Mesh - Making Wi-Fi a more efficient network, handling the mobile explosion

Making Wi-Fi a more efficient network, handling the mobile explosion


In a related Fastnetnews article, it was posited that MU-MIMO (which underpins 802.11ac) could be the underlying tech for the pCell technology, but elsewhere it’s claimed that SDR (Software Defined Radio) is the underlying technology. I guess both could be true as SDR is a hardware platform that can run a variety of signalling schemes.

In fact MU-MIMO is already being proposed to run over SDR and given the reality of setting up an OpenAirInterface LTE base station for 1900 euro or less…things get interesting.

The big deal about 802.11ac, or “gigabit Wi-Fi,” is the promise of much higher speeds for Wi-Fi clients compared to the maximums under 802.11n today. But 11ac is actually several big deals, including something called multiuser MIMO [my-moe], which is part of the “Wave 2” 11ac features expected later this year.

The IEEE 11ac standard adds or extends a number of technologies that, first, give a big boost to the overall data rate, and the accompanying throughput; and second that “protect” and sustain those gains, with a more reliable signal that’s able to support higher throughput over longer distances and gradually and gracefully step down to lower throughput in order to maintain a reliable connection.

With 11ac, channel width jumps to 80 MHz wide (compared to 20 MHz for 11n), and then to 160 MHz wide in Wave 2; supports up to four spatial streams in Wave 1, and up to eight in Wave 2; a much higher density modulation scheme, 256 QAM vs 64 QAM for 11n; and in Wave 2, multiuser MIMO with transmit beamforming.

The basic idea of MU-MIMO is simple: An 11ac access point radio can send data to up to four client radios at the same time (though the clients also have to support MU-MIMO), directing a separate “spatial stream” to each one. By contrast, today’s three-stream 11n access points can only talk to one client at a time, one after the other. Basically, MU-MIMO will make the Wi-Fi network much more efficient; and ensure that, as more users connect to the access point, they will have an overall better “wireless experience.” A few chips for access points and wireless routers are available now and vendors expect the first network and client products to be available starting late this year.

Technically, multi-user MIMO is an 11ac option; vendors are not required to support it. But most observers think that, in short order, it will be a standard feature of almost all 11ac products.

MIMO was first introduced in the 802.11n standard. It took what had been a problem, and turned it into a way to vastly increase Wi-Fi data rates. The problem is called multi-path, which describes what happens when a radio signal is transmitted and then is reflected and bounced around by objects around it, arriving at the receiver in multiple paths at slightly different times and angles. The result is interference and a slower signal.

MIMO exploits the characteristics of multipath, by adding antennas on both the transmit end and the receiving end of a signal. The multiple antennas, via spatial diversity, improve the transmission’s signal-to-noise ratio and improve the reliability of the system by combatting fading. They also, through spatial multiplexing, use the different paths between the antenna groupings to carry vastly more data via two or three or four spatial streams.

Today’s widely deployed 11n access points, routers, and clients all use MIMO, with most access points supporting three spatial streams. But if there are three clients – laptops, smartphones or tablets – associated to a given three-stream 11n access point, only one can send and receive with the access point at a time: in effect, all three spatial streams are “aimed” at only one client. A time slicing algorithm allows fair access by all the clients associated to the access point. Depending on the number of antennas in the client (three for a laptop, one for a smartphone, for example), and hence on the maximum throughput it can support, most of the access point’s capacity may be idle at any given moment.

“A common misconception is that multiuser MIMO makes the network ‘go faster,’” says Matthew Gast, director of emerging technologies at Aerohive Networks and author of “802.11ac: A Survival Guide”. “It really doesn’t. It creates multiple logical connections with devices at the same time. This increases network efficiency.” That means by being able to transmit to as many as four clients at once, the 11ac access point can make full use of its available “gigabit” capacity at every moment.

Currently, Quantenna Communications offers a commercially available 11ac chip, the QSR1000, with four transmit and four receive antennas, supporting four spatial streams and multi-user MIMO, so far the only chip to do so. Rivals Qualcomm and Broadcom, and others, have pledged to bring MU-MIMO to market. Quantenna says it will introduce an 8 x 8 MIMO 11ac chip in 2015.

