Originally Posted: October 1st, 2020

Wi-Fi 101 — Wi-Fi Origins and History

Where Did Wi-Fi Come From?

Wireless networking has a long and interesting origin story. Many separate teams of people worked on the problem for decades. By the early 1990s, many methods had been developed, and the need for a International standard was clear. It took years of work and dozens of proposals to form the original Wi-Fi standard in 1997. It took even more effort over the last 23 years to get to where we are now. What we call “Wi-Fi” is really a series of wireless networking standards, voted on by a worldwide engineering body, the IEEE.

The story of the invention of Wi-Fi depends on who you ask. There is a complicated history leading up to that original 1997 standard, and the truth is hard to decipher. I'm not capable of litigating who invented what, but there are some key players to know about. Where possible, I will point out the people and the organizations behind the technology.

Wired and Wireless Precursors: Say ALOHA to Wireless Networking

In 1968, researchers at the University of Hawaii began to investigate if radio waves could be used for communication between computers. The project was led by professor Norman Abramson, Dr. Franklin Kuo, and a team of faculty and students. It would go on to become known as ALOHAnet.

During this time, wired computer networking was in its infancy, and evolving rapidly. Packet switching was conceived of by Paul Baran in 1964, and first demonstrated by Donald Davies in August 1968. In October 1969, the first ARPANET link was established between UCLA and the Stanford. By June 1971, ALOHAnet was the world's first wireless packet switched data network.(1)

During the development of ALOHAnet, Abramson and his team had to solve many foundational issues with wireless networking, including collisions and failed transmissions. They decided that a random access protocol would be the best way to accomplish this. The ALOHA protocol allowed terminals to send data whenever they had it, and listen to their transmissions to make sure they were received, and acknowledged. If a transmission failed, terminals would resend after a random delay. This was simple and effective, but had problems scaling up to more users. The ALOHA protocol was improved and expanded, and was the inspiration for CSMA/CA.

ALOHAnet was a pioneering technological marvel, but it was experimental and not fit for commercial use. Most critically, unlicensed spectrum for computer communications was not available to the public yet. Since the University of Hawaii didn't have a way to sell the technology, it was put into the public domain. ALOHAnet was decommissioned in 1976, but it inspired others to continue what it started. Most notably, Robert "Bob" Metcalfe studied it for his PhD dissertation.

Through the 1970s, many wired local area network technologies were developed. Some were proprietary, and some were open standards. Most of this early work has gone on to become ANSI or IEEE standards, and formed the basis of the computer communication and the Internet. A standards-based approach wasn't destiny, and we all benefit from the pioneers who pushed for standards over proprietary solutions. All of this groundbreaking work on wired standards helped lay the foundation for wireless.

Key Dates In Wired Networking History

1973: Bob Metcalf and David Boggs created the basis for Ethernet while working at the legendary Xerox Palo Alto Research Center in California. They were inspired by the packet-based approach of ARPAnet and ALOHAnet, and refined that design. On November 11, 1973, the first Ethernet system was functional.

1974: Cambridge Ring was developed at the University of Cambridge, forming the basis of what became Token Ring. Through the 1970s Werner Bux, Hans Müller, and other IBM employees worked on the design of what became IBM's proprietary Token Ring.
1974: Vint Cerf and Bob Kahn publish "A Protocol for Packet Network Intercommunication" which describes TCP and lays the groundwork for TCP/IP.
1975: Xerox files for a patent for Ethernet.
1978: Jonathan B. Postel writes the first specification of IPv4.
1979: Xerox opts to make Ethernet an open standard to help with manufacturing, and to spur sales of it's products. DEC and Intel join the effort.
1980: The IEEE 802.3 committee is formed to make Ethernet an open standard, led by Xerox, DEC, Intel, and other major manufacturers.
1980: Engineer Michael Marcus makes a proposal to the U.S. FCC to open the ISM bands for unlicensed use of spread spectrum modulation and multiple access, going against conventional wisdom at the time. The ISM bands are assumed to have no practical application in communications.
1983: The first IEEE 802.3 Ethernet standard, 10BASE5, was ratified. Wi-Fi uses multiple parts of the IEEE 802 protocol family, and is designed to interwork seamlessly with Ethernet, its wired sibling.
1985: IBM introduced their proprietary version of Token Ring, offering 4 Mbps. Due to the integration with IBM computers, Token Ring was popular for a time in the 1980's, until Ethernet grew into the ubiquitous standard it is today.
1985: the U.S. Federal Communications Commission opened the ISM (Industrial, Science, and Medicine) radio bands of the wireless spectrum for use in communications without a government license. The released frequencies include: 900 MHz, 2.4GHz and 5.8GHz. Other countries and regions eventually followed the FCC's lead. With this spectrum available for anyone to use, the real development of the first commercial wireless network technologies could begin.

