Latency Decoupling in Low-Feedback Multi-User Networks via Overhearing-Driven NOMA

Ahmed Badawy; Amr Mohamed; Mohsen Abedi

REVIEW 5 major objections 7 minor 21 references

Latency Decoupling in Low-Feedback Multi-User Networks via Overhearing-Driven NOMA

T0 review · 5 major / 7 minor · reviewed 2026-07-31 · grok-4.5

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

Pith's one-line read Overhearing and ACK timing let a transmitter finish multi-user deliveries faster without channel knowledge, by decoupling strong users from the weakest link.

desk verdict Clean two-user protocol with real gains in the asymmetric high-rate regime; the abstract’s 50% multi-user claim is unsupported and should be cut or qualified. read the letter →

arxiv 2607.24609 v1 pith:LX3FF4G6 submitted 2026-07-27 cs.IT math.ITmath.OC

Mohsen Abedi , Ahmed Badawy , Amr Mohamed This is my paper · ORCID

classification cs.ITmath.ITmath.OC

keywords low-feedbacknetworksmulti-userlatencyRLNCsymbol-awareNOMAoverhearingACKtimingcompletiontimepowerallocation

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

In low-feedback wireless settings such as non-terrestrial links and massive IoT, the transmitter often cannot get frequent ACKs or channel state, so a single weak user can force everyone to wait. This paper proposes ONOMA, a two-phase scheme that pairs random linear network coding with symbol-aware non-orthogonal multiple access. In phase one the transmitter sends coded packets aimed at one user while others overhear; the order and timing of block ACKs reveal who is strong without any CSI. In phase two it superimposes packets so the strong user, already holding the weak user’s data, regenerates and subtracts those symbols and decodes interference-free. Adaptive power is set from the observed ACK time. Analysis and simulations claim this cuts completion time by up to about a third in two-user cases and half in larger asymmetric networks versus TDMA, multicast, FDMA, inter-session coding, and classical NOMA.

What carries the argument

ONOMA: a phased protocol that first builds side information via overhearing of RLNC packets, then runs symbol-aware NOMA in which the strong user regenerates the weak user’s modulated symbols from known coding vectors and cancels them, plus a power-allocation rule driven by inverting the RLNC completion-time formula on observed ACK times.

What would settle it

In a controlled two-user Rayleigh setup with known asymmetric gains, measure whether SR-ONOMA’s measured completion time stays near the paper’s bounds and beats TDMA/SR-MC by the claimed margins when power is set only from the phase-1 ACK time (and t_max, f), versus when the true gains are given to the allocator; a large gap or loss of the 34% gain would refute the claim.

Watch

Extended reading notes

Core claim

The paper claims that overhearing-driven NOMA (ONOMA) minimizes multi-user completion latency without instantaneous or statistical CSI by using ACK timing to order channels and reconstruction-based symbol cancellation so strong users finish without waiting on the weakest user.

Load-bearing premise

A single observed ACK time, inverted through the expected RLNC completion formula plus a chosen maximum time for the weak user, is treated as a good enough estimate of channel strength to set power so latency truly decouples.

Editorial extensions

If this is right

In asymmetric low-feedback downlinks, overall block delivery time need not track the weakest user’s PtP completion time.
Sparse block ACKs can replace explicit CSI for ordering users and setting NOMA power.
Reconstruction-based cancellation after overhearing can replace classical same-slot SIC when side information is built first.
Scheduling for U>2 can alternate TDMA overhearing rounds with ONOMA pairings as early ACKs identify strong users.
Reported gains are largest at higher rates and severe channel asymmetry, and shrink when channels are both strong or rates are very low.

Reading between the lines

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

The same ACK-timing estimator could be stress-tested under mobility or non-block fading, where one shot time-to-ACK may mix path loss with short-term fades.
If header/coding-vector delivery at the lower rate R' fails often, the strong user’s regeneration step breaks; header reliability is an unemphasized single point of failure.
Extending the idea to uplink grant-free or multi-cell settings would need a different overhearing geometry than the downlink PtP model used here.

Share X Bluesky LinkedIn Reddit HN

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

The useful core is a concrete cross-layer recipe for low-feedback multi-user delivery when the transmitter has no CSI: RLNC to the weak user while others overhear, ACK timing to order channels and set a heuristic λ, then symbol-aware NOMA where the strong user regenerates and subtracts the weak user’s symbols instead of doing same-slot decoding SIC. That assembly is new enough, and the two-user analysis is done properly.

