Networking

Can Wi-Fi keep up with today's applications?

Bandwidth-intensive applications are becoming common place, causing 802.11 devices to be overwhelmed. Is there a solution on the horizon? Michael Kassner reports on the latest technology.

Applications requiring significant bandwidth like streaming media, on-line gaming, and VoIP do not always render well using Wi-Fi connections. Why is that? Internet access should be the bottleneck. It has far less available bandwidth than an 802.11a/g connection. Or does it?

What affects bandwidth

Shoving data through a cable or fiber is simple compared to transmitting it over the ether. Not being confined also makes it difficult to get a handle on a Wi-Fi connection's available bandwidth. The following are some of the reasons why:

  • Wi-Fi transmissions require considerably more management traffic compared to a wired connection, which reduces the bandwidth available for data.
  • Unlike wired Ethernet links, duplex operation (being able to send traffic to both devices simultaneously, doubling bandwidth) is not currently possible with Wi-Fi. Think of two people talking at the same time; neither hears the other.
  • Receiving signal strength plays a huge part in determining the throughput data transfer rate, with more signal strength allowing a better rate. Wired connections are affected only by cable length restrictions.
802.11n helps

802.11n introduced two new ideas that help with the bandwidth problem. One is to use Multiple-Input/Multiple-Output (MIMO) technology. By its nature, radio signals fan out from the transmitting antenna, bumping into objects on the way to the receiver. Up until MIMO was developed, these out-of-sync radio waves would confuse the receiver; however, MIMO can process the different signals. James Wilson, an Intel engineer, explains in his DeviceForge.com article:

"MIMO technology offers the ability to coherently resolve information from multiple signal paths using spatially separated receive antennas. Multipath signals are the reflected signals arriving at the receiver some time after the original or line of sight (LOS) signal has been received.

Multipath is typically perceived as interference degrading a receiver's ability to recover the intelligent information. MIMO enables the opportunity to spatially resolve multipath signals, providing diversity gain that contributes to a receiver's ability to recover the intelligent information."

Increase channel size

802.11n also introduced another important feature, larger channels. Increasing channel bandwidth from 20 MHz to 40 MHz more than doubles the usable bandwidth. The following slide (courtesy of Intel Labs) illustrates that:

Add in MIMO, and 802.11n substantially increases data transfer rates, as can be seen in the following table (courtesy of Intel Labs):

You may be wondering what the two different measurements mean. Here are their definitions:

  • Over-the Air Estimate: Under perfect conditions, it is the amount of data that can be sent from one Wi-Fi radio to another. This is the familiar 54 Mb/s advertised by the 802.11a/g standard.
  • MAC Service Access Point Estimate: Is dependent on the physical environment. 802.11a/g under best conditions may have a transfer rate of 22-25 Mb/s. The transfer rate could be low as 1-2 Mb/s in less than optimal situations.
Multiple streams

Moving to 802.11n equipment will help considerably if your current system is bandwidth-starved. But, is it enough? Wired technology is rapidly moving from 100 Mb/s Ethernet to Gigabit Ethernet. So, 802.11n with its 100Mb/s limit is still lagging behind.

The next big innovation in Wi-Fi is using multiple data streams. You may have seen the designation 3 x3: 3 (a x b: c). This notation helps identify the device's capabilities. The first number (a) represents the maximum number of transmit antennas. The second number (b) is the maximum number of receive antennas. Finally, the third number (c) is the maximum number of data streams the radio can process.

The maximum number of data streams allowed by the 802.11n standard is 4 x 4: 4. A device capable of this designation will have an Over-the-Air Estimate bandwidth of 600 Mb/s.

Starting production

Qualcomm introduced a 4 x 4: 4 chip this past June. Right after 802.11n was ratified, Atheros announced it will have a 3 x 3: 3 chip ready in early 2010. Devices using the Atheros chip will have an Over-the-Air Estimate bandwidth of 450 Mb/s.

