Pith. sign in

REVIEW 2 major objections 6 minor 45 references

5G RAN Slicing with Load Balanced Handovers

T0 review · 2 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read RadioWeaver claims that a single cross-slice load metric, total normalized demand (TND), lets the network hand users over across slice boundaries while still letting every slice achieve its maximum feasible performance.

desk verdict A solid systems paper with a genuinely new load metric (TND) and fair evaluation, but the explicit interference-free assumption is load-bearing and never stress-tested, so the central claim holds only within that stated scope. read the letter →

arxiv 2505.24295 v1 pith:PGKCP54Z submitted 2025-05-30 cs.NI

Yongzhou Chen , Muhammad Taimoor Tariq , Haitham Hassanieh , Radhika Mittal This is my paper
classification cs.NI
keywords 5GRANslicingload-balancedhandoversnetworkmulti-cellresourcequotaallocationproportionalfairnessdatarateOpen
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

This paper tackles a practical problem in dense 5G networks: when users belong to different slices, each with its own service contract and optimization goal, how should the network decide which cell serves each user as people move? The authors propose RadioWeaver, a system that balances load across cells by using a single global criterion, total normalized demand (TND), which converts each slice's diverse objective into a comparable per-cell demand and sums them per cell. RadioWeaver moves 'amenable' users, who can get good-enough signal from another cell, from overloaded to underloaded cells regardless of their slice, then reallocates each slice's quota across cells to match the resulting demand. The paper argues that when this process reaches a fully complementary distribution, where each cell's TND equals its capacity and users keep near-best channel quality, every slice achieves its best possible performance under its own objective. If correct, this would let operators get more out of dense multi-cell deployments without violating slice SLAs.

What carries the argument

The carrying mechanism is the total normalized demand (TND) criterion together with the notion of a fully complementary distribution. TND defines a cell's load as $L_k = \sum_i D_{ik}$, the sum over all slices of each slice's normalized demand, obtained by scaling the slice's per-cell demand ratio by its global quota; this reduces the multi-criteria load balancing problem to a single scalar check of whether a cell is overloaded. The two-phase handover algorithm then drives the network toward the fully complementary state: first move 'amenable' users, whose per-resource data rate at the target cell is at least $\alpha = 0.8$ of the source cell, irrespective of slice; then, if necessary, move other users only when the handover improves that user's slice objective. A swap-based quota allocation assigns each slice its desired per-cell quota when the distribution is fully complementary, and otherwise greedily swaps quota between slices with complementary demand imbalances while preserving per-slice isolation.

What would settle it

Run RadioWeaver in a two-cell deployment where both cells share the same frequency band and measure user throughput per resource block as a function of neighbor-cell load. If a distribution that is fully complementary under TND (each cell's TND equals its capacity) fails to give every slice its predicted maximum performance because per-resource rates drop with neighbor load, the central optimality claim would be disproved in that regime.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1: a user distribution that is both fully complementary and has optimal quality achieves globally maximum performance $P^{\max}_i$ for every slice $S_i$. For each slice, RadioWeaver computes a demand ratio $d_{ik}$ at each cell from the slice's own optimization criterion (weighted proportional fairness, weighted datarate fairness, or fixed datarate demands), multiplies it by the slice's network-wide quota $Q_i$ to get its normalized demand $D_{ik}$, and sums those normalized demands across slices to get a cell's total normalized demand $L_k$. A distribution is fully complementary when every cell's TND exactly equals its radio capacity $R_k$; then each slice can be allocated exactly the quota it wants at each cell, which the authors prove is necessary and sufficient for the slice to reach its maximum possible performance given that distribution. RadioWeaver then tries to reach such a distribution by handing users over from overloaded to underloaded cells, using the macro cell as a relay when small cells do not overlap, preferring users whose channel quality to the target cell is within a threshold of their current cell, and only moving other users when the move improves their slice's own objective.

Load-bearing premise

The result depends on the assumption that every cell operates on its own frequency band that does not interfere with its neighbors, so a cell's throughput depends on its own load and not on what happens in nearby cells; if cells share spectrum, the real cost of moving a user is not captured by TND and the handover decisions could be wrong.

