Pith. sign in

REVIEW 4 major objections 6 minor 59 references

A "Breathing" Mobile Communication Network

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper claims that adjusting antennas' pilot-channel transmit power in real time can make their loads converge to a shared target, cutting load imbalance by more than half in city-scale simulations.

desk verdict The consensus framing is genuinely new and the simulations are promising, but the main convergence proof has a concrete contraction-coefficient error that, as written, invalidates the theoretical consensus claim. read the letter →

arxiv 2411.17290 v1 pith:6VQUOIYL submitted 2024-11-26 eess.SY cs.SYmath.OC

Chao Ge , Ge Chen , Zhipeng Jiang This is my paper · ORCID
classification eess.SYcs.SYmath.OC
keywords loadbalancingCPICHtransmitpoweraverageconsensusbusy-degreemeasurementreportdatanetworkcoverageLaplacianmatrixmulti-agentsystems
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 proposes treating a mobile network's antennas as cooperating agents that adjust their pilot-channel (CPICH) transmit power in real time so that each antenna's load, measured as a "busy-degree," converges toward a common target. The authors model the problem as an average-consensus loop: the Jacobian matrix relating busy-degrees to CPICH powers is shown to be a Laplacian matrix, which lets them solve for power updates via a pseudoinverse (BDBA) or a faster diagonal variant (BFDBA), with a machine-learning coverage check to keep coverage at 99.9%. If the theory holds, the network "breathes"—coverage areas expand and contract with traffic—so under-used antennas absorb load from over-busy ones. Simulations on three large Beijing datasets report that the mean standard deviation of busy-degrees falls by 50–56% and the share of over-busy antennas falls by 56–68%.

What carries the argument

The central object is the Jacobian matrix $A(k)$ with entries $(1/\bar f_i(k))\, \partial f_i/\partial p_j(k)$, which the paper proves is a Laplacian matrix under strong connectivity and continuous traffic density. BDBA approximates this matrix from measurement-report data by counting, for small power perturbations $\pm\epsilon$, how many reports would switch serving antenna, then solves $A(k)\vec u(k)=\vec d(k)$ by pseudoinverse; BFDBA uses only the diagonal entries. A monotone multilayer perceptron trained on MR data supplies the minimum CPICH power needed to keep the network coverage rate above $F^{\mathrm{con}}$, enforcing the coverage constraint in the update step.

What would settle it

Run BDBA on a live network for one week while separately logging PRB utilization and MR counts per antenna; the central claim fails if the correlation between $|M_i(k)|/M$ and actual PRB utilization is weak (e.g., below 0.8) or if the PRB-based standard deviation of busy-degrees does not drop by roughly half with coverage held at 99.9%.

Watch

Extended reading notes

Core claim

The central claim is that dynamic load balancing can be solved without traffic prediction by letting antennas share load through CPICH power updates. Formally, with a continuous average traffic density and a strongly connected antenna graph, the normalized Jacobian matrix $A(k)$ of busy-degrees with respect to CPICH powers is a Laplacian matrix of rank $n-1$, so the power update satisfying $A(k)\vec u(k)=\vec d(k)$ is solvable via pseudoinverse; under slowly varying relative traffic, the resulting closed loop drives all busy-degrees to $z(k)/\sum_j r_j$ for BDBA and to $(1-\tau p_i(k))z(k)/\sum_j r_j$ for BFDBA. The MR-data-based derivative estimates and MLP-based coverage check make this implementable, and the simulations demonstrate the claimed reductions while keeping coverage at 99.9%.

Load-bearing premise

The whole result rests on the assumption that the number of measurement reports an antenna serves is proportional to its real PRB utilization, and that users' reactions to power changes are captured by historical measurement reports.

Editorial extensions

If this is right

  • Under the paper's assumptions, BDBA provably drives all antennas' busy-degrees to the global target $z(k)/\sum_j r_j$, so the network can be steered toward one fairness target without per-cell traffic prediction.
  • BFDBA reaches approximate consensus using only diagonal Jacobian estimates, cutting running time by roughly half on the three datasets while sacrificing only a few percentage points of balancing quality.
  • The reported simulations show the mean standard deviation of busy-degrees reduced by 50–56% and the proportion of over-busy antennas reduced by 56–68% while coverage remains at 99.9%.
  • Because the loop reacts to measured MR data rather than forecast traffic, its performance does not degrade with prediction error, unlike prediction-based load-balancing methods.
  • The MLP-based minimum-power safeguard converts the balance-only problem into a constrained optimization that still meets a hard coverage requirement.

