Pith. sign in

REVIEW 4 major objections 5 minor 34 references

Multi-Layer Perceptron-Based Relay Node Selection for Next-Generation Intelligent Delay-Tolerant Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read This paper claims that replacing Spray and Wait's blind copying with MLP-based relay evaluation improves delivery probability by 7-8 percentage points and cuts average latency by 200-300 seconds in simulated urban emergency scenarios.

desk verdict The real-time relay-selection claim is undercut by temporal leakage in the feature definitions; the reported routing gains are unsupported as stated, but the architecture is clean and the issue is fixable. read the letter →

arxiv 2509.11239 v1 pith:WUXIH7RL submitted 2025-09-14 cs.NI

Zhekun Huang , Milena Radenkovic This is my paper · ORCID
classification cs.NI
keywords Delay-TolerantNetworksrelayselectionSprayandWaitmulti-layerperceptronmachinelearningroutingemergencycommunicationopportunisticreal-timeclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that a delay-tolerant network can route messages more reliably by learning, from past contact and delivery logs, which nodes make good relays, rather than spraying copies blindly. It integrates a feedforward neural-network classifier into Spray and Wait, using seven per-node features—contact frequency, degree, average contact duration, average hop count, average delivery time, relay count, and destination count—to label nodes high- or low-quality and forward only to high-quality nodes. In simulated weekday and holiday emergency scenarios with pedestrians and vehicles, the resulting MLPBasedSprayRouter improves delivery probability by about 7-8 percentage points and reduces average latency by 200-300 seconds compared with baseline Spray and Wait. If correct, this would show that lightweight online ML inference can be bolted onto a standard DTN protocol to make it adaptive.

What carries the argument

The load-bearing mechanism is a median-split labeling scheme plus a two-hidden-layer multi-layer perceptron classifier served through a lightweight HTTP prediction service and cached inside the router. Seven node-level statistics are computed from contact and delivery logs; each node's features are normalized and scored, nodes above the median are labeled high-quality, and the model learns the decision boundary. At runtime the router queries the service at each encounter and forwards only to predicted high-quality peers. The caching mechanism is what makes per-encounter inference cheap enough to sit inside a forwarding loop.

What would settle it

Run the MLPBasedSprayRouter in the same simulator while logging exactly which features are available in each node's local state at each encounter. If the 'real-time' feature set includes values from delivered-message report entries for messages not yet delivered, the claim of real-time prediction is falsified; a concrete test is to remove those three outcome features, retrain on local-only features, and check whether the delivery probability gain over Spray and Wait persists.

Watch

Extended reading notes

Core claim

The central claim is that relay selection in delay-tolerant networks can be treated as a binary classification problem and solved with a shallow neural network trained offline and queried at encounter time. The authors propose MLPBasedSprayRouter, a modified Spray and Wait router that extracts seven features from simulator logs, sends them over HTTP to a cached prediction service, and forwards a message copy only when the peer is classified as a high-quality relay. Their experiments across nine pedestrian/vehicle configurations and two temporal patterns show consistent delivery gains and latency reductions over the static baseline, with the MLP classifier outperforming the SVM and random for

Load-bearing premise

The strongest claim assumes that the seven features—including average delivery time, relay count, and destination count—can be computed in real time at the moment of a forwarding decision, but those values are outcome statistics that only exist after a delivery report is complete.

Editorial extensions

If this is right

  • If the central claim holds, standard DTN protocols can be upgraded by an external prediction service without rewriting the routing core.
  • The 7-8 percentage point delivery gain and 200-300 second latency reduction suggest that selective forwarding beats blind replication even when the number of message copies is limited.
  • MLP beating SVM and random forest at low inference cost hints that shallow neural networks offer a good accuracy-versus-speed trade-off for encounter-time decisions.
  • Stability across weekday and holiday mobility patterns suggests the model generalizes across temporal traffic conditions, not just one scenario.
  • Caching repeated feature patterns implies per-contact overhead can be bounded, making ML-based decisions viable in resource-constrained nodes.

