Pith. sign in

REVIEW 4 major objections 5 minor 21 references

Deep Reinforcement Learning-based Cell DTX/DRX Configuration for Network Energy Saving

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A learned agent picks per-cell sleep-cycle settings, saving up to 45% of radio energy while keeping QoS loss near 1%.

desk verdict A practical RL method for Rel-18 cell DTX/DRX with a good reward idea, but the quantitative claims lean on on-evaluation-set tuning and an untested independence assumption. read the letter →

arxiv 2507.21385 v1 pith:YXOQQ45H submitted 2025-07-28 cs.NI cs.AI

Wei Mao , Lili Wei , Omid Semiari , Shu-ping Yeh , Hosein Nikopour This is my paper · ORCID
classification cs.NIcs.AI
keywords cellDTX/DRXnetworkenergysavingdeepreinforcementlearningDQNcontextualbandit5GRANnear-real-timecontrollerapplicationQoS-delaytrade-off
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

3GPP Release 18 lets a 5G cell periodically go silent between bursts of transmission, but choosing how long the silent gap should be trades energy savings against packet delay. The paper claims that a deep reinforcement learning agent, trained as a one-step decision maker (a contextual bandit) with a carefully smoothed reward, can pick the best cycle length and active duration for each cell under changing traffic. In a 21-cell system-level simulation, the agent achieves up to about 45% energy saving in light load and at least 22% in medium load, while keeping the average data-rate loss at or below 1%. The practical stake is that the same agent could run as a near-real-time controller application, cutting radio access network energy consumption without violating delay-sensitive service guarantees.

What carries the argument

The load-bearing design is the contextual-bandit decomposition: because consecutive observation windows of about one second are treated as nearly independent, the reinforcement-learning problem collapses to a one-step expected-reward estimate, and a deep Q-network with experience replay learns the mapping from RAN observations to the best cell DTX/DRX action. The second load-bearing piece is the approximated QoS reward, a smooth version of a discontinuous threshold reward. The threshold reward is $-ax$ when the delivered-data ratio $y$ meets the target $y_0$, and $y-b$ otherwise; the smooth approximation is $r_{\mathrm{QoS\_approx}} = -\frac{u[1+(\alpha-1)(1-y)]+x}{u+1}$ with $u=\left(\frac{1-y}{(1-y_0)(1-x)}\right)^m$, which behaves like a QoS-only reward when the failed-data ratio is large and like an energy-only reward when QoS is satisfied. Together with the state vector of traffic and transmission-condition measurements, these two mechanisms carry the argument.

What would settle it

Measure lag-1 autocorrelation of per-cell PRB utilization across successive 1-second windows in a real or high-fidelity trace; if the autocorrelation is non-negligible (say above 0.1-0.2), the bandit decomposition is violated and the agent's chosen configuration will deviate from the brute-force optimum. Alternatively, run the trained agent on a trace with 10-60 second bursts (e.g., video-conference sessions) and compare its average reward against an oracle that knows the future traffic; a systematic gap larger than a few percent would refute the 'always selects the best configuration' claim.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the cell DTX/DRX configuration problem—choosing cycle length and on-duration for each cell—can be posed as a contextual bandit and solved by a deep Q-network, provided the reward is a smooth approximation of a discontinuous 'QoS first, then energy' reward. The agent observes per-cell traffic intensity, inter-arrival statistics, packet-size statistics, delay requirements, and transmission capability, and outputs one of a finite set of RRC configurations, including the always-on choice. The smooth reward function makes training stable, and the simulation results show the trained agent keeps data-rate loss within roughly 1% on average while saving up to about 45% energy under light load and 22% under medium load, compared with no cell DTX/DRX. The paper presents the agent as one that 'always tries to select the best possible configuration under any network and traffic conditions,' which is the core claim to be tested.

Load-bearing premise

The whole scheme stands on treating consecutive ~1-second RAN observations as nearly independent, so the next best configuration depends only on the current window; if real traffic has longer-range autocorrelation, the learned Q-values will be biased, and the reported savings also ignore SSB, paging, and system-information overhead during silent periods.