Reading between the lines

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

  • If measurement-report counts do not track true PRB utilization under real traffic, the reported 50–68% improvements may not transfer; a field trial logging actual PRB counters alongside MR counts would be the direct test.
  • The same Laplacian-consensus machinery could extend to other soft parameters such as cell individual offset or joint power/CIO control, since only the Jacobian's structure is used.
  • Combining this breathing framework with base-station sleeping would let under-loaded antennas not only shrink but switch off, compounding the energy savings the paper lists as future work.
  • The convergence bound in Proposition 5.1 suggests balancing quality degrades gracefully with the relative-traffic variation $\delta$, so the algorithm should tolerate the slow daily tide rather than requiring exactly steady traffic.
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

4 major / 6 minor

Summary. The paper proposes a "breathing" mobile communication network in which antennas cooperatively adjust their CPICH transmit powers in real time so that their busy-degrees converge to a common target. The optimization is formulated as a linearized equation involving a Jacobian matrix A(k), which the paper proves to be a Laplacian matrix under continuous-traffic assumptions. Two algorithms, BDBA and BFDBA, are proposed, together with an MR-data/MLP-based fast coverage calculation used to enforce minimum coverage constraints. The theoretical Section 5 claims consensus of busy-degrees under strong connectivity and continuous traffic density, and simulations on three large Beijing datasets report reductions of 50-56% in the mean standard deviation of busy-degrees and 56-68% in the proportion of over-busy antennas.

Significance. If the theoretical and empirical claims were fully established, the paper would make a useful contribution to cellular load balancing: it avoids long-term traffic prediction, links the problem to average consensus, provides a distributed implementation path, and gives a computationally cheap coverage surrogate. The real-data simulations are extensive and the paper is generally clearly structured. However, the central convergence proof rests on an incorrect matrix-norm computation, and the simulation evaluates a proxy quantity rather than the PRB-utilization busy-degree defined in Eq. (2). These issues are load-bearing for the paper's main claims, so the current version cannot be accepted without substantial revision.

major comments (4)
  1. [Section 5, Lemma 5.2] The contraction estimate in Lemma 5.2 is not correct. Equation (11) sets U2(k)=V2(k)=(1/n)1_n as the null-space vectors of the SVD, but these vectors are not orthonormal; an orthogonal SVD requires U2(k)=V2(k)=(1/√n)1_n. With the correct normalization, U1(k)U1(k)^T = I_n − (1/n)1_n1_n^T, so I_n − γU1(k)U1(k)^T = (1−γ)I_n + (γ/n)1_n1_n^T, whose row sums are exactly 1 and whose ∞-norm is 1 for 0<γ≤1. With the authors' non-normalized U2, the matrix displayed in the proof has row sums 1−γ+γ/n, not the claimed 1−((n^2−n+2)/n^2)γ. Either way, the contraction factor used in (25), (37), and (42) is false. Since Lemma 5.2, Theorem 5.1, and Proposition 5.1 all rely on this factor, the consensus proofs as written do not establish the announced convergence.
  2. [Section 5, Eq. (6) and Eq. (22)] A possible repair is to work on the invariant subspace 1^⊥, where the homogeneous part of the iteration contracts by 1−γ. For that repair, one needs d(k) ∈ 1^⊥ and also the traffic-variation term f^rel(p(t+1),t)−f^rel(p(t+1),t+1) in (36) to be in 1^⊥. The paper imposes condition (6) but does not show that it is compatible with the physical conservation law Σ_i f_i(k) r_i = z(k) together with the global target z(k)/Σ_j r_j; when the r_i are unequal, the condition can fail. The difference term is not shown to preserve the zero-sum subspace either. The theorem therefore needs an explicit equal-r_i assumption or a substantially extended argument before the consensus claim can be accepted.
  3. [Section 5, Eq. (21) vs. Step 4, Eq. (17)] The theoretical analysis replaces the clipped update (17) by the unconstrained recursion (21), with the parenthetical assumption that p_i^*(k) always lies between pmin_i(k) and pmax_i. No evidence is given that this assumption holds in the simulations that implement (17). If the minimum-coverage correction or the pmax clipping is active for any antenna in any sampling period, Lemma 5.2 and Theorem 5.1 do not apply to the simulated algorithm. The paper should report how often the constraints activate, or analyze the projected update.
  4. [Section 7.2, Eq. (2)] The simulation busy-degree is not the PRB-utilization busy-degree defined in (2); it is f_i(k) ≈ |M_i(k)|/M, a normalized count of MR samples. The proportionality between MR counts and PRB utilization is asserted but not validated against any PRB data. Moreover, the same MR data are used to estimate the Jacobian in Step 2 and to compute the post-adjustment busy-degree in the evaluation, so the reported reductions of 50–56% in standard deviation and 56–68% in over-busy proportion are established only for the MR-count proxy. A statement of this limitation and a discussion of the expected bias for true PRB utilization and live user behavior are needed.
minor comments (6)
  1. [Section 7.1] The sentence "Therefore, the sampling period is 24" should state that there are 24 sampling periods of one hour each.
  2. [Appendix A and Appendix B] In Appendix A, "We can rewrite (18) as" should refer to Eq. (20), which defines f_i through the traffic density; in Appendix B, "Combining (37) and (38) yields (23)" should refer to Eq. (25).
  3. [Section 6.2] The monotonicity argument for the MLP should state the output-layer activation and require nonnegativity of the final-layer weights; squared weights in the hidden layers alone do not guarantee monotonicity if the output layer weights are unrestricted.
  4. [References] Reference [58] contains a formatting error in the year field ("2009.]").
  5. [Abstract and Section 7.3] The abstract's statement that the coverage rate reaches 99.9% is not backed by a displayed coverage-rate result; the simulations only state the threshold F_con=99.9% in the setup.
  6. [Section 2.1] The graph G is called directed in the notation table but the neighbor relation is described as symmetric; the directed/undirected terminology should be reconciled.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the derivation is self-contained given its assumptions, though a separate correctness concern exists in Lemma 5.2.

