Monday 15 November 2021

Will Distributed FD-MIMO be next big MIMO Enhancement?

We have looked at MIMO quite a few times in this blog. Back in February we looked at some of the advancements that Samsung and Ericsson had been showing here.

Last year, in a blog post, Samsung talked about Distributed Full Dimension MIMO (FD-MIMO). The key points were:

Around that time, the concept of Massive MIMO was proposed in academic papers. These papers proposed the idea of making the signal dimension at the base station much bigger by using a massive number of antennas such that all inter and intra-cell interference asymptotically go to zero. MU-MIMO performance would be improved significantly with a much lower interference level, therefore leading to capacity gain. It looked promising, but no one knew how to bring it to reality, since arranging 10s or 100s of antenna elements in the conventional way (i.e., in the horizontal plane) would lead to a base station that is longer than a bus, so obviously it was not going to work in a real deployment.  

An important breakthrough came when engineers at Samsung noticed that a concept called Active Antenna Systems (AAS), could be exploited to organize 64 or 128 antennas into a 2D active antenna array that is similar in size with a conventional 4-TX system as shown in the middle portion of Figure 1. Such a system is called a Full Dimension MIMO (FD-MIMO) system. Initial evaluation of the FD-MIMO system coupled with high-order MU-MIMO showed a capacity gain by a factor of 3-4 times for a 64 or 128-TX FD-MIMO compared to a 2-TX LTE system, as was summarized in a 2012 Globecom paper , “Fulfilling the promise of massive MIMO with 2D active antenna array”, and later in a 2013 IEEE magazine paper , “Full-dimension MIMO (FD-MIMO) for next generation cellular technology”. 

Samsung has been actively leading the FD-MIMO standardization process in 3GPP from the beginning, including the 3D channel model study in 2012 that paved the way for subsequent system design, the 4G LTE version of elevation beamforming and the FD-MIMO work from 2014, and more recently the 5G NR-MIMO version of FD-MIMO. Samsung has also been a leader in prototyping and testing the feasibility of the technology and was the first to demonstrate an FD-MIMO system supporting 12 simultaneous MU-MIMO users achieving a record aggregate capacity of > 20 bps/Hz in early 2015. These feasibility study result was later published in a 2017 IEEE JSAC paper , “Full Dimension MIMO (FD-MIMO): demonstrating commercial feasibility”.

Initial system level simulations show that the D-FD-MIMO system achieves up to 2 times cell average throughput gain compared to the FD-MIMO system, lifting both cell capacity as well as average user throughput. Such a cellular system can be flexibly deployed to “blanket” a given geographical area and provide better service for both outdoor and indoor users. 

We have developed a hardware prototype and performed field test to verify the feasibility and the performance gain of the D-FD-MIMO system. In the field test, 3 distributed LEGO MIMO RFUs and 7 UE emulators were used. When the number of active RFUs increased from one to three, the overall throughput improved by about 4 times.

A significant amount of work needs to be done before we can accurately quantify the benefits of the D-FD-MIMO technology, but these initial results are certainly promising and show a great potential for this new breakthrough of the MIMO technology.

Back in 2017, Samsung researchers also wrote a paper on this topic, Distributed FD-MIMO: Cellular Evolution for 5G and Beyond, which is available on arXiv here. Quoting from the paper:

Distributed Full Dimension MIMO (D-FD-MIMO) is an evolution of FD-MIMO. A D-FD-MIMO network assumes a cellular structure, where a cell is served by one BS and each BS is connected with a large number of antenna elements, of which individual elements are spatially distributed in the cell. One or more antenna elements are equipped with a digital port, and the signals transmitted and received from all the antenna elements within one cell are jointly processed to perform high order MU-MIMO operation.

Such a cellular system can be deployed outdoors in a city-wide area to provide service to both outdoor and indoor users. It can also be deployed inside the building to serve indoor users only. It is also suitable for providing service in a highly populated area, such as stadiums, shopping centers and airports, where a large number of the users are densely located.

Concepts relating to D-FD-MIMO includes distributed massive MIMO, CoMP (a.k.a. network MIMO) and distributed antenna systems (DAS). Distributed massive MIMO treats the entire network as one cell, featuring an enormous number of access points distributed over a large area, jointly serving all the users. pCell by Artemis can be seen as an implementation of the distributed massive MIMO albeit with a smaller scale in terms of the number of antennas. CoMP relies on the coordination among a few transmission points from the same or different sites to enhance User Equipment (UE) experience at the cell edge. DAS is initially proposed to improve coverage in an indoor cellular communication system, and is sometimes adopted in outdoor scenarios as well. One configuration for outdoor deployment is to have a few antenna arrays distributed throughout the cell to perform MIMO operations. Another DAS configuration deploys a number of individual antenna elements in a distributed manner in each cell of the network, which is similar to the D-FD-MIMO setting. Different from our system-level simulation approach, the analysis theoretically derives the asymptotic sum capacity when the numbers of UE and antennas in each cell both approach infinity with their ratio fixed, and assuming perfect uplink power control.

You can get the PDF of the paper here.

We have written about the Cell-Free Massive MIMO here and here. One of the realizations of D-FD-MIMO is as shown in Ericsson Radio Stripes. 

Researchers on this topic may also be interested in watching Wireless Future Podcast episode 13 on Distributed and Cell-Free Massive MIMO (embedded below). The description says:

In this episode, Erik G. Larsson and Emil Björnson discuss how one can create cell-free networks consisting of distributed massive MIMO arrays. The vision is that each user will be surrounded by small access points that cooperate to provide uniformly high service quality. The conversation covers the key benefits, how the network architecture can be evolved to support the new technology, and what the main research challenges are.

The description also contains some links and the discussion is also interesting to follow. You can jump on to the video directly here.

Related Posts:

No comments:

Post a Comment