Apple is not saying so, but it appears their upgrade to the AirPort Extreme Base Station is a 3 x 3: 3 device. Glenn Fleishman describes what this means in his Ars Technica article:

"The company says only that the latest revision gives you up to 50 percent better Wi-Fi performance and up to 25 percent better range than its immediate predecessor. This conforms with 3x3 antenna arrays, which, even with a two-stream radio, carry data further at higher speeds."

The only problem is that Intel appears to be the only manufacturer currently offering 3 x3: 3 client network adapters.

Final thoughts

Wi-Fi devices that utilize the full potential of 802.11n should eliminate many of the bandwidth problems associated with legacy Wi-Fi devices. The new 802.11n devices will also improve range and reliability. It may be time to unplug the Ethernet cable.

About

Information is my field...Writing is my passion...Coupling the two is my mission.

42 comments
EricTam
EricTam

10Gbps Ethernet is already available and I the price is coming down and hopefully no too long, it will be part of the default desktop NIC card. Therefore, 802.11n will need to run something like 180,180:180. I am going to keep the cable.

seanferd
seanferd

who have questions about, or simply do not understand, their wireless routers. Clear, concise, and informative. For me, it was good to see all that information related in one place.

chanchoth
chanchoth

The maximum number of data streams allowed by the 802.11n standard is 4 x 4: 4. A device capable of this designation will have an Over-the-Air Estimate bandwidth of 600 Mb/s. Where do you get the 600 Mb/s from?

wkankovsky
wkankovsky

Regarding the Capacity vs. SRN chart, comparing 40 Mhz to 20 Mhz, the capacity does not more than double. Rather it appears to increase about 70%, give or take. Or am I reading the chart wrong?

www.hansencc.net
www.hansencc.net

How would 3x3x3 or 4x for that matter work in long distance application where the access points were in a bridge mode vs. a straight access point? Is the 3x3x3 transmission method negated when using directional antennas in clear line of sight? And last, how would it work in a base location with an omni directional antenna(s) talking with multiple remote locations (each with a single directional antenna)?

Jimbo Jones
Jimbo Jones

As long as it operates in the unregulated spectrum, no wireless can be considered a reliable replacement for mission-critical apps. Would you build a hospital imaging system around an ethernet hub that someone could jam from a van parked on the street outside? Does even 802.11n have the bandwidth necessary for bandwidth intensive apps like workstation backups? Notice that vendors of controller/thin-AP solutions are now starting to migrate to a decentralized model where the data traffic no longer flows through the controller as it once did. This is inevitable. Do the math: Say the APs have 600 Mbps OTA. That translates to what? 300 Mbps real-world? Now if you have just ten of these tunneled to your controller, the uplink to your wired network will need to be 3 gig, and any internal firewall would have to keep up with that wire rate. Expect to see firewalls and ACLs off-loaded into the APs. An elephant in the room that nobody seems to want to talk about is the fact that 802.11n is a moving target. n today is not the same as n tomorrow. If I buy 2x2 or 2x3 access points today, then even when 3x3 or 4x4 clients become available the bottleneck will still be my APs, and I'll have to go through another expensive buying cycle. And what about the power to run all those radios? If I buy 802.3af PoE switches today, I'll have to upgrade again to 802.3at switches when I start deploying those power-hungry 4x4s. Better and cheaper to get managed multiport midspans or powered patch panels. At least they won't be as expensive or disruptive to replace.