full rationale

The paper's central derivation—BDBA/BFDBA balancing busy-degrees via Jacobian-based power updates—does not reduce to its inputs. Section 3 derives u*(k)=A+(k)d(k) from the linearized objective A u = d, with d defined as the gap to the target busy-degree, and Section 5 analyzes the closed-loop map d(k+1) = (I - gamma U1 U1^T)d(k) + traffic-variation terms. The convergence target f_i -> z/sum r_j is not built into the update by construction; it must be proved from the Laplacian structure of A. Self-citations [30]-[33] are background consensus results; the one load-bearing citation (Theorem 6.6 of [32] used in Lemma 5.1) is a standard, parameter-free graph-Laplacian rank theorem, not a bespoke result that assumes the paper's conclusion. The simulation's use of the same MR data both to estimate partial derivatives and to evaluate the resulting busy-degrees (Sections 4.1 Step 2 and 7.2) is an in-sample evaluation that limits external validity, but it is not a fitted parameter renamed as a prediction: the reported reductions are computed outcomes of the proposed update, not the objective function being re-announced. The skeptic's objection to Lemma 5.2 (the claimed infinity-norm contraction factor appears inconsistent with the paper's own SVD normalization) is a mathematical correctness issue, not a circularity: the theorem's conclusion does not follow from the displayed proof as written, but the proof is not equivalent to its assumptions.

Assumptions & free parameters 11 free parameters · 9 assumptions · 0 invented entities

The central derivation rests on a strongest-signal user model, continuity of traffic density, strong connectivity, and conservation of total traffic. The empirical claims additionally depend on the MR-count proxy for busy-degree and on the monotone MLP coverage surrogate. No new physical entities are introduced; the breathing network is a metaphor for coverage-area changes.

free parameters (11)
  • epsilon = 0.1
    Finite-difference step for MR-based approximation of the Jacobian in Section 4.1 Step 2; chosen constant, not measured.
  • gamma = 1 in all simulations
    Smoothing factor capping the magnitude of CPICH power updates (Section 4.1 Step 4); set to 1, which disables smoothing.
  • tau = 0.01
    Regularizer in BFDBA's d^(2)(k), creating the (1 - tau p_i) bias in the consensus target (Section 4.2).
  • Delta_p = 1
    Coverage-correction power increment in Section 6.2 Step 1.
  • M (maximum access user number) = 4000 (dataset-A), 5000 (B), 3500 (C)
    Normalizes MR-count busy-degree via f_i approximately |M_i|/M (Section 7.2); per-dataset hand-set value that sets the scale for the over-busy threshold.
  • r_c (coverage threshold) = -90 dBm (example)
    Received CPICH signal strength threshold used to label MR points as covered in Section 6.1 Step 2.
  • F_con = 99.9%
    Minimum coverage requirement in constraint (5) and in simulations (Section 7.2).
  • MLP architecture = 3 hidden layers; neurons 900/460/96 (A), 750/290/75 (B), 2700/1000/150 (C)
    Hyperparameters of the monotone coverage MLP described in Section 6.2 and Section 7.2; selected per dataset, not justified by an external criterion.
  • MR sampling volume V = 30,000,000
    Random sampling scale for MR data used in simulations (Section 7.2), chosen to balance speed and representativeness.
  • over-busy threshold = 0.7
    A busy-degree at or above 0.7 defines an over-busy antenna in Section 7.2; a hand-set threshold for the reported metric.
  • sampling period T = 1 hour
    Discretization period for busy-degree updates; chosen in simulations (Section 7.1).
assumptions (9)
  • domain assumption Each user accesses the antenna with the highest received CPICH signal strength.
    Section 2.1 states this assumption; it drives the Jacobian structure and the MR-based switching counts.
  • domain assumption The average traffic density function g_k(a) is continuous on R^n.
    Section 5 assumes continuity to justify differentiability and exchange of integration order in Lemma 5.1.
  • domain assumption The directed graph G of antennas is strongly connected and its out-degree matrix elements are positive.
    Required for Lemma 5.1 to conclude A(k) is a rank n-1 Laplacian.
  • domain assumption Total network traffic is unchanged when power changes, i.e., sum_i (fbar_i - f_i) = 0.
    Equation (6); used to prove solvability of A(k) u(k) = d(k).
  • domain assumption Network coverage rate F is a monotone increasing function of each antenna's CPICH transmit power.
    Section 2.3 cites [46,47] for this; used in the coverage constraint and the monotone MLP design.
  • ad hoc to paper The theoretical updates in Section 5 assume p_i(k)+gamma u_i(k) stays within [pmin_i, pmax_i], so the clipping in (17) never activates.
    Stated in Section 5 before Eq. (21); the actual BDBA/BFDBA algorithm includes the projection in (17), so the consensus theorems apply to an idealized unconstrained version.
  • domain assumption In Theorem 5.1, the traffic distribution scales as g_k = beta_k g_k*; in Propositions it satisfies the small-variation condition (26).
    These traffic-stationarity conditions are needed for the exact or approximate consensus results.
  • ad hoc to paper In simulations, the average number of accessing users is proportional to the average PRB utilization, so busy-degree can be replaced by normalized MR count.
    Section 7.2 invokes the Law of Large Numbers and traffic self-similarity; this is a practical proxy, not a measured relation.
  • standard math The trained MLP accurately approximates the coverage rate computed by the MR-based method.
    Section 6.2 relies on universal approximation; training curves in Fig. 5 are the only evidence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A "Breathing" Mobile Communication Network." pith.science (2026). https://pith.science/paper/6VQUOIYL

