WirelessBr é um site brasileiro, independente, sem vínculos com empresas ou organizações, sem finalidade comercial, feito por voluntários, para divulgação de tecnologia em telecomunicações
802.11n or UWB?
Autores: Hiroki Yomogita and Tetsuo Nozawa
Origem: http://neasia.nikkeibp.com/neasia/002271. Visite!
Esta página possui 5 figuras grandes. Aguarde a carga se a conexão estive lenta.
802.11n or UWB?
Two major candidates are competing to become the wireless interface destined to feature in home digital equipment: IEEE802.11n, and Ultra-WideBand (UWB).
The day is coming fast when home digital equipment will have wireless interfaces, implementing wireless transmission of high-definition television (HDTV) imagery and high-speed swapping of still pictures and audio content with portable gear.
There are two major candidates for the wireless
interface: 802.11n, a next-generation wireless local area network (LAN) built
on spatial multiplexing using multiple-input, multiple-output (MIMO)
technology, and Ultra-WideBand (UWB) wireless technology using ultra-wideband
technologies such as Wireless Universal Serial Bus (USB).
11n offers a long range of up to 200m, and is viewed as the most likely contender for the home network backbone.
UWB, on the other hand, is likely to make best use of its low-power, high-speed operation in short-range equipment interconnect, such as personal computers (PC) and portable equipment.
At present both 11n and UWB are being supported
as industry standards by multiple groups in competition with each other, with
no clear victor in sight.
In 11n, the TGn Sync industry body primarily composed of home appliance manufacturers is in collision with the World-Wide Spectrum Efficiency (WWiSE) group, comprised of mostly wireless LAN chip manufacturers, wireless LAN vendors and similar firms.
In UWB the situation is similar, with the DS-UWB scheme based on direct-sequence spread-spectrum technology competing with Wireless USB using multiband orthogonal frequency division multiplexing (OFDM) for the position of "standard technology".
Equipment manufacturers have assumed the technologies would continue to coexist, watching and waiting to see if one falls by the wayside.
Recently, however, the boundary between 11n and
UWB has begun to blur rapidly.
As a result of technological refinements to 11n during the competition to become the industry standard, new concepts have been incorporated to provide significant improvements in speed and reduced power consumption. At last 11n technology is approaching the realm of portable equipment application, which until now has been thought to be the domain of UWB.
Mobile 11n Appears
The range of 11n application is steadily
expanding, covering everything from home backbones to small-scale equipment
interconnect (Fig 1). As a result, it seems likely to split into three specs
for different equipment types. In addition to one standard designed to pump
HDTV video streams to audio-visual (AV) home decks, and another for PC and
corporate communications needs, there will be a third for mobile equipment
such as mobile phones and digital cameras.
Fig 1- Toward 1 Gbit
From about 2006 a variety of
wireless interfaces attaining peak data rates over 100 Mbps will appear, and 1
Gbps will grow much closer. In wireless LAN the first IEEE802.11n products,
the next-generation standard, will appear from the second half of 2006.
They will probably start with a peak data rate of about 140 Mbps but rising thereafter to about 600 Mbps.
UWB technology using ultra-wide bandwidth has already achieved a data rate of 100 Mbps or higher with the DS-UWB chipset, but this is scheduled to be bo
The spec for home-use AV equipment will offer a peak data transfer rate of 500Mbps to 600Mbps: high-speed performance thought impossible to achieve without UWB.
The positioning of UWB, on
the other hand, will change significantly depending on radio wave output
regulations. Under current US Federal Communications Commission (FCC) output
power regulations UWB will achieve its designed performance, but an expected
obstacle has cropped up.
The International Telecommunication Union - Radiocommunication Sector (ITU-R), which has been deliberating common spec items for UWB and other wireless services, prepared a draft recommendation in May 2005 setting a stiffer output power regulation on UWB.
The recommendation will not be adopted automatically by regulatory agencies around the world, but it is possible that some nations will impose restrictions on transmission range or other points.
At the same time, though,
there is also considerable development proceeding quietly on UWB that may
significantly improve its potential.
Designed to dramatically extend range with MIMO technology, it makes it possible to aim at not only equipment interconnect applications, but also inter-room connection.
Until now UWB has only offered a transmission range of about 10m at 100Mbps, or 3m at 480Mbps.
