Testbed Development: An Intelligent O-RAN based Cell-Free MIMO Network

Alister Burr; David Grace; Hamed Ahmadi; Josh Shackleton; Kanapathippillai Cumanan; Mostafa Rahmani; Yi Chu

REVIEW 4 major objections 5 minor 1 cited by

Testbed Development: An Intelligent O-RAN based Cell-Free MIMO Network

T0 review · 4 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper reports an O-RAN-based cell-free MIMO testbed that serves commercial 5G phones, with a RIC-hosted xApp controlling which antennas serve each user.

desk verdict A real working O-RAN cell-free uplink testbed that serves commercial 5G phones — the engineering is credible, but the 'intelligent' xApp claim is thinner than the title suggests. read the letter →

arxiv 2502.08529 v1 pith:E6QF634Y submitted 2025-02-12 cs.NI cs.SYeess.SY

Yi Chu , Mostafa Rahmani , Josh Shackleton , David Grace , Kanapathippillai Cumanan , Hamed Ahmadi , Alister Burr This is my paper · ORCID

classification cs.NIcs.SYeess.SY

keywords cell-freeMIMOO-RANopenradioaccessnetworktestbedimplementationMU-MIMOxAppdeepQ-network5Guserequipment

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that cell-free MIMO—many distributed access points jointly serving users instead of one cell at a time—can be built directly on the O-RAN architecture and can serve ordinary commercial 5G phones, not just software-defined radios. It reports a working testbed in which four radio units act as a cell-free array, a modified scheduler lets two phones share the same uplink resources, and a RIC-hosted application (xApp) decides in real time which antennas should contribute to each user's reception. The authors claim this is the first CF-MIMO prototype that combines the O-RAN RIC and E2 interface with MU-MIMO service to commercial 5G user equipment. If that holds, cell-free networking stops being a PHY-only research topic and becomes something operators can optimise through the same open interfaces that O-RAN already standardises.

What carries the argument

The load-bearing mechanism is the closed control loop over the E2 interface, the link between the RIC and the RAN. The DU exposes per-antenna channel metrics as custom key performance measurements and accepts a per-UE antenna on/off mask through a custom RAN control parameter; the xApp's deep Q-network turns those measurements into actions, and the DU applies the resulting antenna selection when it builds the PHY payloads. Underneath that loop, the MU-MIMO change is carried by a scheduler that tracks shared resource elements with two bitmaps and a per-element list of user identifiers, and by a zero-forcing equaliser that needs both users' channel estimates to separate their signals. The xApp's reward is constructed from the trade-off that gives the loop its purpose: more antennas improve post-equalisation SNR but add roughly 50 nanoseconds of processing per antenna per PDU, which matters at split-8 timing margins.

What would settle it

Compare the pretrained DQN's antenna choices against exhaustive search on the live testbed for a scenario that was not in the training simulator—for example, two simultaneous antenna faults or a moving UE—and measure whether the chosen antenna subset actually achieves the claimed throughput/processing-time trade-off; a match to random or worse-than-fixed-antenna behaviour would falsify the intelligence claim.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is an integration result: a cell-free MIMO uplink can run inside an O-RAN-compatible private 5G network serving COTS handsets, and the RIC can steer it live. To get there the authors modified an open-source 5G stack so that multiple radio units connect to one DU, changed the MAC scheduler so two UEs can share the same physical resource blocks, added a zero-forcing equaliser that separates the two MU-MIMO streams using both users' channel estimates, and extended the E2 agent with four per-antenna key performance measurements and a RAN control parameter that switches antennas on or off for a specific UE. On top of this, an xApp embeds a pretrained deep Q-network that maps the KPMs to antenna-inclusion decisions. In the demonstration, unplugging an antenna changed the reported CSI, the xApp excluded that antenna from equalisation, and reconnecting it brought the antenna back into service. The paper's measurements show processor power, PUSCH (uplink data channel) processing time, and total UE throughput all improving as the antenna selection adapts.

Load-bearing premise

The load-bearing premise is that the simple simulator used to pretrain the DQN faithfully reproduces the live testbed's trade-off between antenna count, throughput, and processing time, because the paper gives no evidence that the simulator was validated against real measurements.