@misc{pith2026241117290,
  author       = {Pith},
  title        = {Pith review of: A "Breathing" Mobile Communication Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6VQUOIYL}},
  note         = {Machine review of arXiv:2411.17290}
}
read the original abstract

The frequent migration of large-scale users leads to the load imbalance of mobile communication networks, which causes resource waste and decreases user experience. To address the load balancing problem, this paper proposes a dynamic optimization framework for mobile communication networks inspired by the average consensus in multi-agent systems. In this framework, all antennas cooperatively optimize their CPICH (Common Pilot Channel) transmit power in real-time to balance their busy-degrees. Then, the coverage area of each antenna would change accordingly, and we call this framework a ``breathing'' mobile communication network. To solve this optimization problem, two algorithms named BDBA (Busy-degree Dynamic Balancing Algorithm) and BFDBA (Busy-degree Fast Dynamic Balancing Algorithm) are proposed. Moreover, a fast network coverage calculation method is introduced, by which each antenna's minimum CPICH transmit power is determined under the premise of meeting the network coverage requirements. Besides, we present the theoretical analysis of the two proposed algorithms' performance, which prove that all antennas' busy-degrees will reach consensus under certain assumptions. Furthermore, simulations carried out on three large datasets demonstrate that our cooperative optimization can significantly reduce the unbalance among antennas as well as the proportion of over-busy antennas.

Figures

Figures reproduced from arXiv: 2411.17290 by the authors.

Figure 1
Figure 1. (a) The satellite map (left) and GPS map (right) of a selected area in Beijing, where the red icons denote [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. (a) The standard deviations of busy-degrees of all antennas in Fig.1 from 9:00 to 20:00. (b) The daily [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The flow chart of BDBA and BFDBA. If we consider local optimization, ¯fi(k) (2) can be represented by ¯fi(k) (2) = rifi(k) + P j∈Ni rjfj (k) ri + P j∈Ni rj . For each sampling period k = 1, 2, . . . , we perform the following steps in turn. Step 1 : Calculate antenna i’s busy-degree fi(k) with (2). Step 2 : Approximate the partial derivative of fi with respect to pj with MR data of the k-th sampling period. In pract… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: The structure of the multi-layer perceptrons. [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]
Figure 5
Figure 5. Figure 5: The performances of the MLP corresponding to dataset-A (a); dataset-B (b); and dataset-C (c). [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]
Figure 7
Figure 7. Figure 7: The busy-degrees of the randomly selected four antennas under our algorithms. (a) Antenna 149. (b) Antenna 0 [PITH_FULL_IMAGE:figures/full_fig_p020_7.png]
Figure 10
Figure 10. Figure 10: The running time comparisons with dataset-A (a); dataset-B (b); and dataset-C (c). [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 53 canonical work pages

  1. [1]

    Elastic resource utilization framework for high capacity and energy efficiency in cloud RAN,

    D. Pompili, A. Hajisami, and T. X. Tran, “Elastic resource utilization framework for high capacity and energy efficiency in cloud RAN,” IEEE Communications Magazine, vol. 54, no. 1, pp. 26–32, 2016

    work page 2016

  2. [2]

    Reducing energy consumption by dynamic resource allocation in C-RAN,

    M. Khan, R. S. Alhumaima, and H. S. Al-Raweshidy, “Reducing energy consumption by dynamic resource allocation in C-RAN,” in European Conf. on Networks and Communications (EuCNC)., Paris, France, 2015, pp. 169–174. DOI: 10.1109/EuCNC.2015.7194062

    work page arXiv 2015

  3. [3]