The Wright Brothers of Wi-Fi

By 1985, all the pieces were in place for the development of a universal wireless networking standard. These were the Wild West days of wireless networking. Dozens of companies were competing to be the first to produce a successful product. Many companies came and went, so I'm only going to focus on the biggest and most important ones.

NCR and WaveLAN

In the 1980s, NCR Corporation was a big international company selling computers, ATMs, and cash registers. They wanted to develop a way to have their cash registers work wirelessly. This would give them a competitive advantage, and allow retail stores to avoid the costs and logistics of running cabling to every register. At the same time, cash registers were becoming more computer-like, so NCR sought to develop an International wireless standard that any computer could use. Two of the key engineers assigned to this project were Cees Links and Vic Hayes. Through 1986 and 1987, Vic and the other engineers at NCR developed a product which became known as WaveLAN.

NCR introduced WaveLAN to the world in 1990.(2) While WaveLAN was intended for use in cash registers, it was also positioned as a wireless alternative to Ethernet and Token Ring for PCs. In the 1990s, wireless products were expensive, and hard to use. The early drivers were buggy, and documentation was hard to find. WaveLAN network adapters cost hundreds of dollars, and access points could cost thousands. The high cost and limited usability kept WaveLAN as a niche product, mainly found in the big businesses that could afford them.

WaveLAN operated at 900MHz or 2.4 GHz, and offered speeds of 1 to 2 Mbps. It competed with Aironet's ARLAN and others which offered similar speeds, frequency ranges and hardware. Several companies also marketed wireless bridges and routers based on WaveLAN technology. Wireless networking products became a mess of competing companies and different implementations. The need for an International standard was clear.

NCR wanted to have Vic Hayes make the technical proposals to the IEEE on the companies behalf. Vic disagreed, and fought to stay independent and not do any of the proposals on behalf of NCR. The company agreed, and this actually worked to their benefit. Since Vic wasn't pushing NCR's proprietary solution, he gained the trust of others. Vic ended up becoming the first chairman of the IEEE 802.11 working group, giving him the nickname “Father of Wi-Fi"

IEEE and the 802.11 working group

In 1988, Vic Hayes approached the IEEE to contribute to the 802.4L, an existing committee creating a wireless standard based on Token Ring. As it turned out, that committee had become inactive, and the chairman had quit. This let to Vic becoming the chairman of 802.4L, and the forming of a new working group, 802.11. Ethernet and WaveLAN technology formed part of the technical base of the 802.11 committee, which was officially started in 1990.

WaveLAN and the knowledge of Vic Hayes' group at NCR helped form the basis, but countless others contributed as well. Many other companies submitted proposals for various aspects of the standard, and the group compromised and voted on the best ideas along the way. Most of their documents are still publicly available. The 802.11 group continued to work throughout the 1990s to develop the standard, which was a long and complicated process. It took 7 years and input from countless engineers to form the first Wi-Fi standard, known as 802.11-1997.

The first Wi-Fi standard was in some ways a messy compromise. It allowed for transmissions over Infrared or 2.4 GHz radio waves. It also allowed for two different frequency selection methods, which prevented guaranteed compatibility. In the end, both Frequency Hopping Spread Spectrum (FHSS) and Direct-Sequence Spread Spectrum (DSSS) were added to the standard. The group allowed these options as a way to please the entities involved in the creation, but it made the standard weak.

802.11-1997 never gained much traction. Despite that, the group got to work on its next set of standards, 802.11a and 802.11b, which were developed in parallel. The main difference between them was the frequencies and the modulation schemes they used.

802.11a operates in 5 GHz, using OFDM for data rates up to 54 Mbps.

802.11b operates in 2.4 GHz, using DSSS for data rates up to 11 Mbps.

802.11b hit the market first. Despite the slower speeds, people preferred the availability, lower cost, and longer range of the 2.4 GHz equipment. The success of 802.11b snowballed into the development of 802.11g, and made 2.4 GHz the defacto International wireless spectrum range for years to come.

In the late 1990s, Wi-Fi's success wasn't guaranteed. Early Wi-Fi devices were expensive, slow, and often incompatible with other devices. According to Vic Hayes, he approached many computers makers to persuade them to add Wi-Fi to their computers, but most of them said no. Steve Jobs and Apple wanted to add Wi-Fi to an upcoming laptop, but they wanted it to cost $100, not $500. They argued that the lower cost would make it more marketable to consumers.