Editorial extensions

If this is right

  • Operators can apply one load-balancing rule across a multi-slice network instead of balancing each slice in isolation, and the resulting user-cell mapping lets each slice meet its SLA.
  • In skewed deployments, where some slices have no users with good signal from an underloaded cell, cross-slice movement of amenable users can still balance load; the paper reports 16-365% better slice-level metrics than baselines.
  • RadioWeaver triggers fewer physical handovers than greedy association schemes because it decides the final mapping before issuing handovers, avoiding cascading reassignments.
  • When full complementarity is impossible, the swap-based quota allocation guarantees that no slice is worse off than under a static per-cell quota split.

Reading between the lines

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

  • A testable corollary is that the benefit of TND-based balancing grows with the share of users whose channel quality is good from two or more cells; deployments with strong 'amenable' overlap should show the largest gains, which the paper's own field measurements suggest.
  • The same TND condition could serve as a feasibility check before accepting a slice SLA: a demand distribution is sustainable only if some user-cell mapping keeps TND at or below capacity in every cell, connecting this work to admission control.
  • Because TND assumes each cell's capacity is fixed and independent, the framework would need an effective-capacity correction in spectrum-sharing deployments where moving users changes interference and hence per-resource rates; that is an open extension consistent with the paper's scope statement.
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.

Referee Report

2 major / 6 minor

Summary. The paper presents RadioWeaver, a system for load-balanced handovers in multi-cell 5G RAN slicing with heterogeneous slice objectives. It introduces a global load metric, TND, defined as the sum over slices of per-slice normalized demands, where each slice's demand ratio is computed from its optimization criterion (WPF or WDRF) and the user-to-cell mapping. The paper proves three results: Lemma 1 (per-slice performance is maximized when allocated quota equals normalized demand), Lemma 2 (all slices simultaneously reach their maximum iff the distribution is fully complementary, i.e., TND equals capacity at each cell), and Theorem 1 (a fully complementary distribution with optimal channel quality achieves the globally maximum performance). It then presents a two-step iterative handover algorithm that moves amenable users across cells using the macrocell as a relay, and a swap-based quota allocation scheme. The system is evaluated with trace-driven simulations and a Colosseum testbed, reporting large improvements over several baselines.

Significance. The paper makes a genuinely novel contribution by defining a cross-slice load metric that unifies heterogeneous slice objectives, and it backs the metric with clean mathematical proofs under explicit assumptions. The evaluation is thorough and shows consistent gains over strong baselines, including a variant of the closest related work (MORA++). The main caveat is that the theoretical results and the algorithm's correctness are conditional on the non-interfering-band assumption, which is stated in §2 but not stress-tested; this limits the generality of the central claim. The paper would be a valuable addition to the RAN slicing literature once this scope issue is addressed.

