MU-MIMO, or how OFDM in WiMAX / LTE differs from OFDM in 802.11a / g / n
In continuation of the post Why in WiMax and LTE use OFDM . The author has dismantled almost everything, there is only one small, but essential detail. Since this detail affects the current Wi-Fi standards and promises a breakthrough in the future, I think the reader will be interested.
Let's imagine the following situation: we have a base station (Wi-Fi or WiMax) on which there are 20 clients with, say, Skype. Each client conducts a video conference (bidirectional, unicast) which requires, say, 1 Mb / s of bandwidth and a delay of no more than 150 ms.
Nominal Wi-Fi 802.11n access point allows you to efficiently transmit (throughput, not datarate) from 150 Mb / s to the radio and supports 802.11e QoS for efficient transmission of multimedia traffic. WiMAX allows you to get about 30Mb / s downlink and 20Mb / s uplink per channel. Total, bandwidth per client Wi-Fi: 150/20 = 7.5Mb / s (7.5 times more than required); WiMax client bandwidth: 1.5Mb / s downlink and 1Mb / s uplink. Attention, the question: what technology Skype will work "smoother"?
Oddly enough, Wi-Fi with a huge bandwidth advantage will work worse. And the more customers there are, the more profitable (up to a certain level, naturally) WiMAX will look. This is due to one specific optimization of the work of OFDM in WiMAX.
It is important to remember that OFDM is a multiplexing technology, i.e. seals. OFDM divides the signal into subcarriers, which can be many. In Wi-Fi, components of the same data stream are transmitted over all subcarriers: i.e. compaction is used for redundancy. It happened for a number of technical and historical reasons. Thus, at one point in time, the Wi-Fi access point can serve only one client (not counting multicast / broadcast), even with all the latest MIMO, Spatial Mutiplexing and Beamforming. When transmitting through a “pipe” of 150 Mb / s of data flow coming from, say, the Internet at a speed of 1 Mb / s, our bandwidth is 149.150 ~ = 99.3%. And it's a shame that if we have 20 clients, the access point will serve them in turn with the same terrifying efficiency of 0.7%. While this was still normal for 802.11a / g with a maximum bandwidth in the region of 26-29Mb / s, for 802.11n with its potentially possible theoretical 300Mb / s + at 4x4: 4 MIMO, this is not at all good.
')
WiMax, approaches the question a little differently. First, there are more subcarriers. And secondly - different subcarriers can simultaneously carry data for different clients! Thus, in our situation with Skype, clients will transmit / receive data all together - in parallel - and not sequentially, as in Wi-Fi. This technology is called MU-MIMO (Multi-User MIMO), unlike SU-MIMO (Single User), used in Wi-Fi. Thus, the efficiency of bandwidth use throughout the cell increases markedly. Of course, in reality everything is not so simple - a more complex protocol is required (how to let the client know which subcarriers are his?), More complex signal processing (higher cost of the client and basic equipment), control of roaming, and the MU-MIMO technology itself not only the division of OFDM subcarriers by users. But all this is achievable, although it is beyond the scope of this short post.
Now back to our realities. Today, many homes have 3-4 or more Wi-Fi devices. Some large companies declare that they are going to plug a Wi-Fi module in general into every household appliance unit (it’s not clear, however, why this is needed, and why not to use ZigBee or BT4.0). Most of these devices do not need tens or hundreds of megabits of bandwidth, do not need a minimum delay, but because of SU-MIMO they will ruin the lives of others. And the more devices - the more problems. The good news is that the future 802.11ac WiGig standard will support MU-MIMO. The bad news is the next scheduled upgrade, but in the context of the short life cycle of modern consumer devices - who cares? :)
Let's imagine the following situation: we have a base station (Wi-Fi or WiMax) on which there are 20 clients with, say, Skype. Each client conducts a video conference (bidirectional, unicast) which requires, say, 1 Mb / s of bandwidth and a delay of no more than 150 ms.
Nominal Wi-Fi 802.11n access point allows you to efficiently transmit (throughput, not datarate) from 150 Mb / s to the radio and supports 802.11e QoS for efficient transmission of multimedia traffic. WiMAX allows you to get about 30Mb / s downlink and 20Mb / s uplink per channel. Total, bandwidth per client Wi-Fi: 150/20 = 7.5Mb / s (7.5 times more than required); WiMax client bandwidth: 1.5Mb / s downlink and 1Mb / s uplink. Attention, the question: what technology Skype will work "smoother"?
Oddly enough, Wi-Fi with a huge bandwidth advantage will work worse. And the more customers there are, the more profitable (up to a certain level, naturally) WiMAX will look. This is due to one specific optimization of the work of OFDM in WiMAX.
It is important to remember that OFDM is a multiplexing technology, i.e. seals. OFDM divides the signal into subcarriers, which can be many. In Wi-Fi, components of the same data stream are transmitted over all subcarriers: i.e. compaction is used for redundancy. It happened for a number of technical and historical reasons. Thus, at one point in time, the Wi-Fi access point can serve only one client (not counting multicast / broadcast), even with all the latest MIMO, Spatial Mutiplexing and Beamforming. When transmitting through a “pipe” of 150 Mb / s of data flow coming from, say, the Internet at a speed of 1 Mb / s, our bandwidth is 149.150 ~ = 99.3%. And it's a shame that if we have 20 clients, the access point will serve them in turn with the same terrifying efficiency of 0.7%. While this was still normal for 802.11a / g with a maximum bandwidth in the region of 26-29Mb / s, for 802.11n with its potentially possible theoretical 300Mb / s + at 4x4: 4 MIMO, this is not at all good.
')
WiMax, approaches the question a little differently. First, there are more subcarriers. And secondly - different subcarriers can simultaneously carry data for different clients! Thus, in our situation with Skype, clients will transmit / receive data all together - in parallel - and not sequentially, as in Wi-Fi. This technology is called MU-MIMO (Multi-User MIMO), unlike SU-MIMO (Single User), used in Wi-Fi. Thus, the efficiency of bandwidth use throughout the cell increases markedly. Of course, in reality everything is not so simple - a more complex protocol is required (how to let the client know which subcarriers are his?), More complex signal processing (higher cost of the client and basic equipment), control of roaming, and the MU-MIMO technology itself not only the division of OFDM subcarriers by users. But all this is achievable, although it is beyond the scope of this short post.
Now back to our realities. Today, many homes have 3-4 or more Wi-Fi devices. Some large companies declare that they are going to plug a Wi-Fi module in general into every household appliance unit (it’s not clear, however, why this is needed, and why not to use ZigBee or BT4.0). Most of these devices do not need tens or hundreds of megabits of bandwidth, do not need a minimum delay, but because of SU-MIMO they will ruin the lives of others. And the more devices - the more problems. The good news is that the future 802.11ac WiGig standard will support MU-MIMO. The bad news is the next scheduled upgrade, but in the context of the short life cycle of modern consumer devices - who cares? :)
Source: https://habr.com/ru/post/132247/
All Articles