Editorial extensions

If this is right

  • A near-real-time controller application could deploy the trained agent, periodically reconfiguring each cell's DTX/DRX parameters from RAN measurements over the E2 interface.
  • Operators can expect the largest energy savings (about 45%) at light load, where the sleep gaps are longest, and still save about a fifth of energy at medium load.
  • The smoothed reward makes the approach robust to measurement noise, so the agent does not need per-scenario tuning of the QoS-versus-energy trade-off coefficient.
  • Setting a random start offset per cell in inference mode staggers the active periods of neighboring cells, reducing inter-cell interference compared with synchronized cycles.
  • The same framework should extend to other time-domain energy-saving parameters, provided the observation window is long enough for nearly independent states.

Reading between the lines

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

  • Editorial inference: The bandit assumption is testable and fragile; if real deployments show bursty traffic at 10-second or longer timescales, the one-step DQN would need to be replaced by an MDP with longer memory, since the paper's own justification rests on weak correlations between consecutive 1-second observations.
  • Editorial inference: Because the simulations ignore SSB, paging, and system-information broadcast during non-active periods, the claimed savings are an upper bound; accounting for these transmissions would reduce the net energy gain, especially at very light load where fixed signaling cost is a larger fraction.
  • Editorial inference: The start offset could become a learning target rather than a randomizer; a multi-cell extension that jointly optimizes offsets to coordinate interference would likely improve both QoS and energy over the single-cell policy.
  • Editorial inference: The reward formulation suggests a general recipe for other delay-sensitive energy-saving features: use a smooth approximation of a lexicographic 'QoS first, then energy' objective to stabilize training, and evaluate against an average QoS degradation budget rather than a hard threshold.
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.

Referee Report

4 major / 5 minor

Summary. The paper addresses configuration optimization for 3GPP Release 18 cell DTX/DRX, a time-domain network energy saving feature. The authors formulate the problem as a per-cell choice of DTX/DRX cycle length and on-duration, and propose a DQN trained on a contextual-bandit (CB) model. A smooth approximation of a QoS-threshold reward is designed to balance delivered data ratio and power consumption. The agent is intended to run as an O-RAN xApp. Using a Python system-level simulator with 7 sites and 21 cells, 500 training episodes, and 10 deployment episodes (210 cells), the paper reports up to about 45% energy saving in light load and at most about 1% data-rate loss compared with the no-DTX/DRX baseline. The central claims are that the agent 'always tries to select the best possible' configuration under any network and traffic conditions, and that the approximated QoS reward is more stable and more QoS-resilient than a linear reward.

Significance. If the claims were fully supported, the contribution would be practically valuable: an automated xApp that configures Rel-18 cell DTX/DRX based on RAN observations could reduce RAN energy consumption while protecting delay-sensitive traffic. The paper contains useful ingredients: a plausible formulation of the configuration problem, a smooth reward function that mitigates discontinuity-induced training instability, a realistic system-level simulator, and a comparison across light, medium, and heavy load categories. The authors also honestly acknowledge some limitations, such as ignoring SSB/paging overhead in Footnote 1 and noting the cost of longer simulation in Footnote 3. However, the quantitative headline results rest on an untested independence assumption, reward constants tuned on the same evaluation data, and a QoS measure that is reported through data rate rather than the delivered-data-ratio objective. These issues are fixable but are load-bearing for the paper's strongest claims, so the current evidence falls short of what the abstract and conclusions assert.

