Why aren’t there massive MIMO for FDD?

Hello NR Experts.

Looking for an answer to a boggling question for quite some time…

Why aren’t there massive MIMO for FDD?

What are the technology limitations?

1 Like

There is no technology limitation, but the overhead of CSI report to be sent in UL, and delay between DL CSI signals and UL CSI report makes FDD not suitable for massive MIMO.

1 Like


So channel estimation delay issues.

If no technology limitations, is it safe to assume it’s physics limitations?

The reason I am asking is that Massive MIMO makes sense in FDD since the benefits are huge…

1 Like

I don’t get what you mean by physics limitation.

1 Like

Well, I mean if what you mentioned isn’t a technology limitation, then what is the limitation of not seeing massive MIMO in FDD?

Maybe it is economical for the time being?

1 Like

Using m-mimo with FDD makes the UL capacity decrease significantly.

And in a channel that is changing quickly, the DL ideal capacity will also not be achieved.

1 Like

Eigen based Beamforming used for TDD only.

Grid of beam based solution used for FDD/TDD where fixed set of weight used.

1 Like

I believe that Massive MIMO is useful for both TDD and FDD modes. But TDD is generally used due to its better results estimating channel conditions, given that same frequency for uplink and downlink transmissions is used, besides this reduces the time required for channel estimation and increases system spectral efficiency

1 Like

It would be easier to answer…

FDD Bands has -max up to 2.5 GHz
Whereas TDD Bands are going 3.5 GHz, 4 GHz (And if we refer to latest release of NR-FR1 it goes upto 6 GHz too).

Now point is if Frequency increase we need more power to transfer - Radiation Patterns defined by "Spread " will not suffice overall requirement Because as we know F [incrase] = c / wavelength [decrease], hence Distance Decreases.

As we Increase Frequency = Distance decreased.

Now, how to reach to end user so there is concept called Beamforming

To achieve Beamforming:

  • We need more antenna Power which SPATIAL MUX Antenna Based radio can’t provide so we need Massive Mimo Radios.
  • We need Multiple Number of Layers mapped by Distributed Unit towards Radio to avail more resources.

To get more antenna gain, TX gain we increase the number of antenna Elements . higher the antenna Element - Higher is the gain achived
Also we can project the beam in Azimuth & Elevations can be steered in directions.

1 Like

Massive mimo antennas will be too big at lower frequencies.

At higher frequencies elements are smaller and more can be packed together to support massive mimo.

1 Like

This is the same reason 3gpp restricted number of maximum beams to:

  • FR1 - 4 for upto 3 GHz
  • FR1 - 8 upto 7.12 GHz
  • FR2 - 64 upto 71 GHZ

Now here is one challange.

Do you really need these many beams? Or they really fit in with TDD frame structure we can discuss Seperatley if anyone needs more details on this :innocent:

1 Like

What’s the diference between TDD and FDD?

1 Like

May be RF Experts can speak better than me let me try :upside_down_face:

TDD used for time division duplexing which will help user to utilize more BW for frequency domain as the frequency can be available and resource allocation based on time will be more for UL and DL just what we need is. Synchronization gap

Where as FDD

Utilizes the frequency as one range for DL and other for UL hence Frequency will be not much utilized

Simple story

More Bandwidth = more Frequency available

The above given context was assuming more frequency available so how to deal with distance :hugs:

1 Like

In FDD, 2 separate frequencies are used for DL and UL.

In TDD one single frequency use for both there is Time Gap (Gard Time) between DL and UL.

Just imagine we have 10 MHz frequency.
Then we need to split in FDD for DL and UL.
But in TDD we use hole 10 MHz frequency.

The main difference between FDD and TDD.

1 Like

The question that arise here is why 3gpp didn’t defined fdd bands for higher frequency. Answer to this question defines fundamental reason why fdd is not fit for massive mimo.



Huawei’s Pioneering FDD Massive MIMO Is Reducing Operators’ Cost Per Bit with 3–5x Capacity Increases

Massive MIMO Can be used in two modes FDD and TDD.

For TDD, Downlink beamforming in Massive MIMO relay on uplink pilot measurements—exploiting reciprocity and TDD operation.
For FDD, Massive MIMO uses a predetermined grid of beams with UEs reporting their preferred beams, mostly in FDD operation.

The original Massive MIMO concept assumes TDD and exploits reciprocity for the acquisition of channel state information (CSI) at the BS.

For the question, which is better?

TDD is better in performance due to TDD is using the same frequency for downlink and uplink and as different frequencies have different channel effects, So TDD is better in channel estimation as compared to FDD ( as FDD utilizes different frequencies for each transmission direction).
So TDD requires less channel estimation time, hence has system has more time for data transmission, therefore, increase in system spectral efficiency.


Dell’Oro analyst Stefan Pongratz said Huawei also signaled some optimism about the FDD massive MIMO opportunity, announcing an FDD-based 32T32R system, primarily targeting operators with limited upper mid-band spectrum.

“As a reminder, FDD based massive MIMO systems are not new, and this concept has been around for some time,” wrote Pongratz. “However, FDD based massive MIMO technologies have not gained the same mass market acceptance as TDD based solutions. In addition to the relative efficiency gap between FDD and TDD as a result of leveraging channel reciprocity in TDD systems, FDD based solutions typically also operate in a lower spectrum band, increasing the physical size of the antennas.”

Preliminary Huawei tests suggest that its FDD-based 32T32R massive MIMO antenna can deliver 3-4 times of capacity growth relative to 4T4R.

1 Like

That’s probably due to higher duplex spacing required to overcome out of band emissions and is directly proportional to frequency. At high frequency bands, its computationaly difficult to estimate DL CSIs for UEs and cost-inefficient to deploy different antenna grids for UL and DL at both ue and base station for mid and high band communications due to immensely high duplex spacing.
However Ericsson and some other vendors are coming up with FDD mMimo solutions in both 4G and NR in lower bands.

1 Like

Nokia has it too, since 2018 → Filtronic and Nokia Develop FDD-LTE Massive MIMO Antenna | 2018-04-08 | Microwave Journal

Actually - Huawei and Nokia lock horns in Massive MIMO weigh-in → Huawei and Nokia lock horns in Massive MIMO weigh-in - Rethink

1 Like