major comments (2)
  1. [Section 2; §5.1; Appendix B] The assumption that each cell operates on a non-interfering frequency band (stated in Section 2) is load-bearing for the central results. In §5.1 the demand ratios d_ik are computed from fixed per-RB rates e_ij, and the proofs of Lemma 1, Lemma 2, and Theorem 1 in Appendix B treat e_ij as constants under the user-to-cell mapping. If cells share spectrum, e_ij depends on the interference generated by neighboring cells' transmissions, which in turn depends on the entire user-to-cell mapping; moving a user then changes the effective rates of all users in both the source and target cells, so the TND that justified a handover can differ from the TND after the handover, potentially inverting overload/underload labels. The evaluation in §7.2 places small cells on non-interfering bands, and the Colosseum testbed in §7.5 uses equal channel quality across cells, so this failure mode is never exercised. Because this assumption is explicit but its consequences are not analyzed, the authors should either (a) provide an analysis of the conditions under which TND remains a valid load metric under interference, (b) include a simulation with a simple interference model to quantify the robustness of the algorithm, or (c) clearly delimit the paper's claims to non-interfering deployments in the abstract and conclusion. As it stands, the theorem and the load-balancing decisions are conditioned on a scope that is stated but not examined.
  2. [§5.3; Theorem 1 (§5.2)] The iterative two-step handover algorithm in §5.3 is described as moving users until "the load is balanced ... or further handovers are no longer possible," but no convergence or approximation guarantee is provided relative to the fully complementary optimum defined in Theorem 1. The algorithm is therefore a heuristic that may stop at a distribution that is not fully complementary (as the tail of Fig. 6(a) indeed shows). The paper should clarify the intended relationship between the theorem and the heuristic: is the algorithm designed to converge to a fully complementary distribution whenever one is feasible, or is it a best-effort approach with no such guarantee? A brief formal statement of the stopping condition's suboptimality, or an explicit acknowledgment that Theorem 1 is an upper-bound benchmark rather than an objective the algorithm is guaranteed to reach, would make the contribution more precise.
minor comments (6)
  1. [§5.1(i)] The sentence "We further validated this both mathematically and empirically (details excluded for brevity)" is unsupported; the mathematical validation is given in Appendix B, but no empirical details are provided, so the sentence should be removed or the appendix referenced.
  2. [§5.1(iii)] The reduction of fixed datarate demands to the WDRF formulation by setting weights proportional to demands is stated without justification; a one-sentence derivation would help.
  3. [§5.2] The notation P_max_i is used for both the slice's maximum performance under a given user distribution (Lemma 1) and the global maximum over all user distributions (Theorem 1), which is confusing; distinct symbols would improve readability.
  4. [§7.5] The Colosseum evaluation's restriction to equal channel quality across cells should be explicitly listed as a limitation in the main text, since it means the testbed results do not exercise the channel-quality-aware handover logic.
  5. [Figure 6(d)] The y-axis is not labeled in the text description; it should state "handovers per user per 20-second run" to match the description.
  6. [General] The paper uses "handover" and "handoff" interchangeably; please use a single term throughout.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the TND load criterion is derived from slice objectives via self-contained optimization proofs, and self-citations are confined to implementation or parameterization details.

full rationale

The paper's central derivation chain is self-contained. Demand ratios d_ik are derived per slice from the slice's stated objective (WPF or WDRF) and per-user channel quality e_ij (Section 5.1), with Lemma 1 proved in Appendix B.1 via Lagrange multipliers and a minimax argument, not assumed. Lemma 2 follows from the capacity-feasibility constraint (sum_i Q_ik = R_k) together with Lemma 1, and Theorem 1 adds the observation that the per-slice upper bounds in Lemma 1 increase strictly with every user's e_ij, so a fully complementary distribution with optimal quality attains the global maximum over user distributions. The 'fully complementary' condition is defined in terms of cell capacities and normalized demands (L_k = sum_i D_ik = R_k), not directly in terms of P_max, so the theorem is not a mere restatement of its own definition. The paper does cite prior work [17,44] for the WPF allocation form and [19] for slice parameterization and the simulator, but the relevant proofs are present in the paper, and these citations are used for standard results or implementation scaffolding rather than as the load-bearing justification for the TND optimality claim. The Section 2 assumption of non-interfering frequency bands is an explicit modeling scope restriction: it makes e_ij a fixed function of the user-cell mapping, which is a limitation of applicability, not a circular step. The line 'We further validated this both mathematically and empirically (details excluded for brevity)' is a brevity-omitted detail, but the mathematical validation is actually supplied in Appendix B.1, so it does not hide a circular dependency. No fitted parameter is renamed as a prediction, and no uniqueness claim is imported from the authors' prior work. The evaluation uses external simulators and a hardware testbed against independently defined baselines, with the caveat that the baseline MORA++ shares RadioWeaver's demand computation; that comparison still isolates the effect of the TND-driven load balancing. Overall, the derivation stands on its own equations and proofs.