Reading between the lines

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

  • The paper's 'real-time' framing likely overstates what is available at decision time: features such as average delivery time, relay count, and destination count come from a completed-delivery report, so a node in the field would not know them yet; the reported gains may depend on access to future statistics.
  • A fair field test would restrict features to locally observable encounter history—contact frequency, degree, average duration, and hop counts of messages seen so far—and re-measure delivery; if the gains vanish, the classifier was exploiting offline labels rather than predictive context.
  • The median-split labeling ties 'high-quality' to an end-of-run summary rather than to a per-forwarding decision, so the architecture is better described as offline feature engineering plus online scoring than as online learning.
  • The modular HTTP service design suggests a path to online retraining: as delivery reports accumulate, the model could be refreshed without recompiling the router, but that would require solving the feature-availability problem first.
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 / 5 minor

Summary. The paper proposes an ML-enhanced variant of Spray and Wait routing for delay-tolerant networks (DTNs). The authors extract node-level features from ONE simulator logs (contact frequency, degree, average contact duration, average hop count, average delivery time, relay count, destination count), label nodes via a median split of an aggregate score, and train MLP, SVM, and Random Forest classifiers. The best model, MLP, is deployed via a Flask REST API and integrated into a custom MLPBasedSprayRouter, which forwards messages only to nodes predicted as high-quality relays. Experiments on a simulated emergency scenario with weekday and holiday mobility report that the MLP-based router improves delivery probability by about 7-8% and reduces average latency by 200-300 seconds compared to baseline Spray and Wait, with an unspecified RandomRouter baseline also shown. The paper claims the approach is practical and real-time, and proposes future online learning and a modular testbed.

Significance. If the reported gains were causally valid, the paper would demonstrate a modest but useful improvement to a well-known DTN routing protocol through a modular ML integration. The comparison of three classifiers and the inclusion of simulated weekday/holiday scenarios are also potentially informative. However, the core evaluation is undermined by a severe temporal-leakage problem: the feature set includes delivery outcomes that are not available at routing time, and the labels are derived from the same features. The reported model accuracy and end-to-end delivery improvements may therefore reflect post-hoc information rather than a genuinely adaptive learned policy. Other issues, including an undefined RandomRouter baseline and the absence of statistical variability measures, further prevent the results from being interpreted. The paper currently does not support its central claim.

major comments (4)
  1. [Section 3.2.1] The feature set includes Average Delivery Time, Average Hop Count, Relay Count, and Destination Count, all stated to be derived from the DeliveredMessagesReport after simulation completion. At the moment a forwarding decision is made, these outcome statistics are not known for the message being routed; they become available only after delivery. The paper claims 'real-time feature extraction' in Section 3.1.2, but it never specifies how a node could obtain these values for a candidate peer from causally available information. This is a temporal-leakage problem: the classifier is trained on features that encode the future, and the routing decisions in Tables 4.4-4.5 may be exploiting lookahead information. The authors need to either redefine the features in terms of purely historical, locally observable statistics or provide an explicit mechanism explaining how average delivery time and de
  2. [Section 3.2.1 / 3.2.4] The binary label is assigned by a median split of an aggregate score built from the same normalized features that the classifier consumes. Consequently, the reported classification accuracy (e.g., F1=0.733 for MLP in Table 4.3) measures how well the model reproduces the authors' own scoring rule, not whether it identifies causally good relays. In particular, because the score includes delivery outcomes, a node that has historically delivered messages quickly will be labeled good, and the same historical delivery time is used as a feature. The circularity must be broken by deriving labels from independent ground truth, such as whether forwarding to a node ultimately improved delivery in a held-out causal evaluation, or by using only features available at decision time.
  3. [Section 4.4, Tables 4.4 and 4.5] The RandomRouter baseline is not defined anywhere in the paper: no algorithm description, no reference, no explanation of how it differs from MLPBasedRouter, and no statement of its role. Without this information, the comparison is uninterpretable. Additionally, the protocols were run for five random seeds (Section 4.2), but the tables report only point estimates without standard deviations, confidence intervals, or per-seed results. The claimed +7-8% delivery improvement and 200-300s latency reduction could fall within seed-to-seed variability; the paper must report error bars or a statistical test before these improvements can be accepted.
  4. [Section 4.3] The manuscript does not state whether the mobility configurations used in the router evaluation (weekday and holiday scenarios) are disjoint from the data used to train the classifiers. The dataset is described as being aggregated from 'nine mobility configurations' (Section 3.2.3), and the same types of scenarios appear in the evaluation. If the weekday/holiday runs overlap with the training configurations, the reported delivery gains may reflect memorization of specific scenario traces rather than generalization. The authors should specify the exact training/test split at the scenario level and, ideally, evaluate on unseen configurations.
minor comments (5)
  1. [Section 4.1] Typo: 'pedistrains' should be 'pedestrians'.
  2. [Section 4, table numbering] Table 4.4 and Table 4.5 are referred to as 'Tables 4.4 and 4.5' in the text, but the in-text references use the format 'Table 4.4-4.5' and the table captions do not follow a consistent numbering scheme. Please standardize.
  3. [References] Reference [34] is listed as 'F. Author and S. Author, "A modular dtn testbed for opportunistic networks: Moditones"'—this is a placeholder and must be replaced with the actual citation. Reference [4] also has a typo in the author name ('Vu San').
  4. [Section 3.2.1] The definitions of Average Hop Count and Average Delivery Time are ambiguous: do they refer only to messages that have been fully delivered, or also to messages currently in the buffer? This matters for the causal-feature discussion and should be clarified.
  5. [Figure 3.5] Caption is incomplete: 'Distribution of features across' does not specify across what (presumably high- vs. low-quality nodes) or which scenario/configuration. Please complete the caption.