The Quantenna chip can talk with up to four clients at the same time. Or it can group clients into as many as four groups, again making four simultaneous transmissions, according to Bahador Amir, technical marketing director for the chipmaker. Within each group of, say, five or six or more clients, the Quantenna chip then uses the traditional Wi-Fi convention of talking to each client one at a time in sequence.

An access point in this case could group together the clients furthest away; as distance from the access point increases, the clients have to step down to lower speeds, which takes more network time, “hogging” the bandwidth inefficiently, according to Amir. But with multi-user MIMO, these more distant clients can be grouped, sharing that capacity, while other closer clients are served at higher speeds, at the same time.

Transmit beamforming, which was optional in 11n and is mandatory in 11ac, is a key companion technology, says Aerohive’s Matthew Gast. Transmit beamforming in effect lets the access point “concentrate” or “steer” the RF energy in a specific direction, for a specific client. It does this, first by constantly talking with each client to glean specific channel measurements, which the access point uses to slightly change the signal phasing. That change allows the signal to that client to propagate at a higher effective power level. With MU-MIMO, the access point can thus optimize its signal for up to four clients at the same time.

Multiuser MIMO is arriving at a critical time: The number of Wi-Fi clients is exploding, thanks to smartphones and tablets, the end-user desire to avoid cellular charges, and the end-user demand for video, music, and constant social networking and updating. Not everyone near a Wi-Fi access point may have a Windows or Mac laptop, but nearly everyone is likely to have one or two or more mobile devices with Wi-Fi.

“As clients increase, access points have to cope with that,” says Quantenna’s Amir. “With single-user MIMO, you end up in situations where you saturate the access point, because the access point has to rotate through, say 100 clients. You run out of time, resulting in low-bandwidth service. Multiuser MIMO can improve this dramatically.”

Most smartphones and even tablets with 11ac radios are likely to be single-stream clients (with just one antenna), with a maximum possible data rate of about 433Mbps. Four-stream 11ac access points with transmit beamforming are ideally suited to this kind of client population.

“The one really big question about multiuser MIMO is, how well does it work?” says Craig Mathias, Network World blogger and principal with Farpoint Group, a wireless consultancy. “And the answer is, right now, no one knows. I’ve only seen it in a lab. There are no production deployments. So it’s unclear what benefit you will get.”

Wi-Fi networking is rife with variables, he points out. While an 11ac access point with 80-MHz channels has a theoretical capacity of 1.3Gbps, “in the real world, that will not happen,” he says. “Benchmark tests from the vendors show performance is all over the place. Radio performance is highly variable, and firmware, chips, board layout, drivers, antennas, prevailing radio conditions all have an effect. You cannot predict what performance you will achieve.”

“The main advice I give to clients is, wait and see, and evaluate new [11ac] products as they appear,” Mathias says. “It will take time to see how MU-MIMO will work and, most of all, how it will work in your specific environment.”

1 Like

When Steve Perlman was giving some of his public lecture or intros regarding PCell last he had some very vocal opponents in the audience who were basically saying that PCell was nothing special and MIMO could do it all and more. He had a hell of a time trying to quiet them. Troubling was they looked like professors and other sorts who knew what they were talking about.

I think part of the problem here is that defense has been using this stuff for a long time. The same impulse that had them funding the early internet has them seeking unjammable encypted mesh for the battle field, air, ground and undersea. After all without communications they die in combat. My hope is no bs about defense patents ever comes up, and if it does let us please have instant workarounds. It would be a predictable cheap shot by threatened entities.

PCell does seem to get beyond beam forming, as Artemis pitches like its based on the sensory equivalent of aperception, that instead of so much beam forming it uses interference as a carrier or guide, and its able to do so because of some exotic math and computation. That it can tract things to with in millimeters in 3D space. That its solves latency (sub millisecond for download) bandwidth and crowding issues. This it can ride on present tech and is super cheap and super power efficient. Sounds great except as has been pointed out its power may possibly be expanded to include uploads and to allow us to cut the cord on telcos and cable.

Hilarious if PCell got widely deployed and there was something simple that would allow us to do something like Jailbreak the telcos and cable themselves. The anti jailbreak laws (based on financing claims,) the neutrality attacks and the bribing to prevent municipal internet show how obsolete this industry is. As is the industry in the state in particular to whine about ROI and jobs when its a global last place loser in price and performance.