They derive explicit E[T] expressions for TDMA, multicast, FDMA, inter-session, classical NOMA, and both JR/SR-ONOMA (Eqs. 15–47), with the usual RLNC rank sums and Rayleigh outage, plus appendices that bound residual ranks with the familiar 3/4 factor. Simulations (300 runs, R in {0.5,1,2}, asymmetric Ω) show clear wins versus TDMA and the better baselines in the high-rate asymmetric cases—up to the stated ~34% in Fig. 6c/f. When both channels are strong or R is low, SR-MC is competitive or better; they mostly show that. Classical NOMA with λ=0.5 is correctly noted as infeasible for R≥1, so that comparison is weak but not hidden.

Soft spots are real but bounded. The abstract and conclusion claim “50% in larger asymmetric networks.” Section IV.E is only a qualitative scheduling sketch; every equation and every figure is U=2. That number is asserted, not measured. The ACK-timing Ω̂ inversion (Eqs. 48–49) plus hand-tuned f=1.5 and t_max is a heuristic; when the strong user finishes early the log can go negative and the paper does not say what Algorithm I does. Perfect symbol regeneration (known M, Rc, identical coding vector) is load-bearing and untested under header errors or imperfect sync. None of this sinks the two-user contribution.

This is for people working low-feedback NTN/IoT scheduling and RLNC-NOMA hybrids. The math and citation pattern look solid; circularity is low. I would send it to referees after the multi-user claim is qualified or backed, the power-policy edge cases are handled, and error bars/code appear. Worth engaging for the two-user protocol; do not cite the 50% number as is.

Referee Report

5 major / 7 minor

Summary. The paper proposes ONOMA, a cross-layer transmission scheme for low-feedback multi-user networks combining RLNC with symbol-aware NOMA. In Phase 1, the transmitter sends RLNC-coded packets to one user while others overhear; the relative ACK timing implicitly reveals channel-strength ordering without CSI. In Phase 2, the strong user, having already decoded the weak user's packets, regenerates the weak user's symbols and cancels them, achieving interference-free decoding and "decoupling" user latencies. An adaptive power-allocation policy (Algorithm I) is derived from channel estimates obtained by inverting the expected-completion-time formula on a single observed ACK time. The authors derive expected completion times for ONOMA and five baselines (TDMA, multicast, FDMA, inter-session coding, classical NOMA) in a two-user Rayleigh-fading setting and report Monte Carlo results (300 runs) showing up to 34% completion-time reduction in the high-rate asymmetric regime; the abstract additionally claims up to 50% in larger asymmetric networks.

Significance. If the results hold, the paper offers a practical, feedback-light mechanism for mitigating weakest-user latency domination in asymmetric multi-user networks, a relevant problem for NTN and massive-IoT settings. Notable strengths: the two-user expected-time derivations for both baselines and ONOMA are explicit and reproducible (Eqs. 15–47, Appendices A–D); the ACK-timing channel inference idea is simple and implementable; and the Monte Carlo study (300 runs, sweep over R and channel asymmetry) is consistent with the analytics. However, the significance is currently overstated at the abstract level: the '50% in larger networks' figure is unsupported, the classical-NOMA comparison is degenerate at R≥1, and the power-allocation policy rests on an unvalidated single-sample estimator. With those gaps addressed, the two-user core would be a solid contribution.

major comments (5)