Apple eventually got the $100 Wi-Fi card they wanted. The iBook and AirPort launched on July 21, 1999, a few months before the 802.11a and 802.11b standard were finalized. It was the first mainstream computer designed and sold with integrated wireless networking.

OFDM, CSIRO, and Dr. John O'Sullivan

This is where our story get murky. CSIRO is the Australian federal agency with a long history of scientific research. They have developed many advanced technologies, such as atomic absorption spectroscopy. According to CSIRO, Wi-Fi is an Australian invention. The truth isn't that simple, and it isn't that clear.

CSIRO's key insight came out of their work in radioastronomy. In 1977, CSIRO researcher Dr. John O'sullivan was searching for small exploding black holes. He wrote a paper about how fast Fourier transformations could be used to sharpen images from optical telescopes. As the waves traveled towards the telescope they became scrambled, and CSIRO developed a custom processor to use fast Fourier transformations to unscramble the image. While this technique didn't help him discover the black holes he was looking for, it helped to solve a different problem.

John O’Sullivan and other CSIRO scientists in their lab.

Starting in 1990, John O'Sullivan led a team of CSIRO scientists to develop a high-speed wireless network. Their goal was 100 Mbps, to compete with the best wired networks of the time. This caused them to look past the techniques used by ALOHAnet and WaveLAN, because they wouldn't be fast enough. One technical challenge they faced was how radio waves tend to bounce off surfaces indoors, causing an echo that distorts the signal. Using fast Fourier transformations, they found a way to transmit a signal while reducing that echo. Rather than using one fast wireless channel, they used lots of slower channels. This is known as multicarrier modulation.

Single-carrier modulation systems exploit only one signal frequency to transmit data. Multi-carrier modulation systems divide the whole frequency channel into subcarriers, and the data stream is divided into many low-rate streams transmitted in parallel. Single-carrier modulation has advantages, but multi-carrier modulation is better suited for high-bandwidth, short-range communication like Wi-Fi. In modern Wi-Fi standards, we use multi-carrier modulation in the form of orthogonal frequency division multiplexing, or OFDM.

According to their lawyer in a 2009 patent dispute, CSIRO did not invent Wi-Fi, but it did invent the best way of doing it. CSIRO scientists tested hundreds of techniques until it found a unique combination that worked at high speeds. That combination involved multi-carrier modulation, forward error-correcting, and frequency interleaving to send multiple copies of the data. This specific combination of techniques is the secret sauce the CSIRO applied for a patent for in 1992, and was granted in 1996.

CSIRO didn't invent the underlying techniques, since they were already known. Frequency division multiplexing dates back to the late 1870s, when telegraph companies were trying to increase their capacity. Multi-carrier modulation was described in the 1950s. OFDM is attributed to Bell Labs employee Robert W. Chang back in 1966. Interleaving also dates back to the 1960s. Forward error correction was used when NASA sent the Mariner mission to Mars in 1968.

CSIRO also wasn't the first to combine these techniques and apply them to computer communication. Harris Semiconductor sold modems to the US military using these techniques in the 1980s. Critically, the Harris modem was never patented since interleaving, modulation, and coding had been around for decades by the time Harris came along. According to testimony, Intel official Stephen Saltzman said that outside CSIRO, engineers didn't take the idea of a Wi-Fi patent using these techniques seriously. Experts at Intel had already shipped products based on OFDM with previous employers, and thought that a new patent with OFDM didn't seem credible. Regardless of the merits and the origin of the ideas, US patent #5,487,069 was granted in 1996.

During the development of the 802.11a standard, CSIRO made the IEEE aware of its patent and offered to license it. Beyond that offer, CSIRO never got involved in the creation of the 802.11 standards. Dozens of companies made proposals to the various 802.11 committees, offering to submit their ideas or license their patents. CSIRO never submitted a proposal for the original 1997 standard, or any of the revisions. Despite their lack of involvement in the process and lack of licensing agreement, 802.11a, 802.11g, and later IEEE standards would go on to use OFDM and multi-carrier modulation.

Years later, CSIRO used this as the basis for lawsuits against major networking and technology companies. CSIRO won settlements worth $205 million in 2009, and another $229 million in 2012. If you want more detail on the CSIRO patent fight, Joe Mullin wrote a good piece for ArsTechnica back in 2012. Mark Summerfield wrote a counter-point to Joe's article for Patentology, looking at what the CSIRO patent did and didn't cover. The truth is a messy thing. In this case, success truly did have many fathers.