major comments (4)
  1. [Section III.A and Section IV] The contextual-bandit assumption is asserted without evidence and is internally tensioned by the evaluation protocol. The text claims that consecutive ~1 s RAN observation states are 'approximately independent', which justifies treating the problem as a CB. Training episodes are indeed independent one-step SLS runs, so training states are i.i.d. by construction. However, inferencing episodes consist of 10 consecutive 1 s steps in the same deployment, with per-UE packet size, inter-arrival time, and delay requirement drawn once at episode start (Table IV). The state features therefore contain persistent, cell-specific components across the 10 steps, and the DTX/DRX action at step t determines which packets are delayed into step t+1, creating temporal coupling between action, next state, and reward. A one-step greedy DQN does not account for this delayed effect, so the learned Q-values can be biased relative to the true multi-step control problem. The claim that the agent 'always tries to select the best possible cell DTX/DRX configuration' is not established. I recommend either an autocorrelation analysis of state features across consecutive inferencing steps, a comparison against a full MDP/DQN baseline, or an explicit redefinition of the claim to cover only the one-step setting.
  2. [Section IV, reward tuning and statistical evaluation] The reward constants c=0.75, y0=0.9, m=2, alpha=3 are selected by comparing performance on the same 10 inferencing episodes (210 cells) that generate the headline numbers, and no confidence intervals are reported. Section IV states: 'The best coefficient/constants we found from these comparisons are ...' and then uses those same deployments to report the ~45% energy saving and ≤1% data-rate loss. This is model selection on the evaluation set, so the headline point estimates are optimistically biased. Footnote 3 merely asserts that 210 cells are 'statistically representative enough' without supporting evidence. Please report results on a holdout set, provide confidence intervals or variability measures across the 10 episodes, and account for the number of constant combinations tried. Without this, the phrases 'always maintaining no more than ~1% QoS degradation' and 'up to ~45% energy saving' are not statistically supported.
  3. [Section IV, Figs. 6-7 and Eq. (2)] There is a mismatch between the QoS metric used in the reward and the QoS metric reported in the results. The reward and the problem formulation use the delivered data ratio y (Eq. 2), which counts data in packets received within their delay requirement. The reported QoS result, however, is 'achieved data rate' (Fig. 7). Data rate can remain high even when packets miss their latency deadlines, because delayed packets may still be transmitted and counted in throughput. The conclusion states that the agent maintains 'no more than ~1% degradation of the achieved data rate on average', which is not the same as protecting the delay-sensitive QoS that the whole reward design is built around. Please report the delivered data ratio y and/or the delay-violation ratio for the same experiments, and qualify the 'QoS degradation' claim accordingly.
  4. [Footnote 1 and Section II.B] The energy-saving numbers are computed under a power model that ignores the mandatory residual signaling during non-active periods, namely SSB transmission, random access, paging, and system information broadcast. Footnote 1 justifies this by saying the periodicity considered is much less than 160 ms, but the simulator does not report the actual non-active durations or the power cost of these residual transmissions. Since the paper's headline claims are about absolute energy saving percentages, this omission could materially overstate the gains. I ask the authors to either incorporate a minimal model of residual signaling into the power consumption calculation, or to report a sensitivity analysis showing the impact on the projected savings for representative SSB/paging periodicities.
minor comments (5)
  1. [Section III.B] Equation (6) is introduced with the example constants y0=0.9, m=2, alpha=2, but Section IV reports the best constants as y0=0.9, m=2, alpha=3. Please clarify whether the example in Section III.B is superseded by the tuned value, and state whether alpha=2 was also tried.
  2. [Section III.A] The action space is defined only through the constraints 'cycle length smaller than the minimum latency requirement' and 'on-duration smaller than cycle length'. Because Table IV gives delay requirements of 50/75/100 ms, the effective set of cycle lengths is much smaller than the full Table I. Please list the actual (cycle length, on-duration) pairs used in training and inference so the reader can assess the search space and the meaning of the Q-network output.
  3. [General presentation] The manuscript contains several OCR-style symbol corruptions, for example the author email 'hosein.nikopour@intel.co m', the subscripts in Eqs. (1)-(3), and the subscripted reward notation r_QoS_approx. These make the mathematical definitions harder to read and should be cleaned up in the final version.
  4. [Section IV, Fig. 5] The convergence plot shows mean Q-values only. Adding a plot of the training reward (or loss) over episodes, and also reporting the number of training steps per episode, would make the convergence claim easier to verify.
  5. [Section V] The conclusion says the agent maintains 'no more than ~1% degradation of the achieved data rate on average'. The word 'average' is in tension with 'always' in the same sentence and with the abstract's 'always maintaining'. Please choose precise quantifiers and use them consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation chain is self-contained; the CB independence assumption and reward-constant tuning are validity risks, not circular reductions.

