Wi-Fi taking advantage of more parts of the spectrum
6 mins read
It seems the world is unable to function without wireless communications, whatever the format. Yet only a few years ago, wireless communications was a strange world inhabited by very few people. Things have changed dramatically and the change could be put down to one person.
In the mid 1980s, companies were slowly getting to grips with the concept of Ethernet. This networking technology, which first saw the light of day at the beginning of the decade, brought the ability for personal computers to pass data amongst each other at rates of up to 10Mbit/s. Yet, by the end of the 1980s, companies were looking to develop wireless communications.
The first applications for the technology which was to become IEEE802.11, or Wi-Fi, were not aimed at the business environment. Instead, developers were looking to create technology which addressed point of sale systems and the like.
In its early days, the technology wasn't competitive with Ethernet: data rates were, at the most, 2Mbit/s and few applications were seen. Nevertheless, the technology developed and the first Wi-Fi standard appeared in 1997 under the guise of IEEE802.11. Still supporting no more than 2Mbit/s, the approach had three possible physical layers: infrared; frequency hopping spread spectrum; and direct sequence spread spectrum. Data was transmitted in the ISM band at 2.4GHz; something which still holds today. Earlier, there had been trials at 915MHz but, according to Richard Edgar, senior Wi-Fi product manager with CSR: "It was a very small spectrum and so it didn't take off."
Edgar has been involved in the Wi-Fi market since 1998. "Wi-Fi was really developed to offer wireless Ethernet connection," he recalled, "and people wanted wireless networking for their computers."
This origin can be seen in the construction of the Wi-Fi standard. "Basically, it takes an 802.3 Ethernet frame and wraps an 802.11 frame around it to transport the data," he noted. "At the other end, the 802.11 frame is removed. It was an approach that brought users flexibility and freedom."
But one event – more accurately, one person – provided the impetus that has changed Wi-Fi from being something of a curiosity to its current status in which it is becoming pervasive. That person was Steve Jobs.
Jobs had seen Wi-Fi technology developing and had decided that it should be integrated into a forthcoming product; the iMac, Apple's 'all in one' desktop pc.
"The iMac took Wi-Fi to the mass market," Edgar noted. "The next thing you knew, all the other computer companies were putting Wi-Fi into their products; initially as an add on, but later as a standard offering. And the market has grown from there."
And what a market growth. Demand for Wi-Fi products in 1998 was worth around $2million; by 2001, that had risen to $350m and the industry hasn't looked back.
"Because Wi-Fi 'crossed the chasm'," Edgar commented, "it attracted new entrants; for example, Broadcom with 802.11g. But, while that brought more competition, Wi-Fi remained a pc based technology." And that remained the case until around 2006, when the first Wi-Fi enabled phones began to appear. "Now," said Edgar, "it's everywhere. The world believes in Wi-Fi."
The technology has made what seems to be linear progress. "Inasmuch as 802.11a was developed before 802.11b, there is linearity," Edgar agreed. "However, data rates have grown from 2 to 11 to 54 to 600Mbit/s."
But the developments have not come without their problems. "It took too long to develop 802.11n," Edgar believed. "There were two industry groups and neither could get the 75% vote needed to get their technology ratified. In the end, the groups agreed to work together and most of their options got included in the spec. For instance, 802.11n offers four different methods of beam forming, yet 11n beam forming hasn't really worked in the field."
The work done to date is about to be 'wrapped up' into a new baseline standard. "The IEEE does these 'wrap ups' from time to time," Edgar explained. "The current baseline is IEEE802.11-2007 and the new one will be 802.11-2012, with all current ratified standards included." Reading this document will not be a task for the fainthearted: the publication is likely to run to 2500pages, or a 16Mbyte PDF.
It's where Wi-Fi goes from that point that holds interest; not only in the new variants that are being considered, but also in the competition that may appear from other areas, such as WAPI; the WLAN Authentication and Privacy Infrastructure developed in China.
One of the top line trends is the move away from 2.4GHz in favour of 5GHz and then 60GHz. "There's too much interference at 2.4GHz and, quite simply, there's not enough spectrum," Edgar noted. But the technology is also looking to move into lower frequencies, including the whitespace part of the spectrum – broadly 600 to 800MHz.
Databasing will be central to the use of whitespace (see 802.11af box), but this approach might be extended to the 2.4GHz band. "If that proves successful," Edgar pondered, "it might then get applied at 5GHz."
These moves are encompassed in the development of four new physical layers, or PHYs. "11ac will run at 5GHz and is scheduled for ratification in 2013," Edgar said. "11ad, coming in 2012, will run at 60GHz and is being developed by the Wireless Gigabit (WiGig) Alliance. 11af, planned for 2013, will use whitespace, while 11ah, likely to appear in November 2014, will operate at frequencies of less than 1GHz and find application in smart grid and M2M."