Editorial extensions

If this is right

Future cell-free prototypes can use commercial phones as evaluation endpoints, so performance claims can be checked with standard application-layer throughput rather than PHY-only signals.
Because the RIC and E2 are part of the testbed, existing O-RAN optimisation services such as energy-saving RU switching and traffic steering can be applied to cell-free networks without new hardware.
The exposed per-antenna KPMs and RAN control parameter provide a working template for standardising antenna-association control in the E2 service models.
The measured numbers—0.4 to 2.6 Mbps per useful antenna, up to about 0.2 microseconds and 1 watt saved per excluded antenna—give concrete anchor points for designing xApp reward functions.
The practical MU-MIMO limit is currently about three UEs because the tested phones only channel-estimate correctly on a subset of the eight type-A DMRS ports, so extending DMRS support is a precondition for denser cell-free MU-MIMO.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If this integration pattern spreads, the natural standardisation step is a CF-MIMO-specific E2 service model so that any vendor's xApp can reconfigure antenna groups on any O-RAN DU that supports the model.
The same KPM-to-RAN-control loop could be reused for closed-loop uplink power control or DU/CPU placement optimisation, since the per-antenna measurement and actuation infrastructure already exists.
A natural stress test, not reported in the paper, would move the UEs or corrupt multiple antennas at once to see whether the pretrained DQN generalises beyond the single-antenna-fault demonstration.
A public release of the simulator's dynamics and reward function, plus a comparison of its decisions against exhaustive antenna-subset search on the live testbed, would make the 'intelligent' contribution independently checkable.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Quick read of arXiv:2502.08529. Worth your time: it's one of the first, possibly the first, CF-MIMO testbed that runs the full O-RAN stack (RIC, E2, KPM/RC) and serves commercial 5G handsets with MU-MIMO on the uplink. The comparison table against seven prior testbeds is useful and the differences are real: those were mostly PHY-only, SDR UEs, continuous transmission, no RIC. Here they modified the srsRAN MAC scheduler to overlap two UEs on the same PRBs, added a ZF equaliser, extended the E2 agent with custom KPMs and a control parameter for antenna selection, and ran an OSC RIC with a DQN xApp. The demo video and the throughput/power/processing-time numbers are consistent with a working system. That is the main result and it stands.

Soft spots, in order of size. First, the 'intelligent antenna association' claim: the DQN was pretrained in a 'simple simulator' described in one sentence, with no dynamics, reward function, or validation against the live testbed. The live demo deactivates physically disconnected antennas — something a fixed RSRP threshold would do. Table III is a sweep over antenna count with single 30-second means, no error bars, no 'all antennas on' baseline, and no comparison against a heuristic. So the evidence shows the xApp reacts sensibly to CSI changes; it does not show the learned policy beats a trivial rule. The authors say the reward encodes the same trade-off (throughput vs processing time) that Table III measures, so there is a whiff of circularity in the 'improvement' claim. Second, the DQN details are thin: they give input and action dimensions, but no architecture specifics, no training hyperparameters, and no code release. Third, minor: the paper says the codebase can be provided upon request — no open repo, which limits reproducibility.

None of this breaks the core paper. The engineering contribution is real, and the limitations are mostly about the AI garnish, not the testbed. The paper is honest about the simulator and the DMRS constraints. For a Communications Magazine testbed paper, this is the right shape. I'd send it to review. The weaknesses are fixable with a static baseline and more DQN detail, but even as-is a serious referee can verify the system-level claims and the table is a useful data point for the community.

Recommendation: accept peer review, with a request for a control baseline and simulator description in revision. I'd cite this as a system reference if I worked on O-RAN testbeds; my own work less so.

Referee Report

4 major / 5 minor

Summary. The paper presents the design and implementation of a cell-free MIMO testbed built on the O-RAN architecture. The authors modify the srsRAN DU to support multiple RUs and UL MU-MIMO, integrate an OSC near-RT RIC with customised E2SM-KPM and E2SM-RC messages, and implement a DQN-based antenna-association xApp that selects which distributed antennas contribute to each UE's equalisation. The testbed serves commercial 5G UEs with a full 5G stack, and the paper demonstrates DL/UL throughput, MU-MIMO operation, and measurements of processor power, total UE throughput, and PUSCH processing time as a function of the number of selected antennas.