full rationale

The paper's claimed result—a DQN/CB agent saving up to ~45% energy with no more than ~1% data-rate loss—rests on system-level simulation with reward functions defined in Eqs. (4)–(6), not on any input reused as the output. The reward functions are expressed in terms of measured delivered-data ratio y and normalized power x, and the reported power and data-rate figures are simulator outcomes computed after training, not quantities encoded in the reward constants. The QoS-first behavior of the approximated QoS reward is by design, but a reward objective is not the same as a measured outcome; the paper still needs, and provides, simulation evidence that the trained agent attains small data-rate loss. Section III.A's contextual-bandit independence assumption is an unsupported modeling assumption, and the inferencing protocol (10 consecutive 1 s steps with per-UE traffic parameters drawn once at episode start) introduces state correlation that the CB model ignores; this is a correctness and robustness concern, not circularity, because the conclusion is not assumed in the premise. The only self-citation, [21], appears in future-work discussion and is not load-bearing. The reward constants are tuned by comparing performance on the same 10 inferencing episodes later used for the headline numbers; this is selection on the evaluation set and weakens the evidential value of the quantitative claims, but it does not make any equation reduce to its own input or rename a fitted parameter as a prediction. Accordingly, no circular step meets the quoted-evidence bar, and the derivation chain is self-contained apart from methodological limitations that belong under correctness risk rather than circularity.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

The reported energy-saving and QoS results depend on a chain of modeling assumptions (power model, ignored signaling, CB state independence) and on reward constants that are tuned to the evaluation scenarios. No new physical entities are posited. The ledger is dominated by domain assumptions and hand-set hyperparameters rather than fitted physical parameters.

free parameters (5)
  • Reward coefficient c (linear reward) = 0.75
    Selected by comparing agent performance on the evaluation scenarios; reported in Section IV as the best value.
  • QoS threshold y0 (approximated QoS reward) = 0.9
    Chosen/tuned on the evaluation scenarios; controls the boundary between QoS-first and energy-first behavior.
  • Smoothness exponent m = 2
    Tuned; with m=2 the approximation is not the 'large m' limit described in Section III.B.
  • Reward amplitude constant alpha = 3
    Tuned in Section IV; alpha=2 is also suggested as an example value.
  • Exploration epsilon schedule = 0.9 to 0.05, decay exponent ~50
    Hand-set, standard for epsilon-greedy; affects training dynamics.
assumptions (6)
  • domain assumption Consecutive ~1s observation periods are approximately independent, reducing multi-step RL to a contextual bandit.
    Section III.A; asserted on observation, no supporting data shown.
  • domain assumption Signaling overhead (SSB, paging, system information) during non-active periods is negligible for energy accounting.
    Section II footnote 1; authors note their periodicities are <160 ms but do not model the energy cost.
  • domain assumption Power consumption model: P_DL(s)=110+90s and sleep-mode parameters from [16] apply.
    Section II.B, Eq. (1); taken from industry 3GPP model, not derived in this paper.
  • ad hoc to paper QoS-threshold reward (5) is the theoretically optimal objective.
    Section III.B calls it optimal without proof; it is a hand-designed reward that encodes a policy preference.
  • domain assumption DL-only, TTI=1 ms, slot offset=0, and only RRC signaling are sufficient.
    Section II.A simplifications; limit generality to downlink and coarse timing.
  • ad hoc to paper The 210-cell evaluation set is statistically representative.
    Section IV footnote 3; claimed without confidence intervals or statistical tests.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Reinforcement Learning-based Cell DTX/DRX Configuration for Network Energy Saving." pith.science (2026). https://pith.science/paper/YXOQQ45H

@misc{pith2026250721385,
  author       = {Pith},
  title        = {Pith review of: Deep Reinforcement Learning-based Cell DTX/DRX Configuration for Network Energy Saving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YXOQQ45H}},
  note         = {Machine review of arXiv:2507.21385}
}
read the original abstract