A US tech start-up, however, has developed a technology roughly quadrupling this, to about 40m at 100Mbps.
Now that the competition is surpassing the scale
of merely 11n vs UWB, equipment manufacturers will be forced to make a
decision, selecting the scheme which best fits the specific application and
11n is viewed as being superior in maintaining compatibility with existing wireless LAN, while UWB is generally thought more likely to achieve lower levels of power consumption.
According to Isao Ozasa,
application manager, USB Technical Marketing, PC Peripheral Systems Div, 2nd
Systems Operations Unit, NEC Corp of Japan, "In order to achieve 100Mbps with
11n, because the Internet protocol (IP) processing load is so heavy, we will
need a RISC microcontroller running at an operating frequency of about 200MHz.
With Wireless USB, the same thing would only need 30MHz to 50MHz." In other words, engineers will have to make their selection with consideration for application, peripheral circuit dissipation and other points.
The different methods used to implement transmission diversity by competing WWiSE and TGn Sync illustrates their differing technological approaches. WWiSE has chosen the open loop approach, while TGn Sync (with participation by manufacturers aiming at high-speed interconnect between TV receivers and media servers) goes for the closed loop approach (Fig 2).
Fig 2 - Efficiency or Simplicity?
The key point of contention
betweeen TGn Sync and WWiSE is whether or not to use the Tx BF that TGn Sync
has been pushing. WWiSE is stressing circuit
simplicity and prefers the open loop design, where the transmitter sends data
directly to the receiver (a). TGn Sync, on the
other hand, first uses Tx BF to send a "training" signal to the receiver,
which is analyzed and information on the
transmission path returned.
Based on the information, the transmitter selects the optimal output power and modulation scheme.
This is called a closed loop design because information from the receiver is fed back (b).
The frequency utilization efficiency improves significantly, and the circuitry is more complex, but efficiency can drop when the transmission changes significantly with high-speed motion.
Transmission diversity is a function that increases reliability by using more antennas at the transmitter to achieve longer range. In the open loop design the transmitter does not receive feedback from the receiver, such as distance information, and merely transmits. In the closed loop design, on the other hand, a "training" procedure is used before transmission to allow the transmitter and receiver to share information on channel signal/noise (S/N) ratio. The closed loop design can utilize power more efficiently, making it easier to boost the data rate.
TGn Sync has combined the closed loop approach with MIMO into what they call transmit beamforming (Tx BF). With Tx BF and 256-level quadrature amplitude modulation (QAM) modulation, the system can use four antennas each at transmitter and receiver to boost the peak data rate to 720Mbps for 40MHz channels.
Is Training Effective?
A number of people in WWiSE, however, have raised doubts about Tx BF. A manufacturer engineer taking part in the WWiSE effort explained: "If training is implemented on a per-packet basis, the effective data rate drops way off. The effect is totally unknown when training is used when in motion, because the transmission path state is constantly changing.
When combined with 4x4 MIMO, the processing load on the baseband circuit is considerable, and it can't keep up with MIMO processing." Others are demanding improvements in circuit design, saying it will be essential to design a circuit for antenna mapping, or selecting the modulation scheme and power appropriate for the specific transmission path, in the baseband processing circuit.
In response, one manufacturer engineer in the TGn Sync group said, "Whether or not the proposal can be implemented, including 256-level QAM and an ADC to handle it, is a technical problem for the manufacturers, and not something to be decided in the 11n spec."
MIMO Transceiver Circuit
When it comes to 11n, the hottest topic among semiconductor and component manufacturers right now is miniaturizing the MIMO transceiver circuit.
MIMO technology uses multiple antennas, making for a complex transceiver circuit configuration. Worse, 11n has been a dual-band proposal from the start, using both 2.4GHz and 5GHz, as well as offering a quad-mode spec compatible with 11a/b/g. There is some worry that this could result in an increased number of components needed in the radio frequency (RF) transceiver (Fig 3).
Fig 3- Implementation Problems with 802.11n
IEEE 802.11n faces a host of
problems of ICs, modules and more.
MIMO requires multiple antennas mounted on a uniform interval, presenting problems in module miniaturiztion and assuring antenna mounting space. In addition, two or more RF transceiver systems are needed , increasing power consuption, and the techinical expertise of chip and module manufacturers may have difficulty in implementing the Tx BF function and simplifying algorithm.
the photo shows the 2x3 MIMO CardBus card from Airgo Networks.