[Abstract; §IV.E; §VI] The abstract and §VI claim completion-time reduction of 'up to 50% in larger asymmetric networks,' but the paper contains no analysis and no simulation for any U>2. Every equation from (40)–(51) and every figure in §V is two-user, and §IV.E is a qualitative scheduling sketch only. The U>2 dynamics are not a trivial extension: Phase 2 pairs only two users, the weak user must still accumulate full rank before TDMA resumes, the benefit depends on the channel distribution of the strongest remaining overhearer, and the ACK-timing estimator (48) must be applied repeatedly. Either provide U>2 analysis/simulations supporting the 50% figure, or remove/scope the claim to what is shown.
[§IV.D, Eq. (48)–(49), Algorithm I] Eq. (48) inverts the expected-value formula E[T]=exp(τσ²/(Ω̂P))·F(N) on a single realized ACK time t. When user 2 is strong — precisely the regime where ONOMA helps most — t<F(N)≈N+1.6 occurs with substantial probability (e.g., P(T=50)≈0.17 for N=50, P_succ=0.99), making ln(t/F(N))<0 and the estimate Ω̂(2) negative (or undefined). Algorithm I's behavior in this event is unstated. More generally, a one-shot inversion of a mean-value identity is a high-variance estimator, and no sensitivity analysis of the λ policy to estimation error is given. Since λ directly controls both users' Phase-2 success probabilities (38)–(39), the robustness of this estimator is load-bearing for the claimed gains. At minimum, state the estimator's failure handling and show performance under estimator noise or a simple multi-sample variant.
[§IV.C, Eq. (44)] The JR-ONOMA completion time (44) is written as max{E[T_Ph-2^(1)], E[T_Ph-2^(2)]}, and the text states an upper bound follows by substituting (42)–(43). The true Phase-2 completion time is E[max(T^(1),T^(2))], which is lower-bounded (not upper-bounded) by max of the expectations, and substituting an upper bound on one branch into a max-of-means does not yield an upper bound on the quantity of interest. As written, (44) is not a valid characterization of JR-ONOMA latency. Please either derive E[max] (the two ranks evolve in the same slots under independent channels, so a union-bound or exact expression should be feasible) or clearly label (44) as a heuristic approximation and validate it against the Monte Carlo results.
[§III.E, §V, Figs. 6–7] The claim 'outperforms ... classical NOMA' is vacuous at R≥1: the paper sets λ=0.5 for SR-NOMA, which violates its own feasibility condition (6) (τ(R)≥1), yielding infinite expected transmissions — admitted in §V — so NOMA appears in Figs. 6a–f only as a degenerate baseline. Conversely, at R=0.5 (where (6) holds), SR-ONOMA is beaten by SR-MC and SR-NOMA (Fig. 6g–i). The headline comparison to five baselines therefore holds only in the high-rate asymmetric regime; the abstract and §V conclusions should be scoped accordingly, and ideally a feasible NOMA baseline (rate R<1 with λ satisfying (6), or SIC-feasible power split) should be included for a fair comparison.
[§IV.B, Eqs. (36)–(39)] Interference-free decoding at user 2 (Eqs. 36–37) assumes perfect regeneration of user 1's symbols: user 2 must have decoded user 1's entire Phase-1 block, know M, R_c, and the shared coding vector, and cancel with zero residual. No sensitivity to imperfect cancellation (residual interference fraction, header decoding failure at rate R′, or channel-estimation error in the analog subtraction) is analyzed or simulated. Since this is the mechanism that 'decouples' the latencies, even a brief robustness study (e.g., residual interference at −20 dB) would substantially strengthen the central claim.

minor comments (7)

[Fig. 6 caption] Fig. 6 caption states 'N=5 packets per data block' for R=2 bpcu, but §V text gives N=K̄/R=50. Please correct.
[§V] Ω is a channel-power (variance) parameter; quoting it in dBm (e.g., Ω(2)=−90 dBm) is dimensionally unusual for a Rayleigh variance. Clarify the normalization to noise power σ², since only the ratio τσ²/(ΩP) enters the formulas.
[§III.B–E] Terminology alternates between 'SR-MC'/'SC-MC' and 'SR-NOMA'/'SC-NOMA' (e.g., after (20), (26), (35)). Please unify.
[Throughout] Typos/formatting: 'the the strong user' (§I), 'The, using the tail-sum formula' (§II.B), 'UA V' (§I), 'up to34%' and '50%in' in the abstract, 'R= 2,bpcu' (§V), 'whose proof' punctuation around (19)–(20).
[§IV.B, footnote 1] Footnote 1 on security ('Security is preserved under standard layered encryption assumptions') appears unmotivated — no security threat model is discussed anywhere in the paper. Either motivate or remove.
[§IV.D, §V] The design constants f=1.5 and t_max are free parameters; a short sensitivity sweep of f (and statement of how t_max is chosen in the simulations) would help reproducibility.
[§III.C] In §III.C the FDMA description says rates are set 'as , i.e., R(1)=R(2)=R' — missing text before the comma.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: completion-time bounds follow from standard RLNC rank expectations and Rayleigh outage; simulations are Monte Carlo under the same model, not fits renamed as predictions.