Significance. If fully supported, the paper would be a valuable mid-TRL testbed contribution: the combination of a full 5G protocol stack, COTS UE support, O-RAN RIC/E2 integration, and distributed antenna selection has not been shown together in prior CF-MIMO prototypes. The implementation details of the MAC scheduler bitmaps, the ZF equaliser integration, and the custom E2 KPM/control messages are useful to the community, and the demonstration video provides evidence for the core engineering claim. However, the 'intelligent xApp' contribution is not yet established: the DQN pretraining simulator is essentially undescribed, and the live experiments do not compare against a baseline or a heuristic policy. The engineering claim of a working O-RAN CF-MIMO testbed is credible; the learning/optimisation claim needs substantially stronger evidence.

major comments (4)

[Section III-D] The DQN that drives the antenna-association xApp is pretrained in a 'simple simulator', but the manuscript gives no description of that simulator: its state/action space, reward function, channel model, or how it is validated against the live testbed are all absent. Because the reward encodes exactly the trade-off between throughput and processing time that Table III later reports, the reader cannot tell whether the xApp's decisions reflect learned behaviour or a fixed rule embedded in the reward. This is a load-bearing gap for the 'intelligence' claim; please provide the simulator specification and a quantitative comparison of simulator predictions against the measured antenna-utility relationships.
[Section IV-C, Table III] The performance evaluation reports single 30-second means for processor power, total throughput, and PUSCH processing time, with no error bars, number of repetitions, or statistical test. Given that the power differences between consecutive rows are as small as 0.2-1.0 W and PowerTOP is a coarse-grain logger, the monotonic trends may be within measurement noise. More importantly, there is no baseline with the xApp disabled (all antennas active) and no comparison to a heuristic policy such as fixed RSRP thresholding, so the table does not demonstrate that the DQN improves over a trivial rule. Please add repeated trials, confidence intervals, and at least one baseline policy.
[Section IV-B] The live demonstration only shows the xApp deactivating antennas that the experimenter physically unplugged and reactivating them after reconnection. A rule-based policy that compares PUSCH.PORT.RSRP or SNR to a threshold would behave identically in this scenario, so the demonstration does not validate the learned policy. To support the 'intelligent' claim, please include a scenario where the DQN must make non-obvious trade-offs (e.g., marginal antennas with moderate SNR), and compare its decisions against a thresholding or greedy baseline.
[Section I-B] The novelty claim that this is 'the first implementation of CF-MIMO on the O-RAN architecture' and the Table I comparison rely on the interpretation of 'O-RAN architecture'. The paper later notes interoperability limitations with FlexRIC and the use of a customised OSC RIC, so it would be helpful to state explicitly which O-RAN interfaces (E2AP, E2SM-KPM, E2SM-RC, F1, etc.) are fully standard-compliant and which are modified or proprietary, so that the claim can be assessed against prior work.

minor comments (5)

[Section III-D] The text says the non-standardised KPMs are 'mentioned in subsection D', but they are actually described in Section III-C; please correct the cross-reference.
[Section III-D] There is a typo: 'a simple simulator to pretrained this DQN' should read 'to pretrain this DQN'.
[Section II] The statement that 'all 5 RIC report styles are supported' in Section III-D appears to conflict with the earlier note that FlexRIC does not support all RIC report styles with E2SM-KPM; please clarify whether this refers to the OSC RIC only.
[Section IV-A] The text refers to 'the testbed configurations are summarised below in Table III', but the configuration table is Table II; Table III contains the performance measurements.
[Section I-A] The phrase 'O-RAN compatible fronhaul' contains a typo ('fronhaul' should be 'fronthaul'); please also check for similar typos elsewhere.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the testbed and xApp claims are direct engineering demonstrations, not derived predictions, and the under-described pretraining simulator is a validation gap, not a circular reduction.