Assumptions & free parameters 1 free parameters · 4 assumptions · 0 invented entities

TND is a metric, not a new physical entity. The central claim rests on interference-free operation, fixed slice quotas, and a linear model of user throughput, all explicitly stated but not externally validated beyond the paper's experiments. The only tunable knob is the user-amenability threshold alpha.

free parameters (1)
  • alpha (amenable-user threshold) = 0.8
    Users are considered amenable to handover from a source to target cell only if the ratio of their per-RB data rates is above alpha (Section 5.2). Set to 0.8 in the prototype; sensitivity to 0.9 is examined in Section 7.4. The choice influences the balance between channel-quality preservation and load balancing, so it is a hand-tuned knob that affects reported gains.
assumptions (4)
  • domain assumption Cells operate on non-interfering frequency bands
    Stated in Section 2. The TND load model treats cell resources as independent; cross-cell interference would break the linear throughput model.
  • domain assumption User throughput at a cell equals allocated RBs times per-RB data rate, and each slice's intra-cell scheduler implements the assumed objective (WPF or WDRF)
    Used throughout Section 5.1 and Appendix B to derive demand ratios and prove Lemma 1. If a slice uses a different scheduler, these demand-ratio formulas may not represent its true resource needs.
  • domain assumption Slice global quota Q_i and cell capacities R_k are fixed inputs
    Assumed in Section 5.1 and used in Lemma 2. SLA quotas and capacities are treated as given parameters, not optimized.
  • domain assumption User channel quality (CQI) is available and stable at the coarse timescale of load balancing
    The RIC workflow in Section 5 relies on periodic CQI reports; if CQI changes rapidly within the 500ms and 3dB trigger interval, computed demand ratios may be stale.

how reviews work

0 comments
Cite this review

Pith. "Pith review of 5G RAN Slicing with Load Balanced Handovers." pith.science (2026). https://pith.science/paper/PGKCP54Z

@misc{pith2026250524295,
  author       = {Pith},
  title        = {Pith review of: 5G RAN Slicing with Load Balanced Handovers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PGKCP54Z}},
  note         = {Machine review of arXiv:2505.24295}
}
read the original abstract

With increasing density of small cells in modern multi-cell deployments, a given user can have multiple options for its serving cell. The serving cell for each user must be carefully chosen such that the user achieves reasonably high channel quality from it, and the load on each cell is well balanced. It is relatively straightforward to reason about this without slicing, where all users can share a global load balancing criteria set by the network operator. In this paper, we identify the unique challenges that arise when balancing load in a multi-cell setting with 5G slicing, where users are grouped into slices, and each slice has its own optimization criteria, resource quota, and demand distributions, making it hard to even define which cells are overloaded vs underloaded. We address these challenges through our system, RadioWeaver, that co-designs load balancing with dynamic quota allocation for each slice and each cell. RadioWeaver defines a novel global load balancing criteria across slices, that allows it to easily determine which cells are overloaded despite the fact that different slices optimize for different criteria. Our evaluation, using large-scale trace-driven simulations and a small-scale OpenRAN testbed, show how RadioWeaver achieves 16-365% better performance when compared to several baselines.

Figures

Figures reproduced from arXiv: 2505.24295 by the authors.

