MU-MIMO

 Multi User Multiple Input Multiple Output

    MU-MIMO was first introduced in 802.11ac as an optional feature. In .11ax, downlink MU-MIIMO is mandatory but uplink MU-MIMO is still optional. There has been a lot of marketing around this and many clients do not see the benefit of this. Let's look at what MU-MIMO is, how it operates in a perfect world, and if it's worth the hype.


    At a basic level, MU-MIMO was supposed to solve some inefficiencies in Wi-Fi. When you think about it, the biggest thing that plagues wireless networks is the nature of the technology, it half duplex. One client transmitting at a time. So that means an AP or a client can transmit on one channel at a time, competing for the same airtime. This causes data to move much slower than it could, and it gets drastically worse the more clients that are on a single AP/channel. So how do we solve this? We cannot have collisions (two clients transmitting at the same time) or else traffic doesn't flow. What if we could send data to more than one client at a time? This way, only one device would be talking at a time, but maybe four clients can listen at the same time, only deciphering their own data. Sounds like a great idea!


    Many access points are 2x2:2 or 4x4:4 (T:R:SS, Transmit radios: Receive radios: Spatial Streams). Many clients are *typically* 2x2:2. So what happens if we have a 4x4:4 AP transmitting to a single 2x2:2 client? Half the radios aren't utilized! In MU-MIMO we can allow the AP to use half it's radios for one 2x2 client and half for another, or even split into fourths to transmit to four 1x1 clients.


    To start this process, the AP must facilitate the data exchange to the clients to notify them of being part of a MU-MIMO group. This process prepares the clients to receive a MU-MIMO transmission. First the AP begins the sounding process. The AP sends a Null Data Packet (NDP) to each of the clients, then each client responds with a Feedback Matrix. This allows the AP to perform beam-forming for each client, effectively steering traffic to each of the clients. After this feedback matrix is received, the AP creates a steering matrix to determine how to direct the traffic to each client. The point of this is to make each client hear the transmission that is intended for them the strongest, and not hear the transmissions meant for the other clients (remember a single MU-MIMO data transmission is sent to all clients in the group at once). Each client should hear the transmission meant for them and not or barely hear the transmission for the other clients.


    So here comes the problem, what if those clients are too close to each other? The users will experience inter-user interference due to the downlink transmissions interfering with each other. Due to this, MU-MIMO requires what is called Spatial Diversity. The clients need to be far enough away from each other to avoid interference. In many applications this is not possible. Think classrooms, conferences, and offices. Many times, clients are very densely populated. Because of this, MU-MIMO is rarely seen to actually have an effect in increasing performance. Also, the sounding process of MU-MIMO can negate the increase in efficiency due to overhead. So with MU-MIMO not having a huge benefit and many of our clients only being 1x1 or 2x2, why would I bother purchasing a 4x4:4 access point? I'll go over that in another post!

Comments

Post a Comment

Popular posts from this blog

Following Design Processes

    It is important to use proper design methodology when designing wireless networks. WiFi has evolved numerous times since it was first implemented, yet I still see very outdated or flat out wrong ways of designing a wireless network. The biggest thing I find is that many places do not follow the design process when choosing to install a new or update an existing WLAN. Essentially we are throwing hardware at a design problem and expecting it to work. In this post, I'll cover the proper way to design a wireless network according to CWNP's guidance. Thank you CWDP-303 official study guide from CWNP for providing me with the ability to do so!  Define     This might be one of the most overlooked parts of the design process. This step in the process sets the foundation for a good, well thought out design. Instead of simply getting a floor plan and jumping straight into a network planner, let's understand who or what we are designing for. During this stage, meeting ...
Read more

ExtremeCloud IQ Instant Port

Image
Instant Port enables you to automate your access layer by dynamically assigning VLANs based on the end devices MAC and/or LLDP capabilities. This document will cover the configuration of instant port and show an example.   Configuration of Instant Port is simple. First, create an Instant Port Profile within your network policy then assign that profile to your Switch Template.   ·          The Non-Forwarding VLAN will be used by default to learn the MAC and LLDP capabilities of the end device. ·          The Default Port Type will be used if you wish to provide non-matching devices layer two connectivity. ·          You can choose whether to allow non-matching devices to use either the Default Port Type or the Non-Forwarding VLAN. By using the Non-Forwarding VLAN, unknown devices will be dropped in an unused VLAN without the ability to reach other ...
Read more