full rationale

The paper does not present a derivation chain in which an output is constructed from its own inputs. The central claims are that the testbed serves COTS 5G UEs with MU-MIMO on an O-RAN architecture and that an xApp uses a pretrained DQN to select serving antennas; both are supported by direct measurements and a demonstration video rather than by fitting parameters to the reported outcomes. Table III is a set of averaged logged measurements over different antenna selections, not a prediction made from the DQN or from a fitted model. The xApp's DQN was pretrained in a 'simple simulator' whose dynamics and reward are not specified, and the live evaluation lacks a heuristic or all-antennas-on baseline; this is a legitimate reproducibility and validation weakness, and the manuscript itself states 'We have implemented a simple simulator to pretrain this DQN since training with a live system would take a significant amount of time.' However, no equation, fitted parameter, or self-citation is shown to reduce the reported results to the training reward or to prior work by the same authors. The only self-citation, [14], is used for a ZF equaliser implementation point and is not load-bearing. Therefore no circular step meeting the required evidentiary standard is present; the appropriate finding is no significant circularity, with the noted validation limitations belonging to correctness risk rather than circularity.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central testbed claim rests on domain-level hardware and software assumptions (srsRAN modifications, synchronization) and on an unverified simulator used to train the xApp's DQN. No new physical entities are introduced.

free parameters (1)

DQN reward function weights (throughput vs processing time/energy trade-off) = unspecified
The reward is described qualitatively as a trade-off between antenna count, SNR, and processing time, but the weights or functional form are not given. These choices control the xApp's learned policy.

assumptions (3)

domain assumption The modified srsRAN DU remains 3GPP-compliant enough to serve commercial UEs.
Section III describes deep modifications to the MAC scheduler and PHY; the paper asserts COTS UEs attach and work, but no protocol conformance test is described.
domain assumption Octoclock-distributed 10 MHz and 1 PPS synchronization is sufficient for distributed RU operation.
Section II states all USRPs are synchronized via Octoclock; the CF-MIMO uplink operation depends on this timing without presenting a synchronization-error analysis.
ad hoc to paper The pretraining simulator for the DQN is a faithful proxy for the live testbed.
Section III-D introduces the simulator in one sentence without details; the xApp's live decisions rely on the policy learned in this simulator.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Testbed Development: An Intelligent O-RAN based Cell-Free MIMO Network." pith.science (2026). https://pith.science/paper/E6QF634Y

@misc{pith2026250208529,
  author       = {Pith},
  title        = {Pith review of: Testbed Development: An Intelligent O-RAN based Cell-Free MIMO Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E6QF634Y}},
  note         = {Machine review of arXiv:2502.08529}
}

read the original abstract

Cell-free multiple input multiple output (CF-MIMO) systems improve spectral and energy efficiencies using distributed access points (APs) to provide reliable service across an area equivalent to multiple conventional cells. This paper presents a novel design and implementation of a CF-MIMO network leveraging the open radio access network (O-RAN) architecture based testbed to enhance the performance of interference-prone user. The proposed prototype is developed based on open source software components and unlike many other prototypes, our testbed is able to serve commercial 5G user equipment (UE). The RAN intelligent controller (RIC) allows the cell-free (CF) network to access the embedded artificial intelligence and benefit from the network optimisation techniques that O-RAN brings. The testbed includes an intelligent antenna association xApp which determines the antenna group that serves each UE based on the live key performance measurements. The paper demonstrates the deployment and operation of the CF network and the xApp and discusses how the CF networks can benefit from the O-RAN architecture.

Figures

Figures reproduced from arXiv: 2502.08529 by the authors.

**Figure 2.** Original and modified MAC scheduler logic [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Testbed deployment by 1 W (or less) and the PUSCH processing time was reduced by 0.2 us (or less). With each useful antenna included in the equaliser, the total throughput gained 0.4 to 2.6 Mbps. The results show that the xApp can adapt to the changing configuration and improve all three measurements. TABLE III: Testbed Performance Measurements Antennas selected Processor power consumption (W) Total UE throughput (M… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Enhancing Open RAN Digital Twin Through Power Consumption Measurement
eess.SP 2025-07 conditional novelty 6.0 of 10

Measurements of O-RAN component power consumption under varying PRB load show that power draw is largely load-invariant across Split 8 and Split 7.2b hardware.