3GPP Release 18 cell discontinuous transmission and reception (cell DTX/DRX) is an important new network energy saving feature for 5G. As a time-domain technique, it periodically aggregates the user data transmissions in a given duration of time when the traffic load is not heavy, so that the remaining time can be kept silent and advanced sleep modes (ASM) can be enabled to shut down more radio components and save more energy for the cell. However, inevitably the packet delay is increased, as during the silent period no transmission is allowed. In this paper we study how to configure cell DTX/DRX to optimally balance energy saving and packet delay, so that for delay-sensitive traffic maximum energy saving can be achieved while the degradation of quality of service (QoS) is minimized. As the optimal configuration can be different for different network and traffic conditions, the problem is complex and we resort to deep reinforcement learning (DRL) framework to train an AI agent to solve it. Through careful design of 1) the learning algorithm, which implements a deep Q-network (DQN) on a contextual bandit (CB) model, and 2) the reward function, which utilizes a smooth approximation of a theoretically optimal but discontinuous reward function, we are able to train an AI agent that always tries to select the best possible Cell DTX/DRX configuration under any network and traffic conditions. Simulation results show that compared to the case when cell DTX/DRX is not used, our agent can achieve up to ~45% energy saving depending on the traffic load scenario, while always maintaining no more than ~1% QoS degradation.

Figures

Figures reproduced from arXiv: 2507.21385 by the authors.

Figure 1
Figure 1. Cell DTX/DRX illustration. TABLE I. CELL DTX/DRX RRC PARAMETER VALUES Parameter name Possible values ConfiguraƟon type DTX, DRX, or both AcƟvaƟon status AcƟvated, deacƟvated Cycle length {10, 20, 32, 40, 60, 64, 70, 80, 128, …,10240} ms On-duraƟon Ɵmer {1,…,31}/32 & {1, 2, 3, 4, 5, 6, 8, 10, 20, 30, 40, 50, 60, 80, 100, …, 1600} ms Start offset Any integer ms smaller than cycle length Slot offset {0,1,…,31}/32 ms TA… view at source ↗
Figure 2
Figure 2. Optimal cell DTX/DRX configuration [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Operation of the cell DTX/DRX xApp. Upper: agent design; lower: simplified algorithm flow. Green arrows are for training mode only [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (2 more)
Figure 6
Figure 6. Figure 6: Power consumptions of AI agents and baseline. TABLE IV. SIMULATION PARAMETERS Parameter Value Carrier frequency, channel BW 725 MHz, 5 MHz Subcarrier spacing 15 kHz TTI duraƟon 1 ms Cell transmit power 49 dBm Packet size {125, 150, …, 500} bytes Mean inter-arrival Ɵme …
Figure 7
Figure 7. Figure 7: Achieved data rates of AI agents and baseline [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 21 canonical work pages

  1. [1]

    Green Future Networks: Network Energy Efficiency,

    NGMN Alliance, "Green Future Networks: Network Energy Efficiency," whitepaper, Dec. 2021

    work page 2021

  2. [2]

    SMaRT-5G Project-Sustainable Mobile and RAN Transformation (SMaRT),

    ONF, "SMaRT-5G Project-Sustainable Mobile and RAN Transformation (SMaRT)," Open Networking Foundation, White Paper, 2023

    work page 2023

  3. [3]

    Study on Network Energy Savings,

    3GPP, TR 38.864, "Study on Network Energy Savings," 2022

    work page 2022

  4. [4]

    Network Energy Efficiency Phase 2,

    NGMN Alliance, "Network Energy Efficiency Phase 2," whitepaper, Oct. 2023

    work page 2023

  5. [5]

    Reducing Energy Consumption in LTE with Cell DTX,

    P. Frenger, P. Moberg, J. Malmodin, Y. Jading and I. Godor, "Reducing Energy Consumption in LTE with Cell DTX," in 2011 IEEE 73rd Vehicular Technology Conference (VTC Spring), 2011

    work page 2011

  6. [6]

    A Survey on DRX Mechanism: Device Power Saving From LTE and 5G New Radio to 6G Communication Systems,

    K.-H. Lin, H.-H. Liu, K.-H. Hu, A. Huang and H.-Y. Wei, "A Survey on DRX Mechanism: Device Power Saving From LTE and 5G New Radio to 6G Communication Systems," IEEE Communications Surveys & Tutorials, vol. 25, no. 1, pp. 156-183, 2023

    work page 2023

  7. [7]