    TANGO: traffic-aware network planning and green operation,

    Z. Niu, “TANGO: traffic-aware network planning and green operation,”IEEE Wireless Communications, vol. 18, no. 5, pp. 25–29, 2011

    work page 2011

  4. [4]

    Wireless traffic prediction with scalable Gaussian process: Framework, algorithms, and verification

    Y . Xu, F. Yin, W. Xu,et al., “Wireless traffic prediction with scalable Gaussian process: Framework, algorithms, and verification”, IEEE Journal on Selected Areas in Communications, vol. 37, no. 6, pp. 1291–1306, 2019

    work page 2019

  5. [5]

    Deep Learning-Based Forecasting of Cellular Network Utilization at Millisecond Resolutions,

    A. M. Nagib, H. Abou-Zeid, H. S. Hassanein, A. Bin Sediq, and G. Boudreau, “Deep Learning-Based Forecasting of Cellular Network Utilization at Millisecond Resolutions,” inIEEE International Conference on Communications, Montreal, QC, Canada, 2021, pp. 1–6. DOI: 10.1109/ICC42927.2021.9500858

    arXiv 2021

  6. [6]

    Optimal-Capacity, Shortest Path Routing in Self-Organizing 5G Networks using Machine Learning,

    C. V . Murudkar and R. D. Gitlin, “Optimal-Capacity, Shortest Path Routing in Self-Organizing 5G Networks using Machine Learning,” in IEEE 20th Wireless and Microwave Technology Conference, Cocoa Beach, FL, USA, 2019, pp. 1–5. DOI: 10.1109/W AMICON.2019.8765434

    arXiv 2019

  7. [7]

    An Efficient Physical Resource Block Assignment for Dense Femtocell Networks,

    S. Mishra, R. Thakur, and C. S. R. Murthy, “An Efficient Physical Resource Block Assignment for Dense Femtocell Networks,” in IEEE 79th Vehicular Technology Conference, Seoul, Korea (South), 2014, pp. 1–5. DOI: 10.1109/VTCSpring.2014.7022819. 26 A “Breathing” Mobile Communication Network

    work page arXiv 2014

  8. [8]

    Geometric approach based resource allocation in heterogeneous cellular networks

    X. Huang, S. Tang, D. Zhang, et al., “Geometric approach based resource allocation in heterogeneous cellular networks”, IEEE Transactions on Vehicular Technology, vol. 68, no. 12, pp. 11902–11914, 2019

    work page 2019

Show all 59 references
  1. [9]

    How much energy is needed to run a wireless network?

    G. Auer, V . Giannini, C. Desset, I. Godor,et al., “How much energy is needed to run a wireless network?” IEEE Wireless Communications, vol. 18, no. 5, pp. 40–49, 2011

    2011

  2. [10]

    Energy-efficient repulsive cell activation for heterogeneous cellular networks,

    S. Cho and W. Choi, “Energy-efficient repulsive cell activation for heterogeneous cellular networks,”IEEE Journal on Selected Areas in Communications, vol. 31, no. 5, pp. 870–882, 2013

    2013

  3. [11]

    Toward the 6G Network Era: Opportunities and Challenges,

    I. Tomkos, D. Klonidis, E. Pikasis, and S. Theodoridis, “Toward the 6G Network Era: Opportunities and Challenges,” IT Professional, vol. 22, no. 1, pp. 34–38, 2020

    2020

  4. [12]

    Base station operation and user association mechanisms for energy-delay tradeoffs in green cellular networks,

    K. Son, H. Kim, Y . Yi, and B. Krishnamachari, “Base station operation and user association mechanisms for energy-delay tradeoffs in green cellular networks,”IEEE Journal on Selected Areas in Communications, vol. 29, no. 8, pp. 1525–1536, 2011

    2011

  5. [13]

    Energy savings through dynamic base station switching in cellular wireless access networks,

    E. Oh and B. Krishnamachari, “Energy savings through dynamic base station switching in cellular wireless access networks,” in Proc. IEEE Global Telecommunications Conf. , Miami, FL, USA, 2010, pp. 1–5. DOI: 10.1109/GLOCOM.2010.5683654

    2010

  6. [14]

    Small cell base station sleep strategies for energy efficiency,

    C. Liu, B. Natarajan, and H. Xia, “Small cell base station sleep strategies for energy efficiency,"IEEE Transactions on Vehicular Technology, vol. 65, no. 3, pp. 1652–1661, 2015

    2015

  7. [15]

    Delay-Constrained Energy-Optimal Base Station Sleeping Control,

    X. Guo, Z. Niu, S. Zhou, and P. R. Kumar, “Delay-Constrained Energy-Optimal Base Station Sleeping Control," IEEE Journal on Selected Areas in Communications, vol. 34, no. 5, pp. 1073–1085, 2016

    2016

  8. [16]

    Zone-Based Load Balancing in LTE Self-Optimizing Networks: A Game-Theoretic Approach,