Airgo Networks, Inc of the US, for example, one of the first to introduce MIMO technology, cites the difficulties involved, as summarized by Eiji Takagi, director of Application Technologies: "A 2x3 MIMO transceiver circuit using our chipset just can't be made smaller than the size of about a PCMCIA card. At present, we can't squeeze it into a miniPCI Express card."
On the flip side, this represents an excellent opportunity for component manufacturers with experience in miniaturizing RF front-end circuits. The development of front-end modules (FEM) single-packaging a host of components essential in front ends, such as power amps, low-noise amps (LNA), bandpass filters and antenna switches is getting quite active.
The de facto Standard?
While 11n development seems to be headed for a unified standard, unification of the two UWB schemes at the IEEE802.15 working group seems to be stuck. The key factors are major differences in transmission methods, and long-standing strain between the two groups of manufacturers. Regardless of discussion at the Institute of Electrical and Electronics Engineers (IEEE), manufacturers involved in UWB want to ship chipsets and application products as soon as possible, and are advancing their agendas with the top priority set on becoming the de facto standard.
Both groups are striving to become the industry standard, but DS-UWB seems likely to take the lead in commercialization with the release of a chipset product. The Haier Group of China used a DS-UWB chipset from Freescale Semiconductor, Inc of the US in its liquid crystal display (LCD) TV released in the second half of 2005, for example. According to Freescale chief executive officer (CEO) Michael Mayer, "The reason is simple: we are the only firm offering a UWB chipset product." It is the only chipset with FCC licensing.
Integration of DS-UWB chipsets is also advancing (Fig 4). Freescale, for example, has used low-temperature co-fired ceramic (LTCC) technology to pack the RF transceiver integrated circuit (IC), baseband processor IC, and media access control (MAC) IC into a single package (the so-called system-in-package, or SiP) product about 15mm-square.
Fig 4 - DS-UWB takes the lead in miniaturization
DS-UWB began shipping chip samples from about, and there are already compact modules using the chipset.
Freescale Semiconductors, for example, has already prototyped miniPCI and USM adapter modules.
DS-UWB features a simple transmitter circuit, simplifying the overall circuit configuration. It does not use OFDM modulation/demodulation, which helpes make it number one in low-power operation
Chips to Hit 1.6Gbps
At the end of 2005 Freescale will also ship the XS660 chipset, boosting the data rate by 6x from 110Mbps to 660Mbps, and a chipset for mobile equipment retaining the current 110Mbps data rate but slashing dissipation to only about 100mW.
In 2006 the company plans to release an ultra-high speed UWB chipset hitting a peak data rate of 1.6Gbps. Until now products have only used frequencies between 3.1GHz and 5GHz, but the high-speed variety will instead use the waveband from 6GHz to 10GHz. "We will implement Wireless HDMI, swapping HDMI signals between TV receivers, media servers and other equipment," explained Martin Rofheart, director, Ultra-Wideband Operations at Freescale.
The last remaining problem for the DS-UWB group is to attract a larger group of supporters, which is why they are beefing up their efforts to be adopted as the physical layer for Wireless 1394, the wireless version of IEEE1394, high-speed Bluetooth and other schemes slated to be finalized in around 2006.
The Wireless USB group has already succeeded in building up a strong group of supporters, establishing the Wireless USB standard as expected, incorporating the USB-IF group (a USB standardization organization) under the lead of Intel Corp of the US. They have completed all the preparations needed to become the industry standard, such as picking up organizations defining physical and MAC layers and the WiMedia industry group for interconnectability assurance and logomark issuance. As with DS-UWB, they also hope to be tapped for use as the physical layer in Wireless 1394 and Bluetooth. If the multiband OFDM defined by WiMedia is used for this spec, it will be possible for the physical and MAC layers to be completely WiMedia via multiple interfaces including Bluetooth and Wireless USB.
Intel plans to build Wireless USB into notebook PCs from about 2006, and other manufacturers in the field believe the peripheral equipment market will take off from there.
Airgo Networks: www.airgonetworks.com
TGn Sync: www.tgnsync.org
Wireless USB: www.usb.org
(October 2005 Issue, Nikkei Electronics Asia)