Reference graph

Works this paper leans on

15 extracted references · 13 canonical work pages · cited by 1 Pith paper

[1]

Cell-free mMIMO support in the O-RAN architecture: A PHY layer perspective for 5G and beyond networks,

V. Ranjbar, A. Girycki, M. A. Rahman, S. Pollin, M. Moonen, and E. Vinogradov, “Cell-free mMIMO support in the O-RAN architecture: A PHY layer perspective for 5G and beyond networks,” IEEE Commu- nications Standards Magazine, vol. 6, no. 1, pp. 28–34, 2022

work page 2022
[2]

Making cell-free massive MIMO competitive with MMSE processing and centralized implementation,

E. Bjo¨rnson and L. Sanguinetti, “Making cell-free massive MIMO competitive with MMSE processing and centralized implementation,” IEEE Transactions on Wireless Communications, vol. 19, no. 1, pp. 77– 90, 2019

work page 2019
[3]

The road towards 6G: A comprehensive survey,

W. Jiang, B. Han, M. A. Habibi, and H. D. Schotten, “The road towards 6G: A comprehensive survey,” IEEE Open Journal of the Communications Society, vol. 2, pp. 334–366, 2021

2021
[4]

Interde- pendent cell-free and cellular networks: Thinking the role of cell-free architecture for 6G,

Y. Zhang, Q. Hu, M. Peng, Y. Liu, G. Zhang, and T. Jiang, “Interde- pendent cell-free and cellular networks: Thinking the role of cell-free architecture for 6G,” IEEE Network, 2023

work page 2023
[5]

Intelli- gence and learning in O-RAN for data-driven nextG cellular networks,

L. Bonati, S. D’Oro, M. Polese, S. Basagni, and T. Melodia, “Intelli- gence and learning in O-RAN for data-driven nextG cellular networks,” IEEE Communications Magazine, vol. 59, no. 10, pp. 21–27, 2021

work page 2021
[6]

O-RAN near-RT RIC architecture 4.0,

O-RAN, “O-RAN near-RT RIC architecture 4.0,” ORAN. WG3.RICARCH- R003-v04.00 Technical Specification, Tech. Rep.,

work page
[7]

Hardware implementation of cell-free MIMO and dynamic TDD using the OAI 5G NR codebase,

H. Kamboj, B. Anand, S. Gupta, A. Meshram, S. Balijepalli, and C. R. Murthy, “Hardware implementation of cell-free MIMO and dynamic TDD using the OAI 5G NR codebase,” in WSA 2021; 25th International ITG Workshop on Smart Antennas. VDE, 2021, pp. 1–5

work page 2021
[8]

SDR implementation of a fully distributed cell-free MIMO system,

M. H. Lee, M. C. Ha, H. S. Yoon, C. W. Yun, C. W. Yu, and K. W. Choi, “SDR implementation of a fully distributed cell-free MIMO system,” in 2022 IEEE International Conference on Communications Workshops (ICC Workshops). IEEE, 2022, pp. 1–2

work page 2022

Show all 15 references

[9]

Fully distributed cell-free MIMO systems: Architecture, algorithm, and testbed experiments,

M. H. Lee, C. Yun, G. H. Kim, S. Y. Park, C. W. Yu, and K. W. Choi, “Fully distributed cell-free MIMO systems: Architecture, algorithm, and testbed experiments,” IEEE Internet of Things Journal, 2023

2023
[10]

Full-spectrum cell-free RAN for 6G systems: System design and experimental results,

D. Wang, X. You, Y. Huang, W. Xu, J. Li, P. Zhu, Y. Jiang, Y. Cao, X. Xia, Z. Zhang et al., “Full-spectrum cell-free RAN for 6G systems: System design and experimental results,” Science China Information Sciences, vol. 66, no. 3, p. 130305, 2023

2023
[11]

Experimental performance evaluation of cell- free massive MIMO systems using COTS RRU with OTA reciprocity calibration and phase synchronization,

Y. Cao, P. Wang, K. Zheng, X. Liang, D. Liu, M. Lou, J. Jin, Q. Wang, D. Wang, Y. Huang et al., “Experimental performance evaluation of cell- free massive MIMO systems using COTS RRU with OTA reciprocity calibration and phase synchronization,” IEEE Journal on Selected Areas in...