Circularity Check

2 steps flagged · score 6.0 of 10

Self-defined labels and post-hoc delivery features make the MLP 'prediction' a fit to its own inputs, undermining the reported real-time routing gains.

  1. self definitional [Section 3.2.1, 'Selected Features and Labeling Strategy']
    "Each node is assigned a score by aggregating its normalized feature values (optionally weighted). A binary label is then assigned based on a median split: nodes above the median are labeled as 1 (high-quality), and those below as 0."

    The classifier input is exactly the set of normalized features (contact frequency, degree, average contact duration, average hop count, average delivery time, relay count, destination count) listed earlier in the same section. The label is therefore the median split of the aggregate of those same feature values. Model accuracy (Table 4.3) measures how well the classifier reproduces the authors' scoring rule, not an independently measured relay quality. The subsequent claim that the MLP 'predicts whether a node is suitable as a relay' is a renaming of the score-based median split.

  2. fitted input called prediction [Sections 3.1.2 and 3.2.1]
    "During message forwarding, each node performs feature extraction in real time and submits an HTTP request to the deployed Flask API. ... two internal reports—ConnectivityDtnsim2R and DeliveredMessagesReport—were utilized to derive features ... Average Delivery Time (tdelay): Mean delivery delay for relayed messages."

    The features fed to the 'real-time' classifier include average delivery time, relay count, and destination count taken from DeliveredMessagesReport, which records completed deliveries. At the moment a forwarding decision is made, the message has not been delivered and these statistics do not exist. Either the deployed router uses a different, causally available feature vector than the one used for training (train/serve skew), or it queries the API with post-hoc outcome data (lookahead leakage). In both cases the reported +7-8% delivery gain of MLPBasedRouter (Tables 4.4-4.5) is not evidence for a learned online policy; it is an artifact of fitting to the outcome the router is supposed to predict.

full rationale

The central ML 'prediction' is not an independent forecast. Section 3.2.1 defines the binary label as the median split of an aggregate score computed from the same normalized features that the classifiers consume. Thus the classification performance reported in Table 4.3 measures the model's ability to reconstruct the authors' own scoring rule, a self-definitional target. More seriously, the same section derives features such as Average Delivery Time, Relay Count, and Destination Count from DeliveredMessagesReport, which only exists after messages are delivered. The paper nevertheless claims (Section 3.1.2) that nodes extract these features 'in real time' during forwarding. This is a temporal-leakage inconsistency: either the deployed router cannot compute the trained feature vector at decision time, or it uses future outcome statistics. The +7-8% delivery probability improvement and 200-300s latency reduction (Section 4.4, Tables 4.4-4.5) are therefore not attributable to a causally valid learned policy; they are consistent with an oracle that knows historical delivery outcomes. The comparison against Spray and Wait remains an empirical simulation, but as described it is not a valid online prediction experiment. This is partial circularity - the classifier's target is constructed from its own inputs, and the router's supposed real-time advantage is fitted from post-hoc reports - so score 6.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim rests on hand-defined labels derived from the same features used for prediction, and on the availability of future simulation outcomes at decision time.