full rationale

The paper's load-bearing chain is self-contained and non-circular. Packet success probabilities are the usual Rayleigh outage expressions (Eqs. 3, 5, 14, 30–31, 38–39). RLNC completion times are the standard sum of geometric waiting times until full rank, E[T]=F(N)/P_succ with F(N)≈N+1.6 (Eqs. 10–11). Phase-wise ONOMA and baseline bounds (Eqs. 15–47 and Appendices A–D) apply tower-property inequalities to those same quantities; they do not redefine the target latency in terms of itself. The Phase-2 power policy (Eqs. 48–51, Algorithm I) inverts the expected-completion formula on one observed ACK time and a designer t_max, then picks λ via a hand-set extension factor f=1.5—this is an explicit heuristic design choice, not a fitted parameter later reported as an independent prediction. Simulations (§V, Figs. 6–7) are Monte Carlo averages under the identical channel/coding model, not recalibrations that force the claimed gains. References are external (HARQ, RLNC, NOMA, IoT/NTN surveys); there is no self-citation uniqueness theorem or ansatz smuggled from overlapping authors. Overclaim issues (e.g., the unsupported “50% in larger networks” abstract figure, or classical NOMA at infeasible λ) are correctness/scope problems, not circular reductions of outputs to inputs. Score 0 is therefore appropriate.

Assumptions & free parameters 4 free parameters · 7 assumptions · 2 invented entities

The claim rests on standard wireless/RLNC modeling plus several protocol-specific idealizations (perfect symbol regeneration, ACK-time invertible to Ω, block i.i.d. Rayleigh, capacity-achieving packets). Free knobs are the extension factor f, rate choices, and the λ heuristic. ONOMA itself is the invented protocol entity; it has no independent empirical handle outside this paper’s sims.

free parameters (4)

time extension factor f = 1.5
Hand-chosen multiplier (set to 1.5 in sims) that sets how much longer than Phase-1 E[T] the Phase-2 user targets may run when solving for λ_u and λ_l (Eqs. 50–51, Algorithm I).
power allocation coefficient λ (and bounds λ_l, λ_u) = heuristic mean or random in [λ_l, λ_u]; classical baseline fixes 0.5
Not optimized from a closed-form objective under true CSI; chosen from ACK-inverted Ω̂ via Algorithm I, with random choice when bounds cross. Directly drives Phase-2 success probabilities (38)–(39).
t_max (worst-case completion time for weak-user Ω̂)
Designer-set cap inverted in Eq. 49 to estimate Ω̂^(1) when the transmitter has no CSI; controls how conservative λ_l is.
common rate R and equal rate-splitting defaults = R ∈ {0.5, 1, 2} bpcu in §V
With no CSI, baselines and ONOMA fix R^(1)=R^(2)=R and α=0.5; reported %-gains are tied to the simulated R∈{0.5,1,2} grid.

assumptions (7)

domain assumption Packet error equals outage under capacity-achieving coding and long packets: P_loss = Pr(|h|^2 P/σ^2 < 2^R−1).
Stated in §II Eqs. (2)–(3); all E[T] formulas inherit this.
domain assumption Channels are i.i.d. block-fading Rayleigh, h~(CN)(0,Ω), constant over a packet, independent across users.
§II and multi-user model in §III; enables closed-form exponential success probabilities.
standard math RLNC over GF(2) has expected rank overhead F(N)≈N+1.6 and success probability s_r = P_succ(1−2^{r−N}).
§II.C Eqs. (10)–(11); standard RLNC fact used for every scheme’s E[T].
ad hoc to paper Strong user can perfectly regenerate weak-user channel symbols (known M, R_c, shared coding vector) and cancel them with zero residual interference.
§IV.B Eqs. (36)–(37) and “symbol-aware” claim; security footnote assumes layered encryption but not imperfect CSI/modulation mismatch.
domain assumption Block ACKs are delivered reliably when a user’s decoding matrix hits full rank, at most once per N slots, and ACK timing is observable at the transmitter.
Low-feedback model §II.A and Phase-1 termination rule §IV.A; no ACK error model.
ad hoc to paper Inverting E[T_Ph-1]=exp(τσ^2/(Ω̂P))F(⌈K̄/R⌉) on one observed ACK time yields a usable channel-gain estimate for power control.
§IV.D Eqs. (48)–(49); equates a random hitting time to its expectation.
domain assumption Transmitter has neither instantaneous nor statistical CSI a priori; same rates/power rules applied until ACK timing arrives.
Problem statement §III; motivates all no-CSI baselines and ONOMA’s inference step.

invented entities (2)