    M. Sheng, C. Yang, Y . Zhang, and J. Li, “Zone-Based Load Balancing in LTE Self-Optimizing Networks: A Game-Theoretic Approach," IEEE Transactions on Vehicular Technology, vol. 63, no. 6, pp. 2916–2925, 2014

    2014

  9. [17]

    Load Balancing for Ultradense Networks: A Deep Reinforcement Learning-Based Approach,

    Y . Xu, W. Xu, Z. Wang, J. Lin, and S. Cui, “Load Balancing for Ultradense Networks: A Deep Reinforcement Learning-Based Approach," IEEE Internet of Things Journal, vol. 6, no. 6, pp. 9399–9412, 2019

    2019

  10. [18]

    Load Balancing in Cellular Networks: A Reinforcement Learning Approach,

    K. Attiah, K Banawan, A Gaber, et al., “Load Balancing in Cellular Networks: A Reinforcement Learning Approach,” in IEEE 17th Annual Consumer Communications & Networking Conference (CCNC), Las Vegas, NV , USA, 2020, pp. 1–6. DOI: 10.1109/CCNC46108.2020.9045699

    2020

  11. [19]

    Deep Reinforcement Learning-based CIO and Energy Control for LTE Mobility Load Balancing,

    G. Alsuhli, H. A. Ismail, K. Alansary, M. Rumman, et al., “Deep Reinforcement Learning-based CIO and Energy Control for LTE Mobility Load Balancing,” in IEEE 18th Annual Consumer Communications & Networking Conference (CCNC), Las Vegas, NV , USA, 2021, pp. 1–6. DOI: 10.1109/CC...

    2021

  12. [20]

    Optimized Power and Cell Individual Offset for Cellular Load Balancing via Reinforcement Learning,

    G. Alsuhli, K. Banawan, K. Seddik, and A. Elezabi, “Optimized Power and Cell Individual Offset for Cellular Load Balancing via Reinforcement Learning,” in IEEE Wireless Communications and Networking Conference (WCNC), Nanjing, China, 2021, pp. 1–7. DOI: 10.1109/WCNC49053.2021.9417360

    2021

  13. [21]

    Adaptive Mobility Load Balancing Algorithm for LTE Small-Cell Networks,

    M. M. Hasan, S. Kwon, and J. H. Na, “Adaptive Mobility Load Balancing Algorithm for LTE Small-Cell Networks," IEEE Transactions on Wireless Communications, vol. 17, no. 4, pp. 2205–2217, 2018

    2018

  14. [22]

    Mobility Load Management in Cellular Networks: A Deep Reinforce- ment Learning Approach,

    G. Alsuhli, K. Banawan, A. Elezabi, et al.,“Mobility Load Management in Cellular Networks: A Deep Reinforce- ment Learning Approach," IEEE Transactions on Mobile Computing, vol. 22, no. 3, pp. 1581–1598, 2023

    2023

  15. [23]

    Interpretability of deep learning models: A survey of results,

    S. Chakraborty, et al.,“Interpretability of deep learning models: A survey of results," in 2017 IEEE Smart- World/SCALCOM/UIC/ATC/CBDCom/IOP/SCI, San Francisco, CA, USA, 2017, pp. 1–6. DOI: 10.1109/UIC- ATC.2017.8397411

    2017

  16. [24]

    A survey of methods for explaining black box models,

    R. Guidotti, A. Monreale, S. Ruggieri, et al.,“A survey of methods for explaining black box models," ACM computing surveys (CSUR), vol. 51, no. 5, pp. 1–42, 2018

    2018

  17. [25]

    Characteristics of scalability and their impact on performance,

    A.B. Bondi,“Characteristics of scalability and their impact on performance," in Proceedings of the 2nd interna- tional workshop on Software and performance, 2000, pp. 195–203

    2000

  18. [26]

    Optimizing the power consumption of mobile networks based on traffic prediction,

    S. Dawoud, A. Uzun, S. Göndör, and A. Küpper, “Optimizing the power consumption of mobile networks based on traffic prediction,” in Proc. 38th IEEE Annual Computer Software and Applications Conf., Vasteras, Sweden, 2014, pp. 279–288. DOI: 10.1109/COMPSAC.2014.38

    2014 doi

  19. [27]

    Spatiotemporal modeling and prediction in cellular networks: A big data enabled deep learning approach,

    J. Wang, J. Tang, Z. Xu, Y . Wang,et al., “Spatiotemporal modeling and prediction in cellular networks: A big data enabled deep learning approach,” in Proc. IEEE Conf. on Computer Communications, Atlanta, GA, USA, 2017, pp. 1–9. DOI: 10.1109/INFOCOM.2017.8057090

    2017

  20. [28]

    Spatio-temporal analysis and prediction of cellular traffic in metropolis,

    X. Wang, Z. Zhou, F. Xiao,et al., “Spatio-temporal analysis and prediction of cellular traffic in metropolis,” IEEE Transactions on Mobile Computing, vol. 18, no. 9, pp. 2190–2202, 2018. 27 A “Breathing” Mobile Communication Network

