Dynamic Cooperation Clusters

By deploying many distributed antennas instead of a few multi-antenna base stations, a more uniform communication performance can be achieved over a coverage area. The peak rates might go down but there is a much higher chance of getting a decent rate with 95% probability. This is the main motivation behind Cell-free Massive MIMO, which is the new name for Network MIMO with a large number of antennas (many more than the number of users). The key difference from conventional ultra-dense networks is that the distributed antennas are cooperating to transmit phase-coherently in the downlink and process the received uplink signals coherently. One promising way to deploy these systems is by using radio stripes.

The first papers on Cell-free Massive MIMO assumed that all antennas have access to the downlink data of all users and take part in the uplink signal detection of all users. This is both impractical and unnecessary in a large network, where each user is only physically close to a subset of the antennas. Hence, it makes practical sense that only those antennas that can reach the user with a signal power that is non-negligible compared to the thermal noise should transmit to that user and participate in the detection of its uplink data. 

I designed a framework for this 10 years ago, which I called “dynamic cooperation clusters” (DCC) and it can be readily applied to Cell-free Massive MIMO. The main idea was that every user selects which antennas should serve it in a user-centric manner, which means that any antenna subset can be selected. This stands in contrast to the conventional network-centric approach (which dominated the 4G CoMP literature) where only certain predefined disjoint groups of antennas are allowed to cooperate.

Although the DCC framework is a perfect fit for Cell-free Massive MIMO, the performance analysis that we did 10 years ago was admittedly simplified compared to what is possible with the latest methodology. We considered TDD systems that utilize reciprocity but assumed slowly fading channels that can be estimated without error, thereby avoiding pilot contamination and the computation of ergodic rates. To provide a bridge to the past, we wrote the paper “Scalable Cell-Free Massive MIMO Systems” which revisits the DCC framework in the context of Cell-free Massive MIMO, using the latest analytical methods from the Massive MIMO literature. 

Most importantly, the new paper provides an intuitive way to select the user-centric cooperation clusters based on the uplink pilot transmissions. When a user connects to the network, we suggest that the antenna with the best channel condition is given the responsibility to guarantee the user service. The user is assigned to the pilot that is least affected by pilot contamination in that particular region. Moreover, all antennas serve as many users as there are pilots; at most one user per pilot to limit the negative effect of pilot contamination. Under these assumptions, we show that the users get nearly the same rates as if all the antennas serve all users, but with greatly reduced complexity and fronthaul requirements. In conclusion, scalable and well-performing implementations of Cell-free Massive MIMO are possible!

The following video explains the main ideas:

15 thoughts on “Dynamic Cooperation Clusters”

  1. Respected Professor,

    I hope this message finds you well.

    You are doing a great job especially for young researchers like me.

    I regularly watch every new video. Basically, I need to know about interference avoidance among the dense areas. Could you please recommend some great stuff related to interference avoidance?

    I am looking for your response.

    Thank you so much

    1. I think there is work to be done on spatial interference mitigation in the first two new research directions discussed in this paper:

      Emil Björnson, Luca Sanguinetti, Henk Wymeersch, Jakob Hoydis, Thomas L. Marzetta, “Massive MIMO is a Reality – What is Next? Five Promising Research Directions for Antenna Arrays,” Digital Signal Processing, vol. 94, pp. 3-20, November 2019. (https://arxiv.org/pdf/1902.07678)

  2. Dear Professor,

    I want to thank you for such significant contributions to the topic of massive MIMO.

    I went through the article text, and I would like you to help me understand better some questions I had about Cell-Free Massive MIMO and DCC.

    The Cell-Free Massive MIMO provides a lower ultimate rate but allows a 95% chance of getting a decent service. It seems that the approach is suitable for critical IoT applications (LPWAN). Is this way of thinking correct?

    I have the impression that coherency is more straightforward in the cell-based than free-cell Massive MIMO. Since all antenna elements are close, specialized circuitry performs synchronization in locus, ensuring the phase coherence. Are there solutions for this in Cell-Free? If yes, how they can provide synchronization between too far elements? Can timing/sync work on a distributed algorithm fashion?

    I am looking forward to your feedback!

    Best Regards.

  3. I still haven’t got the main difference between cell-free massive MIMO and CoMP. There are lots of work been done in 3GPP for CoMP. If I want to introduce cell-free massive MIMO into 3GPP standardization, do you think which work I can do? Thanks!

    1. Cell-free massive MIMO is a type of CoMP, similarly to that massive MIMO is a type multi-user MIMO. I am unsure to what extent the 3GPP standardization must change and to what extent it is more of an implementation challenge. I would say that CoMP conventionally is a way to take an existing cellular network and add base station cooperation to it. In contrast, cell-free massive MIMO is an attempt to answer the questions “What if we would be build a distributed MIMO system with CoMP functionality from scratch? What is the best way of doing that, to achieve an architecture where the physical and MAC layer processing are scalable to arbitrarily large network deployments?”

      1. Dear Björnson,

        Thanks for the good post.

        If Cell-free massive MIMO is a type of CoMP, there is nothing to do from 3GPP specification? In particular, LTE CoMP should allow deployment of many TRPs per a cell and TRPs within a cell or in different cells can be coordinated to serve some UEs simultaneously.