free parameters (5)
  • Median split threshold = 50th percentile
    Nodes are labeled high-quality if their aggregate normalized score is above the median, a binary threshold chosen by the authors.
  • MLP hidden layer sizes = 128, 64
    Tuned via grid search; central claim of MLP superiority depends on this configuration.
  • SVM hyperparameters (C, gamma) = C=1, gamma=0.1
    Selected by grid search with 5-fold CV.
  • Random Forest hyperparameters = 200 trees, max_depth=10
    Selected by grid search.
  • Feature weighting in score = equal weights (unstated)
    The score aggregates normalized features 'optionally weighted'; no weights are reported, leaving an unspecified degree of freedom in label construction.
assumptions (3)
  • domain assumption Simulation in ONE with ShortestPathMapBasedMovement represents realistic emergency DTN mobility
    Evaluation claims practical relevance based on a single simulated Helsinki downtown scenario.
  • ad hoc to paper Median split of the aggregate feature score yields a meaningful binary relay-quality label
    Section 3.2.1 defines labels this way; there is no external validation of the label.
  • ad hoc to paper The extracted features are computable at routing time from information available to the node
    Section 3.1.2 assumes real-time feature extraction, but several features (avg delivery time, relay count, destination count) are outcome statistics from the full simulation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Layer Perceptron-Based Relay Node Selection for Next-Generation Intelligent Delay-Tolerant Networks." pith.science (2026). https://pith.science/paper/WUXIH7RL

@misc{pith2026250911239,
  author       = {Pith},
  title        = {Pith review of: Multi-Layer Perceptron-Based Relay Node Selection for Next-Generation Intelligent Delay-Tolerant Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WUXIH7RL}},
  note         = {Machine review of arXiv:2509.11239}
}
read the original abstract

Delay Tolerant Networks (DTNs) are critical for emergency communication in highly dynamic and challenging scenarios characterized by intermittent connectivity, frequent disruptions, and unpredictable node mobility. While some protocols are widely adopted for simplicity and low overhead, their static replication strategy lacks the ability to adaptively distinguish high-quality relay nodes, often leading to inefficient and suboptimal message dissemination. To address this challenge, we propose a novel intelligent routing enhancement that integrates machine learning-based node evaluation into the Spray and Wait framework. Several dynamic, core features are extracted from simulation logs and are used to train multiple classifiers - Multi-Layer Perceptron (MLP), Support Vector Machine (SVM), and Random Forest (RF) - to predict whether a node is suitable as a relay under dynamic conditions. The trained models are deployed via a lightweight Flask-based RESTful API, enabling real-time, adaptive predictions. We implement the enhanced router MLPBasedSprayRouter, which selectively forwards messages based on the predicted relay quality. A caching mechanism is incorporated to reduce computational overhead and ensure stable, low-latency inference. Extensive experiments under realistic emergency mobility scenarios demonstrate that the proposed framework significantly improves delivery ratio while reducing average latency compared to the baseline protocols. Among all evaluated classifiers, MLP achieved the most robust performance, consistently outperforming both SVM and RF in terms of accuracy, adaptability, and inference speed. These results confirm the novelty and practicality of integrating machine learning into DTN routing, paving the way for resilient and intelligent communication systems in smart cities, disaster recovery, and other dynamic environments.

Figures

Figures reproduced from arXiv: 2509.11239 by the authors.

