access points, one on the floor above and the other on the floor below, dovetailed. It was time consuming, but the two technicians were patient. And patience was valuable – especially on this project.
In our conference room meetings, we worked to create better, more effective techniques for Lisa and Mark to use. We developed the procedure that was later used to design many Wi-Fi networks across the US and the world, but then the process was a new one.
On the big conference table were a set of building drawings that Mark and Lisa, working from their measurements, had converted to coverage maps. They used colored shading to show the coverage areas for each building. Blue showed the coverage area of one access point, red the coverage area of another, and green another. The colors made it easier to visualize the access point coverage areas in the three-dimensional space of a building.
Sitting around the table in deep thought, we all tried to think of ways to improve the technicians’ design by adjusting the access points’ locations. Sometimes there was a better idea about their placements. After everyone agreed on a design, it was approved by the team, clearing the way for access points to be installed.
But that still left the problem of channel assignments. We needed to decide which radio channel would be used by each access point because our network now had three channels. Traffic congestion, and therefore performance, of the network depended on making these channel assignments correctly. This additional task was needed because of an event that had happened a year earlier.
1997 was the year that the standards organization of the IEEE adopted its new IEEE 802.11 standard. It was the first standard for wireless LANs and soon became known as “Wi-Fi,” a name that was better for marketing than a string of letters and numbers. Wireless insiders knew that the standard was a big deal. It assured a purchaser of wireless networking equipment that it would be compatible with similar wireless networking gear made by other manufacturers. This hadn’t been true before 1997. Our original AT&T wireless LAN equipment, for example, was incompatible with equipment made by other wireless LAN manufacturers. A network user with equipment made by a different company could not have used Wireless Andrew in those early days.
But the new standard changed things. It gave customers confidence that the new wireless technology was not just a fad. Equipment manufacturers were committed to making compatible products. They believed in the new wireless LAN market. The standard told the world that the new technology was real and was here to stay.
However, the standard meant we would have to move Wireless Andrew to a higher radio frequency. We had been using the 915 MHz UHF band, a band that was unlicensed in the United States but not in many other countries. Wi-Fi equipment would use the higher 2.4 GHz UHF band, one that could be used for unlicensed operation in almost all nations. Companies making the new 2.4 GHz Wi-Fi equipment could confidently expect to sell their products around the world.
We were happy to abandon the 915 MHz band. Spectrum congestion and interference had been causing problems for us. There were too many other signals at 915 MHz – everything from cordless telephones to baby monitors – because no FCC licenses were needed. We called it the “kitchen sink” band – everything was in it save the kitchen sink. And, to make things worse, there was a high power paging tower nearby. The tower’s antenna, as it sent signals to personal pagers, or “beepers,” operated just outside the 915 MHz band and caused us problems.
The 2.4 GHz band was more spacious than 915 MHz. It was a bigger chunk of radio spectrum. 2.4 GHz allowed the use of three Wi-Fi channels – not just the one we had been using. And this permitted frequency reuse. We could set access points to use different radio channels, minimizing interference between access points and providing much better network performance. It was the same technique that had been used for years by the cellular telephone industry to efficiently utilize available spectrum. The technique could now be used with Wireless Andrew, but it would complicate our design process.
But 2.4 GHz came with its own set of problems. It had a different kind of unlicensed interferers – the ones used to cook food. Microwave ovens heat food by exciting the molecules they contain, causing the molecules to do a little dance, increasing the intensity of their already frenetic motion. This increases the temperature of the molecules – and the foods that contain them. And microwave ovens operated at 2.4 GHz, the same frequency used by the new wireless standard. But some microwave ovens were more troublesome than others.
Our experiments found their differences. The mesh encased in the glass of an oven’s door is part of the shielding that prevents radio energy from escaping into the surrounding area. New ovens were pretty well shielded. It was the older, leaky ones that allowed radio emissions to escape, and these were the ones that caused interference. Over time we weeded out the troublemakers on campus and then didn’t have further problems with interfering cookers. And 2.4 GHz cordless phones had begun to appear – another source of radio interference for Wireless Andrew. In the end, these phones would be a bigger problem than microwave ovens.
But overall the new standard was a positive development. I had confidence that Wi-Fi equipment would be a market success. It seemed likely that other universities, companies and government agencies would soon follow our lead and begin to build Wi-Fi networks like ours.
Watching our students use the new technology told me that the demand for Wi-Fi would be big. I received messages from many of them, asking how they could sign up to use the network. The official answer to this question was always the same: “Wireless Andrew is a research network and is intended only for the use of faculty members doing research in the wireless area.” But Carnegie Mellon students are clever, and they found ways to connect to the network. I knew what was going on and didn’t say much about it. Technically lawbreakers, the students were really “early adopters” and up-and-coming technology innovators. I didn’t want to discourage them.
It seemed that the vision of anytime-anywhere high-speed wireless Internet access could soon become reality. I thought about expanding Wireless Andrew to cover the entire campus – my original dream – and opening the network for use by the entire campus community.
By 1999 the Wireless Andrew network was up and running – covering all of the Carnegie Mellon campus. But our work wasn’t finished. Since 1997, when the IEEE 802.11 standard was adopted, the buzz about Wi-Fi networks and Wireless Andrew had increased. Articles in journals and magazines were one reason. Countless phone calls, e-mails and letter inquiries showed up at our offices. Other universities wanted to build their own Wi-Fi networks, and they wanted to know how to do it. At first we liked the attention, but soon the flood of inquiries overwhelmed us.
In an October 1999 e-mail message, a team member complained, “The good news is that Alex’s [writing] efforts on Wireless Andrew worked. The bad news is that we are now getting many requests for visits to see it and to discuss it with us.” We needed an efficient way to handle all the requests.
We decided to ask our office staff to respond to the phone calls and e-mail messages by inviting inquirers to attend monthly briefings describing the Wireless Andrew network, our design methods, and Carnegie Mellon’s wireless research.
The wireless network that was just a dream in 1993 had inspired many to understand what the new technology made possible.
Note: You can read the whole story of Wireless Andrew and its impacts in “Wi-Fi and the Bad Boys of Radio,” a book by Alex Hills. See http://www.amazon.com/Wi-Fi-Bad-Boys-Radio-Technology/dp/1457505606/ref=sr_1_1?s=books&ie=UTF8&qid=1315772834&sr=1-1