        It is not clear to me why Cell-free massive MIMO is for B5G or 6G if we know that it can be realizable by LTE CoMP. If the research is to find efficient way to implement CoMP, it does not seem a good research topic? In fact, https://www.artemis.com/pcell has a very good cell-free implementation already.

        For user-centric coordination cluster, who would decide the cluster serving a UE? I think cell-free still have the cells which are necessary for idle mode operation, time/frequency sync and random access ect. It is just coordination to remove the cell boundary during data communication in connected mode. So if the UE itself determines the coordination cluster, how does the network know?

        Thank you!

        1. I cannot tell how good the pCell implementation is. I’ve seen a few videos with lab experiments and read their white paper. However, it has been rather silent about the company since 2017, so it seems that they were not successful in selling and developing their products. It seems to be a “Level 4” type of implementation (according to my taxonomy in “Making Cell-Free Massive MIMO Competitive With MMSE Processing and Centralized Implementation”).

          There is indeed a lot of connection between LTE CoMP and cell-free Massive MIMO, just as there is a lot of connection between conventional MU-MIMO and co-located Massive MIMO. LTE CoMP was not particularly successful at the time it was developed. It is hard to know how well the latest CoMP implementations perform since the algorithms are proprietary. CoMP has traditionally been network-centric instead of user-centric, so only users that happen to be in the middle of a cooperation cluster is experiencing a cell-free-like situation without inter-cell interference.

          Broadly speaking, Cell-free Massive MIMO is an attempt to build CoMP-like methods (network MIMO) in an optimal way to achieve a user-centric cell-free architecture. Instead of starting from LTE and adding CoMP functionality on top of it, we start from the cell-free architecture and then add all the features to it step-by-step.

          1. Thanks much for the replies.

            As you discussed, LTE CoMP is based on network-centric coordination i.e., the coordination clusters are fixed and disjoint – do you happen to know any pointer to 3GPP spec about this?

            Furthermore, is it correct to say that we can achieve cell-free by changing network-centric coordination to user-centric coordination?

            For user-centric coordination, is it performed by UE or network? (I asked a question earlier but did not get your view).

            I appreciate your feedback.

          2. I don’t have a pointer to the 3GPP spec.

            The intention with user-centric coordination is that every UE should be served by all the access points within its area of influence. Exactly how to implement that is up to the implementation. One option is to let the UE pick access points, and then let the network confirm them. Another option is to let the network make the selection based on measurements from uplink transmissions. I don’t think the best implementation is known.

  4. Thanks Prof. Björnson for further info.

    Your post on massive MIMO vs. MU-MIMO is very insightful (http://ma-mimo.ellintech.se/2017/10/17/six-differences-between-mu-mimo-and-massive-mimo) – Massive MIMO is an practical implementation to achieve expected gain of MU-MIMO by utilizing channel hardening and favorable propagation.

    Based on your discussion earlier, it seems that cell-free massive MIMO is an implementation to achieve expected gain of distributed MIMO (e.g., CoMP). Do you mind sharing X-differences between cell-free massive MIMO and distributed MIMO? Thank you.

      1. Thank you. In 5G Rel-16, multiple TRPs, where each TRP can have massive MIMO, can coordinate to serve multiple UEs simultaneously (aka 5G CoMP). Hence, it would be good if we can make assumption that each AP/TRP can have massive MIMO in distributed MIMO/CoMP when comparing with cell-free massive MIMO. I look forward to your list.

  5. Dear Professor Björnson,

    I am currently studying the concept of cell-free massive MIMO and am eager to implement it. In one of your publications, specifically “Performance of Cell-Free Massive MIMO with Rician Fading and Phase Shifts,” you conducted a study on the Rician channel for the AP-UE link.

    I am interested in applying this concept to the link between ground-based User Equipment (UE) and an airborne Access Point (AP). This specific link is described using the Rician channel model.

    If I follow your simulation described in the “functionCellFreeSetup.m,” I believe that “The height of the Access Point” in my case would represent the altitude at which the airplane flies. Additionally, for “The height of the User Equipment,” I assume I can adopt the value you used (1.5m).

    The mathematical representation of the received signal in cell-free massive MIMO appears analogous to the “classical” MIMO: y=h⋅x+n. As such, it seems reasonable to assume that implementing my cell-free massive MIMO setup may not require entirely new constructs. In the context of MATLAB simulation, the representation remains akin to the standard MIMO equation, y=h⋅x+n, albeit with multiple AP and UE. This includes the selection of the most favorable AP for each UE based on channel conditions.

    Thank you for your comprehensive work in this area. I would greatly appreciate any guidance or insights you can provide.

    1. Yes, you can change the AP height as you described.

      I agree that the MIMO equations are the same in cell-free massive MIMO and “classical” MIMO. The difference is in the channel modeling and practical constraints such as serving each UE by a subset set of the antennas, different power constraints per AP, and possibly the use of distributed precoding/combining methods instead of centralized methods (which are the one most similar to “classical” MIMO).

Leave a Reply

Your email address will not be published. Required fields are marked *