Kris.J
Kris.J

A wireless LAN should never be considered a replacement for - only a complement to - a wired LAN.

veli-pekka.ketonen
veli-pekka.ketonen

YES - WI-FI CAN Think of mobile networks, they have a cell radius of several hundred meters to tens of kilometers. One base station has maximum data capacity from 1 Mbit/s to 15 Mbit/s (note: HSPA, basic 3G much less) between mobile terminal and base station. Latencies are around 100-200ms. While this is a decent number after years with 600ms latency, it is still limiting application performance. Base Station to Radio Network Controller transmission connections often are only few multiples of 1.5Mbit/s or 2 Mbits/s and thus limit the performance significantly from the above mentioned aggregate throughputs. Practical aggregte throughput per base station is a few Mbit/s. 802.11/a/g already delivers 15 Mbit/s practical throughput with a latency of about 4 ms within a radius of tens of meters. 802.11n adds more maximum througput to this and coverage, if mobile devices one day have MIMO support. Important issue here is that curretly lacking user experience in Wi-Fi networks is not due to lack of maximum supported throughput or latency. It is about how this capacity is used in practice. Wireless networks and their performance need to be continuously managed and optimized. Unfortunately WiFi networks are typically not managed at all. Neither have the enteprise grade WLAN equipment proper capability for performance and service level management. After fixing the AP's to ceiling, network is just left alone and expected to deliver 100Mbit/s throughput for years with 99.9999% availability. Unfortunately this does not happen. You should look for proper WLAN performance management solutions, if you want to keep your network running and users happy. As the founder of 7signal, I'm proud to say that with our 7signal Sapphire solution, YES WI-FI CAN!

Mycah Mason
Mycah Mason

Michael, Nice articel, thanks. Do you know if there are currently any of these newer N-devices out there? (Even in pre-production etc?) Can you provide a link to a plain English article that gives some more info about the newer implementation of N? Thanks!

Michael Kassner
Michael Kassner

I have a particular interest in Wi-Fi as having been a ham for almost 45 years now. Edit: Spelling

Michael Kassner
Michael Kassner

I have done that using wired connections, but not Wi-Fi. They must have a GUI that allows you to authenticate.

Michael Kassner
Michael Kassner

Sorry, I did not have the time to go into all the details. Like most technology, a newer version is going to be better. Regardless of MIMO, 802.11n has better receivers and transmitters. I have setup several 8-10 kilometer bridge links. using 802.11n devices and they work better than my 802.11g links.

Michael Kassner
Michael Kassner

I feel that 802.11n is completely capable of handling any of the situations you mention. Wi-Fi is also very cost-effective simply due to needed as many Ethernet cables. Every client that I have converted to Wi-Fi has seen increased productivity and the employees become very fond of the new mobility.

foster.ryan
foster.ryan

You wouldn't put mission critical traffic, especially a hospital. It doesn't matter what frequency you are on or what you use to filter interference, someone can jam it.

rsimms
rsimms

I have found that I generally get better performance from 802.11g devices with high gain antennas than from the 802.11n devices with fixed antenna.

Michael Kassner
Michael Kassner

For the insight. I agree that many people have the wrong perception about Wi-Fi.

technocrat25
technocrat25

n today?s high-tech world connectivity to networks is extremely important for organizations to survive. Networking is possible of various machines like computers, laptops, cell phones etc. This term paper talks of Wi-Fi, the wireless technology that allows internet connection to be broadcasted through radio waves. Its purpose serves directly to the users looking for internet access devoid of any cords or wires. What is Wi-Fi? Wi-Fi is a wireless technology brand that is owned by Wi-Fi Alliance Wi-Fi Alliance is a consortium of separate and independent companies agreeing to a set of common interoperable products based on the family of IEEE 802.11 standards. Wi-Fi certifies products via a set of established test procedures to establish interoperability. Those manufacturers that are members of Wi-Fi Alliance whose products pass these interoperability tests can mark their products and product packaging with the Wi-Fi logo.

Ron K.
Ron K.