2023
[12]

Implementa- tion of a cloud-based cell-free distributed massive MIMO system,

D. Wang, C. Zhang, Y. Du, J. Zhao, M. Jiang, and X. You, “Implementa- tion of a cloud-based cell-free distributed massive MIMO system,” IEEE communications magazine, vol. 58, no. 8, pp. 61–67, 2020

2020
[13]

S38.212. NR; Multiplexing and channel coding, v18.2.0,

3GPP, “S38.212. NR; Multiplexing and channel coding, v18.2.0,” Tech- nical Specification, 2024

2024
[14]

Deep reinforcement learning-based sum rate fair- ness trade-off for cell-free mMIMO,

M. Rahmani, M. Bashar, M. J. Dehghani, A. Akbari, P. Xiao, R. Tafa- zolli, and M. Debbah, “Deep reinforcement learning-based sum rate fair- ness trade-off for cell-free mMIMO,” IEEE Transactions on Vehicular Technology, vol. 72, no. 5, pp. 6039–6055, 2022. ACKNOWLEDGEMENTS Thi...

2022
[2024]

Available: https://orandownloadsweb.azurewebsites.net/ specifications

[Online]. Available: https://orandownloadsweb.azurewebsites.net/ specifications

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [1]

Cell-free mMIMO support in the O-RAN architecture: A PHY layer perspective for 5G and beyond networks,

V. Ranjbar, A. Girycki, M. A. Rahman, S. Pollin, M. Moonen, and E. Vinogradov, “Cell-free mMIMO support in the O-RAN architecture: A PHY layer perspective for 5G and beyond networks,” IEEE Commu- nications Standards Magazine, vol. 6, no. 1, pp. 28–34, 2022

work page 2022

[2] [2]

Making cell-free massive MIMO competitive with MMSE processing and centralized implementation,

E. Bjo¨rnson and L. Sanguinetti, “Making cell-free massive MIMO competitive with MMSE processing and centralized implementation,” IEEE Transactions on Wireless Communications, vol. 19, no. 1, pp. 77– 90, 2019

work page 2019

[3] [3]

The road towards 6G: A comprehensive survey,

W. Jiang, B. Han, M. A. Habibi, and H. D. Schotten, “The road towards 6G: A comprehensive survey,” IEEE Open Journal of the Communications Society, vol. 2, pp. 334–366, 2021

2021

[4] [4]

Interde- pendent cell-free and cellular networks: Thinking the role of cell-free architecture for 6G,

Y. Zhang, Q. Hu, M. Peng, Y. Liu, G. Zhang, and T. Jiang, “Interde- pendent cell-free and cellular networks: Thinking the role of cell-free architecture for 6G,” IEEE Network, 2023

work page 2023

[5] [5]

Intelli- gence and learning in O-RAN for data-driven nextG cellular networks,

L. Bonati, S. D’Oro, M. Polese, S. Basagni, and T. Melodia, “Intelli- gence and learning in O-RAN for data-driven nextG cellular networks,” IEEE Communications Magazine, vol. 59, no. 10, pp. 21–27, 2021

work page 2021

[6] [6]

O-RAN near-RT RIC architecture 4.0,

O-RAN, “O-RAN near-RT RIC architecture 4.0,” ORAN. WG3.RICARCH- R003-v04.00 Technical Specification, Tech. Rep.,

work page

[7] [7]

Hardware implementation of cell-free MIMO and dynamic TDD using the OAI 5G NR codebase,

H. Kamboj, B. Anand, S. Gupta, A. Meshram, S. Balijepalli, and C. R. Murthy, “Hardware implementation of cell-free MIMO and dynamic TDD using the OAI 5G NR codebase,” in WSA 2021; 25th International ITG Workshop on Smart Antennas. VDE, 2021, pp. 1–5

work page 2021

[8] [8]

SDR implementation of a fully distributed cell-free MIMO system,

M. H. Lee, M. C. Ha, H. S. Yoon, C. W. Yun, C. W. Yu, and K. W. Choi, “SDR implementation of a fully distributed cell-free MIMO system,” in 2022 IEEE International Conference on Communications Workshops (ICC Workshops). IEEE, 2022, pp. 1–2

work page 2022