    2018

  21. [29]

    Wireless Traffic Prediction With Scalable Gaussian Process: Framework, Algorithms, and Verification,

    Y . Xu, F. Yin, W. Xu, J. Lin, and S. Cui, “Wireless Traffic Prediction With Scalable Gaussian Process: Framework, Algorithms, and Verification,”IEEE Journal on Selected Areas in Communications, vol. 37, no. 6, pp. 1291–1306, 2019

    2019

  22. [30]

    The smallest possible interaction radius for synchronization of self-propelled particles,

    G. Chen, Z. Liu, and L. Guo, “The smallest possible interaction radius for synchronization of self-propelled particles,” SIAM Review, vol. 56, no. 3, pp. 499–521, 2014

    2014

  23. [31]

    Small noise may diversify collective motion in vicsek model,

    G. Chen, “Small noise may diversify collective motion in vicsek model,”IEEE Transactions on Automatic Control, vol. 62, no. 2, pp. 636–651, 2017

    2017

  24. [32]

    Critical connectivity and fastest convergence rates of distributed consensus with switching topologies and additive noises,

    G. Chen, C. Chen, G. Yin, and L. Y . Wang, “Critical connectivity and fastest convergence rates of distributed consensus with switching topologies and additive noises,” IEEE Transactions on Automatic Control, vol. 62, no. 12, pp. 6152–6167, 2017

    2017

  25. [33]

    Linear stochastic approximation algorithms and group consensus over random signed networks,

    G. Chen, X. Duan, W. Mei, and F. Bullo, “Linear stochastic approximation algorithms and group consensus over random signed networks,” IEEE Transactions on Automatic Control, vol. 64, no. 5, pp. 1874–1889, 2019

    2019

  26. [34]

    Dynamic load balancing for distributed memory multiprocessors,

    G. Cybenko, “Dynamic load balancing for distributed memory multiprocessors,” Journal of Parallel and Distributed Computing , vol. 7, no. 2, pp. 279–301, 1989. [Online]. Available: https://www.sciencedirect.com/science/article/pii/074373158990021X

    1989

  27. [35]

    A survey of various propagation models for mobile communication,

    T. K. Sarkar, Z. Ji, K. Kim, A. Medouri, and M. Salazar-Palma, “A survey of various propagation models for mobile communication,” IEEE Antennas and propagation Magazine, vol. 45, no. 3, pp. 51–82, 2003

    2003

  28. [36]

    An overview of outdoor propagation prediction models,

    S. Hamim and M. F. Jamlos, “An overview of outdoor propagation prediction models,” in Proc. Int. Symp. on Telecommunication Technologies, Langkawi, Malaysia, 2014, pp.385–388. DOI: 10.1109/ISTT.2014.7238240

    2014

  29. [37]

    Mobile small cell deployment for next generation cellular networks,

    S. F. Chou, T. C. Chiu, Y . J. Yu, and A. C. Pang, “Mobile small cell deployment for next generation cellular networks,” in 2014 IEEE Global Communications Conference , Austin, TX, USA, 2014, pp.4852–4857. DOI: 10.1109/GLOCOM.2014.7037574

    2014

  30. [38]

    Static and dynamic channel assignment using neural networks,

    K. Smith and M. Palaniswami, “Static and dynamic channel assignment using neural networks,” IEEE Journal on Selected Areas in Communications, vol. 15, no. 2, pp. 238–249, 1997

    1997

  31. [39]

    The capacity region of the gaussian mimo broadcast channel,

    H. Weingarten, Y . Steinberg, and S. Shamai, “The capacity region of the gaussian mimo broadcast channel,” in Proc. Int. Symp. on Information Theory, Chicago, IL, USA, 2004, pp. 174. DOI: 10.1109/ISIT.2004.1365211

    2004 arXiv

  32. [40]

    Automated optimization of service coverage and base station antenna configuration in umts networks,

    I. Siomina, P. Varbrand, and D. Yuan, “Automated optimization of service coverage and base station antenna configuration in umts networks,” IEEE Wireless Communications, vol. 13, no. 6, pp. 16–25, 2006

    2006

  33. [41]

    Network selection algorithm based on link quality parameters for heteroge- neous wireless networks,

    K. Ahuja, B. Singh, and R. Khanna, “Network selection algorithm based on link quality parameters for heteroge- neous wireless networks,” Optik, vol. 125, no. 14, pp. 3657–3662, 2014

    2014

  34. [42]

    Joint allocation of resource blocks, power, and energy-harvesting relays in cellular networks,

    S. Jangsher, H. Zhou, V . O. Li, and K.-C. Leung, “Joint allocation of resource blocks, power, and energy-harvesting relays in cellular networks,” IEEE Journal on Selected Areas in Communications, vol. 33, no. 3, pp. 482–495, 2015

    2015

  35. [43]

