Wi-Fi’s Enigma: MU-MIMO



Without a doubt 802.11ac Wave1 has definitively ruled Wi-Fi tech-talk throughout the year, with “Gigabit Wi-Fi” declared as the wireless buzzword of 2014 (voted on by the staff at On The Fly Wi-Fi). Sure, other Wi-Fi trends like analytics, social and cloud have tried to grab the elusive buzzword top spot but nothing has ruled our chatter more than Eight-O-Two-Dot-Eleven-A-C.

802.11ac has raised the bar in terms of we can achieve with our wireless LAN networks. Greater speed and capacity, over previous generations, that will help carry us into the future – and trust me; we will need every ounce of Wi-Fi technology we can get our hands on as the surge in device and application use will continue to grow and push the limit. If you haven’t heard about IoT, you’re well advised to start learning about it. Check out OTFW Episode 11 or Episode 12 with @DevinAkin to glean some excellent IoT insights.

2015 is going to see shipment of 802.11ac Wave2 technology. As we’ve learnt this year, 802.11ac Wave1 was an evolutionary step from 802.11n enhancing two elements to achieve gain: Wider Channels (now 80Mhz) and Denser Modulation (256QAM).

As Wave2 enters the foray we are hearing about speeds upwards of 7Gbps over-the-air with the advent of even wider channels (160Mhz) and support for more spatial streams (up to 8). The combination of these elements, using math theory, allows us to calculate this given 7Gbps. Hopefully, most of you reading this are aware of current 5Ghz channel limitations and the unlikelihood of seeing mobile devices supporting more than 2-3 streams. The real world results will be less than the math calculates, with real-world achievement of roughly 2Gbps over –the-air (still a great feat, don’t read my view as coming from a negative angle).

So why all the fuss with 802.11ac wave2?

The real hype of 802.11ac Wave2, and my early entry for Wi-Fi buzzword of 2015 is MU-MIMO (Multi-User MIMO). Why the hype? Well, MU-MIMO is a paradigm change in how WLANs operate – many experts suggest that MU-MIMO is a revolutionary step in Wi-Fi – and I agree.

In essence, MU-MIMO provides ability to communicate to multiple stations (client devices) at the same time, on the same channel. This is revolutionary, as Wi-Fi has traditionally operated as a shared medium where a single device transmits at its prescribed time slice. This now-called “SU-MIMO” (single user MIMO) is how we’ve done it for years using fairly elegant algorithms to limit collisions and errors (and, BTW, we never actually used this SU-MIMO till now – just saying).

Many offer a simple analogy from our wired-side counterparts – going from SU-MIMO to MU-MIMO is similar to going from a hub to a switch. While not an exact truth, it is a reasonable way to bridge your understanding into the world of MU-MIMO.

What else do we know about MU-MIMO? Well, we’ve all seen some iteration of the following image depicting MU-MIMO. Many of us can describe this, and illuminate excitement about the possibilities.



So, given that we all “know” what the picture above is showing us, what are some of the unknowns and implications we’re not hearing about? What is the enigma of MU-MIMO? Here’s my list:

1. Single direction only:

Early marketing of MU-MIMO has failed to adequately inform that initial implementations only support Tx from AP to client. There is no support of multiple clients Tx’ing back to up the AP at the same time – currently bi-directional MU-MIMO is an unknown with potential for release at an unknown date (and yes, it will require new AP and client hardware – call it Wave3 if you’d like, just don’t shoot the messenger (and yes, .11ax might include it in a few years)).

2. Implication of Single Direction?

At this point it remains unclear the implication of being able to transmit to multiple stations, but reverting to SU-MIMO when client’s Tx back. In today’s world tons of traffic from clients go upstream – consider social sharing (photos/vids), two-way video (like facetime), and cloud storage. We simply don’t know what type of environments MU-MIMO is best suited for, and where the most gains will be achieved.