Figure 1
Figure 1. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The diagrams represent the coverage areas and user distribution across two cells (𝐶1 represented in blue and 𝐶2 represented in green). Users are split among two slices (𝑆𝐴 depicted by triangles and 𝑆𝐵 depicted by circles). The color of each user represents the corresponding serving cell under different handover schemes. Red squares highlight the extra handovers caused by load balancing for different schemes. RAN Quo… view at source ↗
Figure 3
Figure 3. Per-slice RAN quota at each cell and the corresponding per-user metric under different load balanced handover schemes. load at 𝐶2 due to the poorer channel quality, while 𝑆𝐵 has a higher load at 𝐶1 with more users. We can exploit this comple￾mentary demand distributions to adjust their relative quotas at the two cells [33]. Specifically, given the 12:8 demand ratio across the two cells for slice 𝑆𝐵, we can allocate … view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Overall Workflow of RadioWeaver cell of each user is fixed and given, and does not consider the complementary problem of load balancing. It frames per￾cell quota allocation across slices as a global optimization problem, which it solves using an off-the-shelf solver. A…
Figure 5
Figure 5. Figure 5: The two-steps handovers of RadioWeaver over multiple rounds. (a)(b) show Step1 and Step2 in Round1, and (c) shows Step1 in Round2. more RBs (higher 𝑛𝑖𝑗). Therefore, the demand ratio of such a slice 𝑆𝑖 at cell 𝐶𝑘 can be denoted as 𝑑𝑖𝑘 = Í 𝑢𝑖 𝑗 ∈𝐶𝑘 𝑤𝑖 𝑗 /𝑒𝑖 𝑗 Í 𝑢𝑖 𝑗 𝑤𝑖 𝑗…
Figure 6
Figure 6. Figure 6: Comparing RadioWeaver and baselines: (a) The CDF graph of cells’ load ratios. The load balancing effect on slice-level performance when compared to NoLB in (b) DRF slices and (C) WPF slices. (d) The number of physical handovers triggered per user across the experiment.…
Figure 7
Figure 7. Figure 7: Comparing RadioWeaver and baselines: (a) The reduction in median FCT of web flows. (b) The weighted PF metric improvement of WPF slices with different slice quotas in Scenario3. MORA++ are the highest, owing to their greedy user-cell assignment that creates a cascade o…
Figure 8
Figure 8. Figure 8: The experiment settings and performance improve￾ment of RadioWeaver compared with other baselines in mul￾tiple scenarios. skewed geo-distribution (where some slices have no users in the coverage regions of some small cells). Scenario 4 has balanced geo-distribution of …
Figure 9
Figure 9. Figure 9: The average throughput of every UE in two slices. The legend shows the max-min fairness index(maximal user throughput divided by minimal user throughput) of NoLB, NaiveLB, and RadioWeaver. (as shown in Fig9). Suppose NaiveLB randomly transfers two UEs in 𝑆0 from 𝐶1 to …
Figure 10
Figure 10. Figure 10: The normalized SINR values measured at 15 different locations for cellular operator A 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Location ID 0.00 0.25 0.50 0.75 1.00 Normalized SINR Value 2120Mhz 2175Mhz 751Mhz 1962.5Mhz [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 11
Figure 11. Figure 11: The normalized SINR values measured at 15 different locations for cellular operator B Since U is fully complementary, it indicates that the TND at each cell equals to the cell’s capacity, i.e., ∀𝑘, Í 𝑖 𝐷𝑖𝑘 = 𝑅𝑘 . The allocation strategy is to assign 𝐷𝑖𝑘 to 𝑆𝑖 in 𝐶𝑘 , …

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

45 extracted references · 44 canonical work pages

  1. [1]

    https: //www.ericsson.com/en/reports-and-papers/microwave- outlook/articles/maximizing-capacity-in-spectrum-limited- networks

    Frequency reuse in limited spectrum networks. https: //www.ericsson.com/en/reports-and-papers/microwave- outlook/articles/maximizing-capacity-in-spectrum-limited- networks

    work page

  2. [2]

    https://www.o-ran.org/

    O-RAN Alliance. https://www.o-ran.org/

    work page

  3. [3]

    Technical Report TS 38.133 V16.4.0, 3rd Generation Partnership Project (3GPP), 2020

    5G; NR; Requirements for support of radio resource management (3GPP TS 38.133 version 16.4.0 Release 16). Technical Report TS 38.133 V16.4.0, 3rd Generation Partnership Project (3GPP), 2020. Release 16

    work page 2020

  4. [4]

    Technical Re- port 36.814, 3rd Generation Partnership Project (3GPP), 2020

    Further advancements for e-utra physical layer aspects. Technical Re- port 36.814, 3rd Generation Partnership Project (3GPP), 2020. Version 9.0.0

    work page 2020

  5. [5]

    https://www.abiresearch.com/press/5g-data- traffic-explosion-will-drive-5g-small-cell-deployments-to-13-million- by-2027, 2022