You have the power of the Internet at your fingertips. A search using Google and the keyword '802.11n' returned 7,310,000 hits.

foster.ryan
foster.ryan

me=legally use 1500w on channels 1-7 of the 2.4ghz band, access-point=0.5w; if you know what "FM cancellation" (aka fm interference cancellation) is that's the end of it right there. The 802.11 protocol is amazing in it ability to deal with interference, but when the successfully transmitted packets drop to zero, there is nothing any protocol can do to help. This can happen on any band, it just might not be legal.

foster.ryan
foster.ryan

What I had meant by mission critical was devices crucial to the health or survival of someone in the hospital. For instance, on the normal 2.4ghz 802.11, channels 1-7 are in a ham band. It is little used band, but potentially I could unwittingly be talking to someone in the car in the parking lot, and my legal limit of 1500w out could stop all those devices easily. Thankfully, this would be an annoyance, and not a life threatening event. Now out of that band and in a 'regulated band' as the person I was replying to meant it, there won't be any accidental disruption, but someone could very easily build a transmitter with a wide bandwidth to take out all traffic on that band and be in that same parking lot or even inside the building. If there were any devices that were actually mission critical, this person with malice could potentially not only annoy the staff, but could kill someone. Besides malice, I don't think anyone would ever willingly stake their life on their cell phone connection. That's essentially the same thing. Add into that the ease that the phone can be jammed (regardless of legality,) and that's your network-- no matter what band it is on. I doubt there will ever be mission critical traffic on wireless networks in any hospital. I am sure that there will be devices of convenience.

Michael Kassner
Michael Kassner

When I had my triple bypass. My monitor was wireless. All the nurses had wireless devices as well. They would be alerted via them as well as use them to voice record notes.

Michael Kassner
Michael Kassner

You will see larger gains due to better receiving and transmitting hardware. You lose some of the leverage from MIMO. But, I use 802.11n for several long distance links up to 10 kilometers.

Kim SJ
Kim SJ

I'm a great fan of WiFi, but I'm even more of a fan of wired networks. Here are some of my reasons: 1. Radio spectrum is a scarce resource It's shared with everyone in the vicinity, and it's really useful for hand-held and portable devices. Why make them work less well by using it for things you could use wire for? 2. Radio is hard to debug If things start not to work well, it's impossible for the average Joe to know where to start fixing it. Especially since it may be your neighbor's kit which is causing the problem. My brother had a WiFi music distribution system where the client in the kitchen would only work if he parked his car in the right place. If a wired link goes down a) it may well only affect one client, and b) the number of places to look for the fault is very small. Back in the days of coax Ethernet, I recall someone did a study comparing the Mean-time-to-repair of coax and structured wiring. The shared medium came in at 4 hours, compared to 45 mins for Cat3 (as it then was). [Wish I could find a source for that study] 3. Wired Ethernet is probably greener In particular, using PoE to avoid the use of inefficient plug-top PSUs and/or batteries saves a lot of power base-load. 4. Streaming As more and more Internet usage involves continuous streams of data, worst-case bandwidth becomes the critical factor. A performance which might be excellent for file transfer could be awful for watching a video stream. And the more people who are streaming, the more dramatic the effect of a big file download over the same wireless will be. 5. Bandwidth needs In the UK, Virgin have already run trials of 200Mb/s broadband connections. And for those who say "why would you want that much", I say you already have 1Gb/s connection to your desktop, why would you want less to the outside world? Probably only because today it costs too much. Samsung estimated that ultra-high-definition, 3-D TV will be taking 300Mbits/s within the decade. 6. Raw capacity My house has a 24-port gigabit-switch. So that's 1Gbit/s to _every_ port, all the time. Using the silly "it's all mine" calculation beloved of wireless proponents, that's 24Gbit/s of network bandwidth in my house. Or, more sanely, almost twice as much point-to-point bandwidth as even the best wireless under ideal conditions. 7. Health issues I know this is a "maybe". But certainly, high-enough radio energy levels can do damage. We don't really know what safe levels are. Experience says that official limits for safe levels of most things trend downwards over time, as more and more obscure damage mechanisms are identified. The Romans thought it was safe to use lead containers with wine. Not so long ago we thought it was safe to use lead compounds in motor fuels, paints, etc. AND FINALLY... http://xkcd.com/653/ says it all. :-)