3. MU-MIMO Grouping?

At this point we are unsure optimal grouping of client devices – some suggest 3 single stream devices, other suggest 4 single stream devices in an MU-MIMO Group or derivatives therein (such as two 2 stream devices in a group).

4. Spatial Isolation Required?

It remains unknown as to how much spatial isolation will be required between client devices to successfully join a single MU-MIMO group, and decode their intended signal. In other words, will devices need to be 3 feet, 5 feet, 10 feet apart to participate and be able to distinguish the stream intended for them against noise. Certainly devices in too close of proximity to each other won’t be able to decode their stream from a nearby neighbor – it’ll simply be noise.

5. AP ability/overhead to decide SU-MIMO vs. MU-MIMO?

We simply have no insight as to how well APs will decide to build MU-MIMO groups vs. the tradeoff off just using the channel in SU-MIMO traditional mode. Consider that most clients are mobile, and moving – the calculation to keep up with devices and decide how to build an MU-MIMO group, dynamically in the real world, is a mystery today.

This is my main list of unknown factors when discussing MU-MIMO. Got something else? Add it to the comment section please.

The result of this list leaves open a very simple question: how much gain, in the real world should we expect with MU-MIMO? The best guess answer, today, is that 802.11ac Wave2 APs will provide near 2x performance from that found today in 802.11ac Wave1. That’s really including all the elements of a Wave2 AP, and putting it in the real world – including a big, and important, assumption that all client devices in operations have 802.11ac Wave2 radios as well (never forget about client-side capabilities as THE critical factor when answer the “how fast is the AP” question).

Circles to the point, and essentially a question I’m being asked a lot these days – should I wait for Wave2?

The answer remains fairly simple. If you have a need, or project, that you need to roll out now choose Wave1. It’s by no means a bad decision. 802.11ac Wave1 will provide a huge boost, over legacy 11a/b/g/n, in any environment, period. If you’re not buying immediately, and can wait for Wave2 in 2015 then you’re answer might be in hand. The important point, hopefully made here, is that Wave2 APs will absolutely be better than Wave1…. However…. the performance gain you should expect in the real world will likely be less than 2x and definitely not the 15x factor that some suggest between 802.11ac Wave1 and Wave2.

Rest assure that 2015 is going to be a bigger and more exciting year for Wi-Fi – I’m looking forward to it!

I want to take a moment to thank all of you who have participated in the On The Fly Wi-Fi community this year. Please keep coming back in 2015, and bring a friend with you!

Happy Holidays and Happy New Year!

4 Responses to “Wi-Fi’s Enigma: MU-MIMO”

  1. @kuzojoe says:

    I think the difficult obstacle for getting to Wave 2 will be infrastructure to the AP. Needing 10G links to each AP is not only expensive, but for many of us not practical. I also don’t believe you will see 8 spatial stream devices for the common user. Great article.

  2. Hemant says:

    During MU-MIMO, Tx power of radio gets shared among beams. Plus, imperfect null forming causes cross beam interference. So, SNR at client in MU-MIMO would typically be less than the SNR for the same client in SU-MIMO. That raises the question: Is it efficient to do two MU-MIMO Tx (which occur at lower speed due to lower SNR and all associated complexities of MU-MIMO) or two sequential SU-MIMO Tx (which occur at higher speed due to higher SNR and without additional complexities)?

    There are also secondary factors on each side like: Tx power of radio being typically lower for higher MCS (higher SNR of SU-MIMO), but then SU-MIMO can also benefit from diversity gain due to number antennas being higher than streams.

    Which way the efficiency balance tips in various scenarios is another factor unknown (to me).

  3. François says:

    That’s a great comment Hermant (and a great article Mike :)). I guess, it is going to depend on the environment. You will be more likely to deploy Wave2 in a high density environment where you know that you will have sufficient Tx power. Some other environments will probably be less suited for Wave2. It would be interesting to get some Wave2 APs and test different configurations/environments in 2015!

Leave a Reply