In Edgar's view: "There's a sea change in Wi-Fi; it's exploding into new markets. People want things to work faster and if there's spectrum available, Wi-Fi should have it." The mantra appears to be 'one MAC, many PHYs'.
As part of this 'explosion', Wi-Fi is moving away from its traditional home in the pc and networking industry. "It's becoming an all encompassing technology," Edgar ventured, "and is starting to take share in markets that were seen as complementary – even Bluetooth."
The consequence is that design will become harder. "At 60GHz, cmos design is quite tricky and one of the challenges for silicon companies is that they're getting to the point where shrinking the rf part isn't economic any longer. It's unlikely we'll see the rf element shrink beyond 45nm, but we'll still want the digits to shrink," he said.
It's therefore likely that Wi-Fi chips will become SoCs. "But if these are done properly, with the peripherals integrated, then this will simplify product design," he concluded.
IEEE802.11ac
This will offer 80 and 160MHz bandwidth options, building on the 40MHz available in 802.11n. "It will support eight antennas," said Edgar, "as well as multiuser MIMO." With MU-MIMO, multiple stations with multiple antennas can transmit and receive simultaneously.
The standard will only be available at 5GHz as there is insufficient bandwidth at 2.4GHz. "While 5GHz has lots of channels, the advantages begin to disappear as bandwidth increases," he noted. 802.11ac will use a 64QAM constellation as standard, with 256QAM as an option. "But the error vector magnitude with 256QAM will be -32dB," he added, "so silicon developers will have to work hard to keep the noise inside the chip down to as little as possible."
Maximum data rate is 6.933Gbit/s, but one channel in a 64QAM system will sustain 325Mbit/s. "Most smartphones can't cope with that," Edgar said.
Contiguous and non contiguous channels are also enabled. "At 5GHz, there are things like radar and weather satellites," Edgar pointed out. "So if a radar is operating, 11ac can't work in that part of the spectrum. By splitting 160MHz into two 80MHz bandwidths, 11ac transmissions can have high data rates while avoiding radars." Having said that, Edgar added there were questions about whether 160MHz channels would be deployed.
Potential applications include whole home networking, where a lot of interest is being shown by consumer electronics companies. "Smartphones, networking and pcs are transitioning to 11ac," Edgar noted, "and we may see laptops on the market in time for Christmas 2012."
IEEE802.11ad
802.11ad will operate at 60GHz, which will pose technical problems. "At 60GHz, radio waves hate oxygen," Edgar pointed out. "Energy gets absorbed, so the challenge will be for antennas to become more focused."
11ad is likely to support a personal basic service set mode. "This can be seen as a personal area network," Edgar explained. "Because 11ad will essentially be short range, this will be good for personal communications as there's no interference."
802.11ad may also enhance Wi-Fi Direct, the technique that allows Wi-Fi devices to talk to each other without the need for hot spots.
Using OFDM, the draft standard is expected to support 6.8Gbit/s. "But you'll need 32 antennas to get the highest speeds," Edgar warned, "so beam forming is a 'must'." However, gain increases with antenna area, which supports higher powered links.
Consumer electronics is a potential user. "I'm sceptical about 60Hz being good for video transmission," Edgar continued. "Previous attempts have been made at this frequency and I'm not sure how Wi-Fi will be different. While it could be good for linking a laptop to a monitor, it won't be a whole house application."
Other applications could include high speed file download and networking. "If it works, you will download movies in seconds," he said.
It's likely an 11ad radio will have to be a system in package, with the antennas on top of the chip.
IEEE802.11af
Targeting those parts of the spectrum vacated by analogue tv transmission, 802.11af will offer longer range communications. The concept is reliant upon a central database to allocate channels. "An access point (AP) has to negotiate with that database," Edgar explained. "The AP sends its coordinates – including Z, because that's important – and the database sends a list of available channels. The AP then assigns a channel to a device."
In the UK, Ofcom has added a time element, restricting access to certain channels at particular times to protect wireless microphones used at concerts.
"802.11af may also be used to offload mobile phone traffic," Edgar suggested. "Whitespace is being looked at as a way to do this because of the range it offers. Along with support smart grid applications and sensors, 11af may also enable rural broadband and it could have application in interactive tour guides."
IEEE802.11ah
Operating at frequencies of 1GHz or less, 802.11ah will have a transmission range of up to 1km and could support up to 6000 associated stations. "There's a lot of license exempt spectrum available," Edgar asserted, "and this may be an area which suits software defined radio."
The draft standard is likely to prove attractive to companies developing technology aimed at smart grid applications. "The problem is there are more than 100 potential markets," Edgar said, "and each isn't very large. But, together they're big. Developing silicon to serve these markets will not be easy, neither will software development."
Although capable of supporting data rates of up to 600Mbit/s, 802.11ah applications may actually be handling something around 100kbit/s.