pgit
pgit

Can't speak for the fellow you questioned, but I try to push wired networks over wireless and it's paid off. Wireless is slower and is far more error prone. It's also inherently insecure. Despite encryption or other measures I can't help thinking that the signal is out there and openly available to Joe Public, if primarily in raw form. And there's bound to be someone out there that can decrypt anything. I read somewhere the vast majority of nation's "spying" activities world wide are actually corporate espionage, looking for proprietary info to get a leg up on the competition. A wire is just so much more peace of mind... I'm sure I'll get blasted for saying this bit, but I don't believe wifi, cell phones, cordless phones and other devices emitting microwaves into your surroundings are a good idea. I've been following it and more and more articles come out all the time showing a 1 to 1 correlation with cell phones and brain tumors. I bet that's what got Ted Kennedy. He was an early adopter of cell phone technology, had a longer exposure than most. There's also a lot of young folk, in their 20's, coming up with brain tumors and other odd, heretofore unseen maladies. Here's really why cell phones were 'approved' by all 'authorities' with oversight of the technology: http://arstechnica.com/telecom/news/2009/12/sprint-fed-customer-gps-data-to-leos-over-8-million-times.ars

Michael Kassner
Michael Kassner

You are exactly right. It is a good thing the Wi-Fi Alliance is around. 802.11 would be no where near what it is today without them.

Jimbo Jones
Jimbo Jones

My hospital has had for many years a wireless telemetry system like your heart monitor; iirc it operates in a 600 MHz band. For legal reasons and for patient care it is very important to get as close to 100% of the monitor data as possible, so it is mission-critical to a certain extent. But it is not absolutely vital in the sense that a patient would physically suffer if the data flow stopped. There are backups. So there are varying degrees of mission-critical-ness. I also agree with Foster that even in the regulated spectrum, or with an underground fiber conduit for that matter, you're taking a gamble. But life is full of risk-benefit tradeoffs. You just have to be aware of the risks and their relative values to your organization, so you can plan accordingly.

Michael Kassner
Michael Kassner

The creed of an amateur radio op is to only use enough power to get the job done. Besides, I would be scared to death of a 1KW S band amp. That is the largest one that I know of. I absolutely agree with your assessment of mission critical. I also would consider the wireless heart monitor I had after my triple bypass to be very mission critical. At least to me.

foster.ryan
foster.ryan

There are regional restrictions in the 70cm band and over, but the federal limit is 1500 watts pep for all licensees higher novice all bands over 30mhz, and 1500w pep on all bands except 60m and 30m for General and above. I couldn't find a link for the fcc rules that wasn't a book you would have to pay for, but on this ARRL bandplan it does give the same information but less detailed. http://www.arrl.org/FandES/field/regulations/Hambands_color.pdf If you can find it, I know it has to be somewhere on this FCC page: http://wireless.fcc.gov/index.htm?job=rules_and_regulations (If you do manage to find it, your a better man than I.) Your completely right, it is a ridiculous amount of power on such short wavelengths, and would melt plastic-- I don't think anyone would realistically be using that much power on such a short wavelength, but it is legal. When I was in the army, we used to cook hotdogs on the hf antennas in seconds so I can't even imagine what 1500w at 2.4ghz would do. (400w rtty hf radios, after amp, before the amp it was 100mw.) You can lose a connection through anything, but mission critical traffic for a hospital also shouldn't be leaving on a wan line either, but if it did, (besides say having a microwave backup like critical infrastructure like power plants have, who do have mission critical data over fiber wan generally have as a backup) it is a lot harder for someone with bad intentions to cut that line than to jam a network with rf. There is no network that someone can't take down, but it is a lot easier for someone to do so undetected. I think the real issue in this conversation is what is mission critical-- in this example, of the hospital, the mission could be summarized as saving lives, so no systems on which someone could have their life depend-- anything that can't go down without hurting someone, not just inconveniencing them, that is the mission critical traffic. The mission critical traffic for a different industry may not be near as important and could be done wirelessly, but in the example he gave, we are talking about peoples lives, not just a loss of profit or productivity.