    5G Data Traffic Explosion Will Drive 5G Small Cell Deployments to 13 million by 2027. https://www.abiresearch.com/press/5g-data- traffic-explosion-will-drive-5g-small-cell-deployments-to-13-million- by-2027, 2022

    work page 2027

  6. [6]

    https://www.lightreading.com/digital-transformation/us-cell-towers- and-small-cells-by-the-numbers, 2022

    US cell towers and small cells: By the numbers. https://www.lightreading.com/digital-transformation/us-cell-towers- and-small-cells-by-the-numbers, 2022

    work page 2022

  7. [7]

    Technical report, NOKIA, 2023

    5g spectrum bands explained — low, mid and high band. Technical report, NOKIA, 2023

    work page 2023

  8. [8]

    Measurement report event a3 in 5g nr

    5G-RRC-LAYER. Measurement report event a3 in 5g nr. https://wirelessbrew.com/5g-nr/measurement-report-event-a3- in-5g-nr/ , 2025

    work page 2025

Show all 45 references
  1. [9]

    Agrawal, A

    R. Agrawal, A. Bedekar, S. Kalyanasundaram, N. Arulselvan, T. Kold- ing, and H. Kroener. Centralized and decentralized coordinated schedul- ing with muting. In 2014 IEEE 79th V ehicular Technology Conference (VTC Spring), pages 1–5, 2014

    2014

  2. [10]

    Al-Falahy and O

    N. Al-Falahy and O. Y . Alani. Technologies for 5g networks: Chal- lenges and opportunities. IT Professional, 19(1):12–20, 2017

    2017

  3. [11]

    Balasingam, M

    A. Balasingam, M. Kotaru, and P. Bahl. Application-Level service assurance with 5g RAN slicing. In 21st USENIX Symposium on Net- worked Systems Design and Implementation (NSDI 24), pages 841–857, Santa Clara, CA, Apr. 2024. USENIX Association

    2024

  4. [12]

    D. Bega, M. Gramaglia, A. Banchs, V . Sciancalepore, and X. Costa- Pérez. A machine learning approach to 5g infrastructure market opti- mization. IEEE Transactions on Mobile Computing , 19(3):498–512, 2020

    2020

  5. [13]

    Bhushan, J

    N. Bhushan, J. Li, D. Malladi, R. Gilmore, D. Brenner, A. Damnjanovic, R. T. Sukhavasi, C. Patel, and S. Geirhofer. Network densification: the dominant theme for wireless evolution into 5g. IEEE Communications Magazine, 52(2):82–89, 2014

    2014

  6. [14]

    Blogspot

    T.-K. Blogspot. Handover optimization. https://telecom- knowledge.blogspot.com/2015/01/handover-optimization.html, 2015

    2015

  7. [15]

    Bonati, S

    L. Bonati, S. D’Oro, S. Basagni, and T. Melodia. Scope: an open and softwarized prototyping platform for nextg systems. In Proceedings of the 19th Annual International Conference on Mobile Systems, Applica- tions, and Services, MobiSys ’21, page 415–426, New York, NY , USA,

  8. [16]

    Bonati, P

    L. Bonati, P. Johari, M. Polese, S. D’Oro, S. Mohanti, M. Tehrani- Moayyed, D. Villa, S. Shrivastava, C. Tassie, K. Yoder, A. Bagga, P. Patel, V . Petkov, M. Seltser, F. Restuccia, A. Gosain, K. R. Chowd- hury, S. Basagni, and T. Melodia. Colosseum: Large-Scale Wireless Experi...