ONOMA (Overhearing-driven NOMA) protocol
purpose: Cross-layer scheme that phases RLNC overhearing, ACK-timing ordering, reconstruction-based NOMA, and optional SR Phase-3 PtP to decouple user completion times.
Defined in §IV; performance claims are properties of this constructed protocol, not of a pre-existing standard object.
ACK-timing-based channel estimate Ω̂ and Algorithm I λ policy
purpose: Replace explicit CSI with inverted expected-completion formulas and a two-sided λ rule.
§IV.D; no external calibration dataset or measurement campaign validates the estimator.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Latency Decoupling in Low-Feedback Multi-User Networks via Overhearing-Driven NOMA." pith.science (2026). https://pith.science/paper/LX3FF4G6

@misc{pith2026260724609,
  author       = {Pith},
  title        = {Pith review of: Latency Decoupling in Low-Feedback Multi-User Networks via Overhearing-Driven NOMA},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LX3FF4G6}},
  note         = {Machine review of arXiv:2607.24609}
}

read the original abstract

Conventional wireless protocols such as Hybrid Automatic Repeat Request (HARQ) rely on frequent and timely feedback, which becomes impractical in low-feedback regimes including non-terrestrial networks and massive IoT. This limitation is particularly critical in heterogeneous multi-user systems with unknown and asymmetric channels, where a single weak user can dominate the overall completion latency. We propose Overhearing-driven Non-Orthogonal Multiple Access (ONOMA), a novel cross-layer transmission scheme that minimizes latency without requiring instantaneous or statistical CSI at the transmitter. ONOMA integrates Random Linear Network Coding (RLNC) with symbol-aware NOMA and explicitly exploits overhearing and acknowledgment timing. In the first phase, users overhear RLNC transmissions, and the relative timing of acknowledgments is used to implicitly infer channel strength ordering. In the second phase, symbol reconstruction enables interference-free decoding for strong users, effectively decoupling user latencies. An adaptive power allocation policy is derived from acknowledgment timing-based channel estimates. Analytical and simulation results show that ONOMA outperforms TDMA, multicast, FDMA, inter-session, and classical NOMA, reducing completion time by up to 34% in two-user and 50% in larger asymmetric networks.

Figures

Figures reproduced from arXiv: 2607.24609 by the authors.

**Figure 2.** A RLNC-based (network-encoded) packet structure; a) Packet struc [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Baseline transmission scheme with successive refinement with user 1 weaker than user 2: (a) Phase 1: baseline two-user transmission intended for [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

Figures from the paper (4 more)

**Figure 4.** Figure 4: ONOMA transmission scheme with user 1 weaker than user 2: (a) Phase 1: PtP packet transmission to user 1, while user 2 overhears the transmission [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 5.** Figure 5: Building blocks of the transmitter in Phase 2 of a two-user RLNC-based ONOMA transmission scheme [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: Comparison of the expected completion time of the SR-ONOMA scheme with the baseline transmission schemes in a downlink two-user low-feedback [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

**Figure 7.** Figure 7: Comparison of the expected completion time of the SR-ONOMA scheme with the most competitive baseline schemes, including TDMA and SR-MC, () () () [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references

[1]

Classi- fication of LTE uplink scheduling techniques: An M2M perspective,

M. A. Mehaseb, Y . Gadallah, A. Elhamy, and H. Elhennawy, “Classi- fication of LTE uplink scheduling techniques: An M2M perspective,” IEEE Commun. Surv. Tutorials, vol. 18, no. 2, pp. 1310–1335, 2016

2016
[2]

IMT vision–framework and overall objectives of the future development of IMT for 2020 and beyond,

M. Series, “IMT vision–framework and overall objectives of the future development of IMT for 2020 and beyond,”ITU-R Rec. M.2083, Sep. 2015

2020
[3]

IEEE 802.11be Wi- Fi 7: Feature summary and performance evaluation,

X. Liu, Y . Dong, Y . Li, Y . Lin, and M. Gan, “IEEE 802.11be Wi- Fi 7: Feature summary and performance evaluation,”IEEE Commun. Standards Mag., vol. 9, no. 2, pp. 1–7, Jun. 2025

2025
[4]

6G wireless systems: Vision, requirements, challenges, insights, and opportunities,

H. Tataria, M. Shafi, A. F. Molisch, M. Dohler, H. Sj ¨oland, and F. Tufvesson, “6G wireless systems: Vision, requirements, challenges, insights, and opportunities,”Proc. IEEE, vol. 109, no. 7, pp. 1166–1199, Jul. 2021

2021
[5]

6G wire- less communication systems: Applications, requirements, technologies, challenges, and research directions,

M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y . M. Jang, “6G wire- less communication systems: Applications, requirements, technologies, challenges, and research directions,”IEEE Open J. Commun. Soc., vol. 1, pp. 957–975, Nov. 2020

2020
[6]

What should 6G be?

