Cell-free massive MIMO is likely one of the technologies that will form the backbone of any xG with x>5. What distinguishes cell-free massive MIMO from distributed MIMO, network MIMO or cooperative multi-point (CoMP)? The short answer is that cell-free massive MIMO works, it can deliver uniformly good service throughout the coverage area, and it requires no prior knowledge of short-term CSI (just like regular cellular massive MIMO). A longer answer is here. The price to pay for this superiority, no shock, is the lack of scalability: for canonical cell-free massive MIMO there is a practical limit on how large the system can be, and this scalability concerns both the power control, the signal processing, and the organization of the backhaul.
At ICC this year we presented this approach towards scalable cell-free massive MIMO. A key insight is that power control is extremely vital for performance, and a scalable cell-free massive MIMO solution requires a scalable power control policy. No surprise, some performance must be sacrificed relative to canonical cell-free massive MIMO. Coincidentally, another paper in the same session (WC-26) also devised a power control policy with similar qualities!
Take-away point? There are only three things that matter for the design of cell-free massive MIMO signal processing algorithms and power control policies: scalability, scalability and scalability…
Further scalability issues are addressed in “A New Look at Cell-Free Massive MIMO: Making It Practical With Dynamic Cooperation” (https://arxiv.org/abs/1906.10853) that will appear on PIMRC 2019.
And in the journal version https://arxiv.org/pdf/1908.03119.pdf
Why do we apply power control in cell-free Massive MIMO?
As I know, power control is used to control the power of pilot in uplink to harness pilot contamination. What are the other reasons of using power control?
What you describe is one recent use case for power control, but the same thing is normally done also in the transmission of uplink data. The classical reason to use power control is to avoid near-far effects, where the received signals from cell-center users are much stronger than the signals from cell-edge users. In Massive MIMO, we can optimize the power control towards a performance metric, such as maximum sum rate or max-min fairness. You can find some references to papers in the following blog post: https://ma-mimo.ellintech.se/2018/10/02/are-pilots-and-data-transmitted-at-the-same-power/
Section 7.1 in my book “Massive MIMO networks” is also discussing these issues.
I have two question about this blog:
1. Why cell-free MIMO has less channel hardening as mentioned in the research ‘Ubiquitous Cell-Free Massive MIMO Communications’?
2. Another thing you mentioned that cell-free MIMO can provide uniformly good service. But in my opinion, whether it can provide fair service is depended on the power policy you adopt. Which means that even a celluar MIMO with adopt max-min power allocation policy can provide fair service. Isn’t it?
1. Channel hardening appear when you combine the signals from many antennas that have similar average SNRs. In cell-free massive MIMO, you will have many antennas but they are spread out and therefore have different SNRs. This is investigated in the following paper: https://arxiv.org/pdf/1710.00395
2. Yes, in both cases you need to utilize a max-min power allocation policy, but the rate that every user gets under such a policy will be different depending on the network architecture. I think what Erik means is that the rate will be larger in a cell-free system than a cellular system since the antennas are spread out and therefore there is a lower risk that a user has a bad channel to all its neighboring access points.
Thanks a lot!
Hi,
I have two questions,
1- Can you explain more about the backhauling? How the backhaul traffic transmit toward the network core?
2- What is the radio stripes? Is it a cable? Fiber?
1) I think that it is the connections between the access points and central processing units that Erik is referring to here.
2) Radio stripes are described in this paper https://jwcn-eurasipjournals.springeropen.com/track/pdf/10.1186/s13638-019-1507-0 and also in this video https://www.youtube.com/watch?v=ZepQAw21HfA
I have two questions:
1. How can we enhance the spectral efficiency in cell free Massive MIMO using power control strategies?
2. How can we achieve better power control for distributed as well as centralized scenarios which is scalable?
1. Since many users are active simultaneously, their transmissions will interfere with each other. Hence, it is not optimal to always transmit with maximum power, but power control can be used to find a better operating point – such as max-min fairness or maximum sum rate.
2. There are many centralized and distributed algorithms, some of which are distributed. We summarize them in the last chapter of the book “Foundations of User-Centric Cell-Free Massive MIMO”.
Hello.
I am working on a simulation of multiple CPUs’ interconnection ( Fig1 in the linked paper “Scalable Cell-Free Massive MIMO System” (1)). My first goal is to reproduce Fig 1 in the reference “Scalability Aspects of Cell-Free Massive MIMO” (2).
How did you simulate the interconnection of CPUs? Did you create an array with an N-number of a single CPU CF mMIMO? But then how can I show that one AP can be served by several CPUs?
Hi! I suggest that you contact Giovanni Interdonato directly. He is now a professor at the University of Cassino and Southern Lazio.
Hello.
I am reading your publication, Scalable Cell-Free Massive MIMO Systems,” and I have found your posted simulation codes. Thank you very much. This is helpful for understanding the concept described in this paper.
I have one question regarding this simulation. I have noticed in The transmitted signal was not used in the simulation. From my understanding the transmit signal ( data + pilots) is a random vector which is transmitted via a channel ( multiplication with channel matrix as you described in the paper) and at the receiver side we can estimate it. In the simulation, no vector was transmitted to the receiver, you only use pilots, right? Could you please explain in detail, how it works?
Thank you very much for your help
Your observations are correct. A large portion of the book is focused on deriving expressions for the spectral efficiency that can be achieved using different precoding/combining methods. Thanks to these expressions, we can evaluate communication performance without having to generate data signals and decode them. We just need to generate channel realizations, channel estimates, and precoding/combining vectors to compute the expectations in the spectral efficiency expressions.
Thank you very much for your reply.
I have another question. In the paper, section III, the power control coefficients are set as it is given in eq 4. Was it derived by authors or is it a known equation? Where can I find a detailed explanation of this equation and how it was derived?
Hi!
I think you are referring to another equation than (4), but the short answer is that there is no new power control method in the paper. We use methods from previous work. For example, the heuristic but well-performing method from reference [33].
Sorry.
Eq 4 I mentioned is from “Scalability Aspects of Cell-Free Massive MIMO” you mention in your post ( Fig.’s reference). In the pdf file, it is p 3
The discussion starts that “power control is applied independently in each AP…”
Ok, now I get it.
Yes, that formula is proposed in “Scalability Aspects of Cell-Free Massive MIMO” (there is no derivation but only arguments for why it is a good heuristic). Simulations are utilized to compare different functions f() and noticing that sqrt(gamma) is a good choice. Many papers on cell-free massive MIMO refers to this paper when using this power control strategy.
Similar strategies had been proposed earlier in other contexts, but not for cell-free massive MIMO, to the best of my knowledge.
Thank you very much for your reply and help with my questions.
in your simulation for the book “Foundations of User-Centric Cell-Free Massive MIMO”, I found a parameter that describes the height difference between an AP and a UE (in meters): distanceVertical = 10;
Is it the distance between UE and AP or the heights of antennas? or something else?
It means that the APs are deployed 10 m above the plane where the UEs are. So the minimum AP-UE distance becomes 10 m, which is achieved when the horizontal distance is 0 m.