    2021

  9. [17]

    Caballero, A

    P. Caballero, A. Banchs, G. de Veciana, and X. Costa-Pérez. Multi- tenant radio access network slicing: Statistical multiplexing of spatial loads. IEEE/ACM Transactions on Networking , 25(5):3044–3058, 2017

    2017

  10. [18]

    Y . Chen, B. Ai, Y . Niu, R. He, Z. Zhong, and Z. Han. Resource allocation for device-to-device communications in multi-cell multi- band heterogeneous cellular networks. IEEE Transactions on V ehicular Technology, 68(5):4760–4773, 2019

    2019

  11. [19]

    Y . Chen, R. Yao, H. Hassanieh, and R. Mittal. Channel-Aware 5g RAN slicing with customizable schedulers. In 20th USENIX Symposium on Networked Systems Design and Implementation (NSDI 23) , pages 1767–1782, Boston, MA, Apr. 2023. USENIX Association

    2023

  12. [20]

    5G; 5G System; Network Slice Selection Ser- vices (3GPP TS 29.531 version 15.1.0 Release 15)

    ETSI. 5G; 5G System; Network Slice Selection Ser- vices (3GPP TS 29.531 version 15.1.0 Release 15). https: //www.etsi.org/deliver/etsi_ts/129500_129599/129531/ 15.01.00_60/ts_129531v150100p.pdf, 2018

    2018

  13. [21]

    5g; nr; base station (bs) radio transmission and re- ception

    ETSI. 5g; nr; base station (bs) radio transmission and re- ception. https://www.etsi.org/deliver/etsi_ts/138100_138199/ 138104/15.02.00_60/ts_138104v150200p.pdf, 2018

    2018

  14. [22]

    Foukas, M

    X. Foukas, M. K. Marina, and K. Kontovasilis. Orion: Ran slicing for a flexible and cost-effective multi-service mobile network architecture. In Proceedings of the 23rd Annual International Conference on Mobile Computing and Networking, MobiCom ’17, page 127–140, New York, NY ...

    2017

  15. [23]

    Foukas, G

    X. Foukas, G. Patounas, A. Elmokashfi, and M. K. Marina. Network slicing in 5g: Survey and challenges. IEEE Communications Magazine, 2017

    2017

  16. [24]

    Gomez-Miguelez, A

    I. Gomez-Miguelez, A. Garcia-Saavedra, P. D. Sutton, P. Serrano, C. Cano, and D. J. Leith. Srslte: An open-source platform for lte evolution and experimentation. In Proceedings of the Tenth ACM International Workshop on Wireless Network Testbeds, Experimental Evaluation, and C...

    2016

  17. [25]

    Gudipati, D

    A. Gudipati, D. Perry, L. E. Li, and S. Katti. Softran: software defined radio access network. HotSDN ’13, page 25–30, New York, NY , USA,

  18. [26]

    Gulati, S

    S. Gulati, S. Kalyanasundaram, P. Nashine, B. Natarajan, R. Agrawal, and A. Bedekar. Performance analysis of distributed multi-cell coordi- nated scheduler. In 2015 IEEE 82nd V ehicular Technology Conference (VTC2015-Fall), pages 1–5, 2015

    2015

  19. [27]

    Gures, I

    E. Gures, I. Shayea, M. Ergen, M. H. Azmi, and A. A. El-Saleh. Ma- chine learning-based load balancing algorithms in future heterogeneous networks: A survey. IEEE Access, 10:37689–37717, 2022

    2022

  20. [28]

    M. M. Hasan, S. Kwon, and J.-H. Na. Adaptive mobility load balancing algorithm for lte small-cell networks. IEEE Transactions on Wireless Communications, 17(4):2205–2217, 2018

    2018

  21. [29]

    M. M. Hasan, S. Kwon, and S. Oh. Frequent-handover mitigation in ultra-dense heterogeneous networks. IEEE Transactions on V ehicular Technology, 68(1):1035–1040, 2019

    2019

  22. [30]

    Kokku, R

    R. Kokku, R. Mahindra, H. Zhang, and S. Rangarajan. Nvs: a vir- tualization substrate for wimax networks. In Proceedings of the Six- teenth Annual International Conference on Mobile Computing and Networking, MobiCom ’10, page 233–244, New York, NY , USA, 2010. Association for ...