S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, “What should 6G be?”Nat. Electron., vol. 3, no. 1, pp. 20–29, Jan. 2020

2020
[7]

What will 5G be?

J. G. Andrews, S. Buzzi, W. Choi, S. V . Hanly, A. Lozano, A. C. K. Soong, and J. C. Zhang, “What will 5G be?”IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1065–1082, Jun. 2014

2014
[8]

Distributed machine learning for early HARQ feedback prediction in cloud RANs,

B. G ¨oktepe, C. Hellge, T. Schierl, and S. Stanczak, “Distributed machine learning for early HARQ feedback prediction in cloud RANs,”IEEE Trans. Wireless Commun., vol. 23, no. 1, pp. 31–44, 2023

2023

Show all 21 references

[9]

Outage performance of hybrid satellite–terrestrial relaying networks with rateless codes in co-channel interference environments,

N. V . Toan, T. T. Duy, P. N. Son, D. T. Hung, N. Q. Sang, and L.-T. Tu, “Outage performance of hybrid satellite–terrestrial relaying networks with rateless codes in co-channel interference environments,” inProc. IEEE Int. Conf. Syst. Sci. Eng. (ICSSE), Ho Chi Minh City, Vietn...

2023
[10]

Grant-free non-orthogonal multiple access for IoT: A survey,

M. B. Shahab, R. Abbas, M. Shirvanimoghaddam, and S. J. Johnson, “Grant-free non-orthogonal multiple access for IoT: A survey,”IEEE Commun. Surv. Tutorials, vol. 22, no. 3, pp. 1805–1838, Third Quarter 2020

2020
[11]

Internet of things in the 5G era: Enablers, architecture, and business models,

M. R. Palattella, M. Dohler, A. Grieco, G. Rizzo, J. Torsner, T. Engel, and L. Ladid, “Internet of things in the 5G era: Enablers, architecture, and business models,”IEEE J. Sel. Areas Commun., vol. 34, no. 3, pp. 510–527, Mar. 2016

2016
[12]

Comprehensive evaluation of the IEEE 802.15.4 MAC layer performance with retransmissions,

A. Faridi, M. R. Palattella, A. Lozano, M. Dohler, G. Boggia, L. A. Grieco, and P. Camarda, “Comprehensive evaluation of the IEEE 802.15.4 MAC layer performance with retransmissions,”IEEE Trans. V eh. Technol., vol. 59, no. 8, pp. 3917–3932, Oct. 2010

2010
[13]

Iterative soft-Kalman channel esti- mation for fast time-varying MIMO-OFDM channels,

E. P. Simon and M. A. Khalighi, “Iterative soft-Kalman channel esti- mation for fast time-varying MIMO-OFDM channels,”IEEE Wireless Commun. Lett., vol. 2, no. 6, pp. 599–602, Dec. 2013

2013
[14]

Space-time block coded spatial modulation,

E. Basar, U. Ayg ¨ol¨u, E. Panayirci, and H. V . Poor, “Space-time block coded spatial modulation,”IEEE Trans. Commun., vol. 59, no. 3, pp. 823–832, Mar. 2011

2011
[15]

Toward improving the scalability of LR- FHSS: Transmission policies and performance analysis,

L. B. Rocha, J. M. de Souza Sant’Ana, M. A. Ullah, S. Montejo- S´anchez, and R. D. Souza, “Toward improving the scalability of LR- FHSS: Transmission policies and performance analysis,”IEEE Open J. Commun. Soc., vol. 6, no. 1, pp. 7075–7090, Jan. 2025

2025
[16]

Massive MIMO for maximal spectral efficiency: How many users and pilots should be allocated?