    Backhaul-aware small cell DTX based on fuzzy Q-Learning in heterogeneous cellular networks,

    A. De Domenico, V. Savin, D. Ktenas and A. Maeder, "Backhaul-aware small cell DTX based on fuzzy Q-Learning in heterogeneous cellular networks," in 2016 IEEE International Conference on Communications (ICC), 2016

    work page 2016

  8. [8]

    Reinforcement learning for interference-aware cell DTX in heterogeneous networks,

    A. De Domenico and D. Kténas, "Reinforcement learning for interference-aware cell DTX in heterogeneous networks," in 2018 IEEE Wireless Communications and Networking Conference (WCNC), 2018

    work page 2018

Show all 21 references
  1. [9]

    R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed., The MIT Press, 2018

    2018

  2. [10]

    Machine Learning Based C-DRX Configuration Optimization for 5G,

    P. Bruhn and G. Bassi, "Machine Learning Based C-DRX Configuration Optimization for 5G," in Mobile Communication - Technologies and Applications; 25th ITG-Symposium, 2021

    2021

  3. [11]

    On using Deep Reinforcement Learning to balance Power Consumption and Latency in 5G NR,

    K. Boutiba and A. Ksentini, "On using Deep Reinforcement Learning to balance Power Consumption and Latency in 5G NR," in ICC 2023 - IEEE International Conference on Communications, 2023

    2023

  4. [12]

    Radio Resource Control (RRC) protocol specifications,

    3GPP, TS 38.331, "Radio Resource Control (RRC) protocol specifications," 3GPP, Technical Specification Group Radio Access Network, NR, 2024

    2024

  5. [13]

    O-RAN: Towards an Open and Smart RAN,

    O-RAN Alliance, "O-RAN: Towards an Open and Smart RAN," whitepaper, Oct. 2018

    2018

  6. [14]

    Network energy saving techniques in time, frequency and spatial domain,

    CATT-R1-2211210, "Network energy saving techniques in time, frequency and spatial domain," 3GPP TSG RAN WG1#111, Toulouse, France, 2022

    2022

  7. [15]

    Multiplexing and channel coding,

    3GPP, TS 38.212, "Multiplexing and channel coding," 3GPP Technical Specification Group Radio Access Network, NR, 2024

    2024

  8. [16]

    Enabling Network Power Savings in 5G- Advanced and Beyond,

    T. Islam, D. Lee and S. S. Lim, "Enabling Network Power Savings in 5G- Advanced and Beyond," IEEE Journal on Selected Areas in Communications, vol. Vol.41, no. No.6, pp. 1888-1899, June 2023

    2023

  9. [17]

    PyTorch Reinforcement Learning (DQN) Tutorial,

    A. Paszke and M. Towers, "PyTorch Reinforcement Learning (DQN) Tutorial," [Online]. Available: https://docs.pytorch.org/tutorials/intermediate/reinforcement_q_learning .html

  10. [18]

    Adam: A Method for Stochastic Optimization,

    D. P. Kingma and J. Ba, "Adam: A Method for Stochastic Optimization," in International Conference on Learning Representations (ICLR), 2015

    2015

  11. [19]

    Study on Licensed-Assisted Access to Unlicensed Spectrum,

    3GPP, TR 36.889, "Study on Licensed-Assisted Access to Unlicensed Spectrum," 2015

    2015

  12. [20]

    Gymnasium Documentation,

    Farama Foundation, "Gymnasium Documentation," [Online]. Available: https://gymnasium.farama.org/

  13. [21]

    Graph Reinforcement Learning for QoS-Aware Load Balancing in Open Radio Access Networks,

    O. Semiari, H. Nikopour and S. Talwar, "Graph Reinforcement Learning for QoS-Aware Load Balancing in Open Radio Access Networks," 2025. [Online]. Available: https://arxiv.org/abs/2504.19499. 4 Compared to the case without cell DTX/DRX, ICI may increase in a heavied loaded cell...

    2025 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.