    2010

  23. [31]

    Y . Li, B. Cao, and C. Wang. Handover schemes in heterogeneous lte networks: challenges and opportunities. IEEE Wireless Communica- tions, 23(2):112–117, 2016

    2016

  24. [32]

    Y . Li, Z. Yuan, and C. Peng. A control-plane perspective on reducing data access latency in lte networks. In Proceedings of the 23rd Annual International Conference on Mobile Computing and Networking , Mo- biCom ’17, page 56–69, New York, NY , USA, 2017. Association for Compu...

    2017

  25. [33]

    Mahindra, M

    R. Mahindra, M. A. Khojastepour, H. Zhang, and S. Rangarajan. Ra- dio access network sharing in cellular networks. In 2013 21st IEEE International Conference on Network Protocols (ICNP) , pages 1–10, 2013. 13 Y ongzhou Chen1, Muhammad T aimoor T ariq1, Haitham Hassanieh 2, Rad...

    2013

  26. [34]

    Misra, A

    R. Misra, A. Gudipati, and S. Katti. Quickc: practical sub-millisecond transport for small cells. In Proceedings of the 22nd Annual Interna- tional Conference on Mobile Computing and Networking , MobiCom ’16, page 109–121, New York, NY , USA, 2016. Association for Com- puting ...

    2016

  27. [35]

    W. Nam, D. Bai, J. Lee, and I. Kang. Advanced interference man- agement for 5g cellular networks. IEEE Communications Magazine, 52(5):52–60, 2014

    2014

  28. [36]

    G. Piro, L. A. Grieco, G. Boggia, F. Capozzi, and P. Camarda. Simulat- ing lte cellular systems: An open-source framework.IEEE Transactions on V ehicular Technology, 60(2):498–513, 2011

    2011

  29. [37]

    Pontus, T

    W. Pontus, T. Claes, O. Antonino, and Q. Liwei. L1/l2 trig- gered mobility – the new way of doing handover in 5g ad- vanced. https://www.ericsson.com/en/blog/2024/8/5g-advanced- handover-triggered-mobility, 2024

    2024

  30. [38]

    Prado, F

    A. Prado, F. Stöckeler, F. Mehmeti, P. Krämer, and W. Kellerer. En- abling proportionally-fair mobility management with reinforcement learning in 5g networks. IEEE Journal on Selected Areas in Communi- cations, 41(6):1845–1858, 2023

    2023

  31. [39]

    A. Rao. 5g network slicing: crossdomain orchestration and manage- ment will drive commercialization. https://www.cisco.com/c/dam/ en/us/products/collateral/cloud-systems-management/network- services-orchestrator/white-paper-sp-5g-network-slicing .pdf, 2020

    2020

  32. [40]

    J. X. Salvat, L. Zanzi, A. Garcia-Saavedra, V . Sciancalepore, and X. Costa-Perez. Overbooking network slices through yield-driven end-to-end orchestration. In Proceedings of the 14th International Conference on Emerging Networking EXperiments and Technologies. Association for...

    2018

  33. [41]

    Xie and K

    Y . Xie and K. Jamieson. Ng-scope: Fine-grained telemetry for nextg cellular networks. Proc. ACM Meas. Anal. Comput. Syst. , 6(1), feb 2022

    2022

  34. [42]

    Y . Xu, G. Gui, H. Gacanin, and F. Adachi. A survey on resource allocation for 5g heterogeneous networks: Current research, future trends, and challenges. IEEE Communications Surveys & Tutorials , 23(2):668–695, 2021

    2021

  35. [43]

    C. Yang, W. Wang, Y . Qian, and X. Zhang. A weighted proportional fair scheduling to maximize best-effort service utility in multicell network. In 2008 IEEE 19th International Symposium on Personal, Indoor and Mobile Radio Communications, pages 1–5, 2008

    2008

  36. [44]

    U is fully complemen- tary

    Q. Ye, B. Rong, Y . Chen, M. Al-Shalash, C. Caramanis, and J. G. Andrews. User association for load balancing in heterogeneous cel- lular networks. IEEE Transactions on Wireless Communications , 12(6):2706–2716, 2013. 14 5G RAN Slicing with Load Balanced Handovers A Measuremen...

    2013

  37. [2013]

    Association for Computing Machinery

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.