E. Bj ¨ornson, E. G. Larsson, and M. Debbah, “Massive MIMO for maximal spectral efficiency: How many users and pilots should be allocated?”IEEE Trans. Wireless Commun., vol. 15, no. 2, pp. 1293– 1308, Feb. 2016

2016
[17]

DRL-based energy minimization in fast HARQ with finite blocklength codes and feedback delay,

X. Wu, W. Wang, and D. Qiao, “DRL-based energy minimization in fast HARQ with finite blocklength codes and feedback delay,”IEEE Trans. Wireless Commun., vol. 24, no. 8, pp. 6917–6930, Aug. 2025

2025
[18]

A random linear network coding approach to multicast,

T. Ho, M. M ´edard, R. Koetter, D. R. Karger, M. Effros, J. Shi, and B. Leong, “A random linear network coding approach to multicast,” IEEE Trans. Inf. Theory, vol. 52, no. 10, pp. 4413–4430, Oct. 2006

2006
[19]

Completion delay of random linear network coding in full-duplex relay networks,

R. Su, Q. T. Sun, Z. Zhang, and Z. Li, “Completion delay of random linear network coding in full-duplex relay networks,”IEEE Trans. Commun., vol. 70, no. 12, pp. 7843–7857, Dec. 2022

2022
[20]

FlexNC+ RecNet: Flexible network (re)coding in cloud-native 5G: Design and testbed measurements,

O. Lhamo, T. V . Doan, E. Tas ¸demir, M. Attawna, G. T. Nguyen, P. Seeling, M. Reisslein, and F. H. P. Fitzek, “FlexNC+ RecNet: Flexible network (re)coding in cloud-native 5G: Design and testbed measurements,”IEEE Trans. Netw. Serv. Manag., vol. 22, no. 6, pp. 5756 – 5774, Dec. 2025

2025
[21]

Cost-efficient FEC scheme for time-sensitive multi-hop transmissions in overlay networks,

C. Xu, J. H. Wang, R. Li, H. Wu, J. Wang, J. Zhang, and K. Zheng, “Cost-efficient FEC scheme for time-sensitive multi-hop transmissions in overlay networks,”IEEE Trans. Mobile Comput., vol. 24, no. 9, pp. 8025–8038, Sep. 2025

2025

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.

[1] [1]

Classi- fication of LTE uplink scheduling techniques: An M2M perspective,

M. A. Mehaseb, Y . Gadallah, A. Elhamy, and H. Elhennawy, “Classi- fication of LTE uplink scheduling techniques: An M2M perspective,” IEEE Commun. Surv. Tutorials, vol. 18, no. 2, pp. 1310–1335, 2016

2016

[2] [2]

IMT vision–framework and overall objectives of the future development of IMT for 2020 and beyond,

M. Series, “IMT vision–framework and overall objectives of the future development of IMT for 2020 and beyond,”ITU-R Rec. M.2083, Sep. 2015

2020

[3] [3]

IEEE 802.11be Wi- Fi 7: Feature summary and performance evaluation,

X. Liu, Y . Dong, Y . Li, Y . Lin, and M. Gan, “IEEE 802.11be Wi- Fi 7: Feature summary and performance evaluation,”IEEE Commun. Standards Mag., vol. 9, no. 2, pp. 1–7, Jun. 2025

2025

[4] [4]

6G wireless systems: Vision, requirements, challenges, insights, and opportunities,

H. Tataria, M. Shafi, A. F. Molisch, M. Dohler, H. Sj ¨oland, and F. Tufvesson, “6G wireless systems: Vision, requirements, challenges, insights, and opportunities,”Proc. IEEE, vol. 109, no. 7, pp. 1166–1199, Jul. 2021

2021

[5] [5]

6G wire- less communication systems: Applications, requirements, technologies, challenges, and research directions,

M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y . M. Jang, “6G wire- less communication systems: Applications, requirements, technologies, challenges, and research directions,”IEEE Open J. Commun. Soc., vol. 1, pp. 957–975, Nov. 2020

2020

[6] [6]

What should 6G be?

S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, “What should 6G be?”Nat. Electron., vol. 3, no. 1, pp. 20–29, Jan. 2020

2020

[7] [7]

What will 5G be?

J. G. Andrews, S. Buzzi, W. Choi, S. V . Hanly, A. Lozano, A. C. K. Soong, and J. C. Zhang, “What will 5G be?”IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1065–1082, Jun. 2014

2014

[8] [8]

Distributed machine learning for early HARQ feedback prediction in cloud RANs,

B. G ¨oktepe, C. Hellge, T. Schierl, and S. Stanczak, “Distributed machine learning for early HARQ feedback prediction in cloud RANs,”IEEE Trans. Wireless Commun., vol. 23, no. 1, pp. 31–44, 2023

2023