    A sleep-scheduling scheme for enhancing QoS and network coverage in IEEE 802.15. 4 WSN,

    T. Semprebom, C. Montez, G. Araújo, and P. Portugal, “A sleep-scheduling scheme for enhancing QoS and network coverage in IEEE 802.15. 4 WSN,” in Proc. IEEE World Conf. on Factory Communication Systems, Palma de Mallorca, Spain, 2015, pp. 1–4. DOI: 10.1109/WFCS.2015.7160579

    2015

  36. [44]

    Coverage rate calculation in wireless sensor networks,

    S. S. Kashi and M. Sharifi, “Coverage rate calculation in wireless sensor networks,”Computing, vol. 94, no. 11, pp. 833–856, 2012

    2012

  37. [45]

    Intelligent mobile communication network plan for 5G based on insight big data,

    C. G. Shan, Z. J. Ming, H. X. Yi, L. Nan, and L. Lei, “Intelligent mobile communication network plan for 5G based on insight big data,” inProc. 7th Int. Conf. on Signal and Information Processing, Networking and Computers, Springer Singapore, 2021, pp. 1068–1074

    2021

  38. [46]

    On Received-Signal-Strength Based Localization with Unknown Transmit Power and Path Loss Exponent,

    G. Wang, H. Chen, Y . Li and M. Jin, “On Received-Signal-Strength Based Localization with Unknown Transmit Power and Path Loss Exponent,” IEEE Wireless Communications Letters, vol. 1, no. 5, pp. 536–539, 2012

    2012

  39. [47]

    T. S. Rappaport, Wireless Communications: Principles and Practice, Prentice-Hall, 1996

    1996

  40. [48]

    Strang, Linear algebra and its applications, 4th ed

    G. Strang, Linear algebra and its applications, 4th ed. Belmont, CA: Thomson, Brooks/Cole, 2006

    2006

  41. [49]

    Bullo, Lectures on network systems, 1.4 ed

    F. Bullo, Lectures on network systems, 1.4 ed. Kindle Direct Publishing, 2019

    2019

  42. [50]

    Stochastic geometry based models for modeling cellular networks in urban areas,

    C.-H. Lee, C.-Y . Shih, and Y .-S. Chen, “Stochastic geometry based models for modeling cellular networks in urban areas,” Wireless networks, vol. 19, no. 6, pp. 1063–1072, 2013

    2013

  43. [51]

    Spatial stochastic models and metrics for the structure of base stations in cellular networks,

    A. Guo and M. Haenggi, “Spatial stochastic models and metrics for the structure of base stations in cellular networks,” IEEE Transactions on Wireless Communications, vol. 12, no. 11, pp. 5800–5812, 2013

    2013

  44. [52]

    Modeling spatial structure of wireless communication networks,

    J. Riihijarvi and P. Mahonen, “Modeling spatial structure of wireless communication networks,” inProc. IEEE Conf. on Computer Communications Workshops , San Diego, CA, USA, 2010, pp. 1–6. DOI: 10.1109/INF- COMW.2010.5466702. 28 A “Breathing” Mobile Communication Network

    2010

  45. [53]

    V . A. Zorich,Mathematical analysis I, 2nd ed. Berlin: Springer, 2016

    2016

  46. [54]

    Approximation theory of the MLP model in neural networks,

    A. Pinkus, “Approximation theory of the MLP model in neural networks,”Acta Numerica, vol. 9, pp. 143–195, 1999

    1999

  47. [55]

    Approximations of functions by a multilayer perceptron: a new approach,

    J. G. Attali, G. Pagès, “Approximations of functions by a multilayer perceptron: a new approach,”Neural networks, vol. 10, no. 6, pp. 1069–1081, 1997

    1997

  48. [56]

    Binary cross entropy with deep learning technique for image classification,

    U. Ruby and V . Yendapalli, “Binary cross entropy with deep learning technique for image classification,”Int. J. Adv. Trends Comput. Sci. Eng, vol. 9, no. 10, pp. 5395–5397, 2020

    2020

  49. [57]

    On the self-similar nature of Ethernet traffic (extended version),

    W. Leland, M. Taqqu, W. Willinger, and D. Wilson, “On the self-similar nature of Ethernet traffic (extended version)," IEEE/ACM Transactions on networking, vol. 2, no. 1, pp. 1–15, 1994

    1994

  50. [58]

    The large deviation approach to statistical mechanics,

    H. Touchette, “The large deviation approach to statistical mechanics," Physics Reports, vol. 478, no. 1, pp. 1–69, 2009.]

    2009

  51. [59]

    Deployment Algorithms for UA V Airborne Networks Toward On-Demand Coverage,

    H. Zhao, H. Wang, W. Wu, and J. Wei, “Deployment Algorithms for UA V Airborne Networks Toward On-Demand Coverage," IEEE Journal on Selected Areas in Communications, vol. 36, no. 9, pp. 2015–2031, 2018. 29

    2015

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.