Michael Kassner
Michael Kassner

I need for you to explain what you mean. As an aside, I am an amateur radio operator, which allows me use 100 watts on 2.4 GHz. I do a lot of satellite work using those frequencies. 1500 watts would melt plastic. Sure, losing connections is an obvious issue. Yet I have had a solid fiber optic connection drop due to being cut by a backhoe. Guess which one can be repaired faster.

Michael Kassner
Michael Kassner

I did not say anything about exceeding any countries regulations. As an amateur radio operator, I am acutely aware of my responsibility in this regards. That is why directional antennas are in use, to avoid the 360 degree spreading of RF when you are only trying to connect to a single remote link.

mpaignigiannis
mpaignigiannis

There is so much misconcept about RF and everyone believes that WiFi is a god-sent solution for all our problems. Well RF will never provide the broadband speeds that can be delivered through cable/UTP and that's it. Period. It is a shared medium. Not only between your users but users of the companies in the nearby buildings! It's purpose is to provide solutions where laying cables is far more expensive (adhoc seminar rooms, hotel lobbies, recreation facilities etc) Do not try to use the same bullet for all the jungle!! Let me note some misconceptions a)ok 802.11n sounds great but.... 40MHz do you understand how much bandwidth is that?? You consume 2/3 of the whole ISM band!!! The channels are non overlapping so numbers 14channels for Japan or 13channels europe or 11 for US are misleading!! In densely populated areas (ie European cities) it is rather ordinary to receive 10-15 access points in the near vicinity of your appartment (different ones in each veranda)!!! And if you try to deploy an enterprise wifi net for journalists at an important event PLEASE AVOID channels 1-3 because the moment those hand held wireless TV cameras appear you will have a completely disrupted net!!! They transmit so loudly that the band becomes completely useless even with APs of the highest fidelity / sensisivity. 2) American companies who sell equipement worldwde should at least respect EUROPEAN REGULATIONS. The limit is 20dBm EIRP people for Europe! Not 1Watt as in US. Only CISCO respects that and forbids shipment of "american configured" devices to european markets. I am tired of hearing of people using high gain antennas and believing that they are legitimate. EIRP is calculated at the point of highest gain! The technology is built for building HOTSPOTS not "hot islands" or hot acres as I like to say. Hot islands... (real islands surrounded with water you know...) yes they exist. Would you like to know that they operate at power levels close to GSM basestations ??? Users: please respect that as far as Rf is concerned the mass population is the guinnea pig. We are not aware if and how it can affect our health the environement etc. Do not use a technology for uses that it never was designed for. If you want 100Mbps lay 100meters of UTP. The chiepest and easiest. Regards

Kim SJ
Kim SJ

Thank you Michael. As for differing values I agree we probably have different standpoints, and yes the "correct" solution will be different for different needs. As for typical users' skillsets, yes I am unusually well versed in the art. My view of how the market must develop is that as far as possible IT systems should look after themselves. In my experience, where an RJ45 socket is available, it is much easier and quicker to plug in than to configure WiFi. Yes, it takes more effort to install, but that is done by a professional, and thereafter ordinary users do not even need to know simple magic like network IDs or keys. And you can let guests plug in, secure in the knowledge that once they have left the premises they have also left your network!

Michael Kassner
Michael Kassner

Thank you for the comments. I see your points and could argue them, but it would not help. That is because we place different values on the bullet items you discussed. I also see that you are not what we would call a normal user. You have to take that into consideration as well. Would you expect everyone to have your skillset or not have anything?

Michael Kassner
Michael Kassner

I have a slightly different perspective though. I am more concerned about the security of my 3G connection than any WPA2 encrypted connection.

CharlieSpencer
CharlieSpencer

I was trying to encourage Kris.J to post something to support his statement. I think there will be a time when wireless will be as fast as wire. Even then, where I work we'll still require a VPN connection (or whatever replaces it) that we won't require with wire. That's another password for my users to remember, and another reason why they'll still prefer wire.

Editor's Picks