802.11 Timeline

1986: The 802.4L working group begins investigating a wireless standard based on Token Ring.

1990: The 802.11 committee was formed and NCR employee Vic Hayes became the first chairman.
1993: AT&T made the first large-scale deployment of WaveLAN at Cernegie Mellon university. Dr. Alex Hills then started a wireless research initiative to provide coverage to 7 buildings on campus. This was an important project for researching wireless networking and its scalability.
1996: Australia’s CSIRO patented a technique for reducing multi-path interference of radio signals transmitted for computer networking. This technique finds its way into the 802.11a standard, fueling future patent disputes.
June 1997: the first version of the 802.11 standard is ratified. This first version used similar techniques to WaveLAN, offering up to 2 Mbps of speed and optional interoperability.
1998: Equipment using the original 802.11 standard is rare and expensive, and no major consumer products adopt it.
July 1999: Apple introduced the iBook and AirPort, the first major consumer products to use the 802.11b standard.
1999: The Wireless Ethernet Compatibility Alliance (WECA) is formed to certify that products comply with the 802.11 standards.
September 1999: The 802.11a and 802.11b standard are ratified.
- 802.11a used OFDM for 54 Mbps link rates over 5 GHz.
- 802.11b used DHSS for 11 Mbps link rates over 2.4 GHz.
Late 1999: The WECA re-brands 802.11 as Wi-Fi, certifies it's first products, and becomes known as the Wi-Fi Alliance.
Early 2000s: 2.4 GHz equipment using 802.11b was adopted more broadly, and 5GHz 802.11a falls behind in popularity. This cascaded into the next standard, 802.11g, which opted to focus on 2.4 GHz to maintain compatibility 802.11b devices.
2003: 802.11g is ratified, applying 64-QAM modulation and OFDM to 2.4 GHz for link rates up to 54 Mbps.
2009: 802.11n is ratified, introducing MIMO, frame aggregation, up to 4 spatial streams, 40 MHz channels, optional 5 GHz support, and data rates up to a hypothetical 600 Mbps.
-Most 802.11n devices top out at 2 spatial streams, due to 3 or 4 spatial streams being unreliable with the 802.11n standard.
2013: 802.11ac is ratified, and broken into two waves. Wave 1 certification begins.
- Wave 1 certified devices add 80 MHz channels, up to 3 spatial streams, 256-QAM modulation, and explicit beamforming, for data rates up to 1.3 Gbps over 5 GHz.
- Wave 2 certified devices add down-link MU-MIMO, optional 160 MHz channels and a 4th spatial stream for data rates up to 2.34 Gbps over 5 GHz.
- 802.11ac devices use 802.11n over 2.4 GHz for backwards compatibility.
2014: Wi-Fi device shipments reach 10 billion.
2016: 802.11ac Wave 2 certifications begin.
2017: Wi-Fi device shipments reach 20 billion.
2017: The first draft of the 802.11ax standard is published, and manufacturers begin to make devices based on the draft specification.
- 802.11ax adds OFDMA, 1024-QAM modulation, up to 8 spatial streams and bi-directional MU-MIMO, for data rates up to 1200 Mbps over a single spatial stream in a 160 MHz channel.
2018: The Wi-Fi alliance renames the existing Wi-Fi standards.
- 802.11b = Wi-Fi 1
- 802.11a = Wi-Fi 2
- 802.11g = Wi-Fi 3
- 802.11n = Wi-Fi 4
- 802.11ac = Wi-Fi 5
- 802.11ax = Wi-Fi 6
2019: Wi-Fi device shipments reach 30 billion.
April 2020: The US FCC votes to release 6 GHz spectrum for unlicensed use, giving 1200 MHz of additional spectrum for use in wireless networks.
November 2020: Wi-Fi 6 is expected to be officially ratified, and work on Wi-Fi 7 (802.11be) is underway.

Footnotes:

1 - ALOHAnet used 100 KHz channels, operating at 9600 baud on the 407 and 413 MHz UHF frequencies. It originally connected seven computers across four islands, which communicated wirelessly with a central computer in a star network topology.

2 - AT&T would go on to buy NCR in 1991, only to spin off the remnants of the division as NCR and Lucent by 1996.

Wi-Fi 101 — Wi-Fi Origins and History

Wi-Fi 101 — Wi-Fi Origins and History

Where Did Wi-Fi Come From?