Figure 3
Figure 3. presents a series of kernel [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 1 canonical work pages

  1. [1]

    A delay-tolerant network architecture for challenged internets,

    K. Fall, “A delay-tolerant network architecture for challenged internets,” Proceedings of CCR, pp. 27-34, 2003

    2003

  2. [2]

    Spray and wait: an efficient routing scheme for intermittently connected mobile networks,

    T. Spyropoulos, K. Psounis, and C. S. Raghavendra, “Spray and wait: an efficient routing scheme for intermittently connected mobile networks,” in Proceedings of the 2005 ACM SIGCOMM workshop on Delay-tolerant networking, 2005, pp. 252- 259

    2005

  3. [3]

    Machine learning for routing in delay-tolerant networks: A survey,

    N. Mohamed and S. Boudjit, “Machine learning for routing in delay-tolerant networks: A survey,” Computer Networks, vol. 191, p. 108011, 2021

    2021

  4. [4]

    Low-cost mobile personal clouds,

    M. Radenkovic and H. H. Vu San, "Low-cost mobile personal clouds," 2016 International Wireless Communications and Mobile Computing Conference (IWCMC), Paphos, Cyprus, 2016, pp. 836-841, doi: 10.1109/IWCMC.2016.7577166

    arXiv 2016

  5. [6]

    Optimization of energy consumption in delay-tolerant net- works,

    W. Junran and et al., “ Optimization of energy consumption in delay-tolerant net- works,” Journal of Network and Computer Applications, 2023

    2023

  6. [7]

    Peoplerank: Social opportunistic for- warding,

    A. Mtibaa, M. May, C. Diot, and M. Ammar, “Peoplerank: Social opportunistic for- warding,” IEEE INFOCOM, pp. 1-5, 2010

    2010

  7. [8]

    Follow-me cloud: When cloud services follow mobile users,

    T. Taleb and A. Ksentini, “Follow-me cloud: When cloud services follow mobile users,” IEEE Communications Magazine, 2022

    2022

  8. [9]

    Epidemic routing for partially-connected ad hoc networks,

    A. Vahdat and D. Becker, “Epidemic routing for partially-connected ad hoc networks,” Duke University, Technical Report CS-2000-06, 2000

    2000

Show all 34 references
  1. [10]

    Opportunistic networking: Data forwarding in disconnected mobile ad hoc networks,

    L. Pelusi, A. Passarella, and M. Conti, “ Opportunistic networking: Data forwarding in disconnected mobile ad hoc networks,” IEEE Communications Magazine, vol. 44, no. 11, pp. 134-141, 2006

    2006

  2. [11]

    A review of machine learning ap- plications in delay-tolerant networks,

    A. Bhavani, A. V. Ramana, and A. Chakravarthy, “A review of machine learning ap- plications in delay-tolerant networks,” Journal of Network and Computer Applications, vol. 193, p. 103439, 2022

    2022

  3. [12]

    An improved hybrid routing protocol combining manet and dtn,

    M. W. Kan and Y. W. Chung, “An improved hybrid routing protocol combining manet and dtn,” 2020

    2020

  4. [13]

    A machine learning-based protocol for efficient routing in opportunistic networks,

    D. K. Sharma, S. K. Dhurandher, I. Woungang, and R. K. Srivastava, “A machine learning-based protocol for efficient routing in opportunistic networks,” IEEE Systems Journal, 2016

    2016

  5. [14]

    Milena Radenkovic and Andrew Grundy. 2011. Congestion aware data dissemination in social opportunistic networks. SIGMOBILE Mob. Comput. Commun. Rev. 14, 3 (July 2010), 31–33, doi: 10.1145/1923641.1923653

    2011

  6. [15]

    Cognitive Caching at the Edges for Mobile Social Community Networks: A Multi-Agent Deep Reinforcement Learning Approach,

    M. Radenkovic and V. S. H. Huynh, "Cognitive Caching at the Edges for Mobile Social Community Networks: A Multi-Agent Deep Reinforcement Learning Approach," in IEEE Access, vol. 8, pp. 179561-179574, 2020, doi: 10.1109/ACCESS.2020.3027707

    2020

  7. [16]

    Enabling Real -time Communications and Services in Heterogeneous Networks of Drones and Vehicles,

    M. Radenkovic, V. S. Ha Huynh, R. John and P. Manzoni, "Enabling Real -time Communications and Services in Heterogeneous Networks of Drones and Vehicles," 2019 International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), Barcelona, Spain, 2...

    2019

  8. [17]

    and Radenkovic M

    Huynh V. and Radenkovic M. (2019). Interdependent Multi -layer Spatial Temporal -based Caching in Heterogeneous Mobile Edge and Fog Networks.In Proceedings of the 9th International Confe rence on Pervasive and Embedded Computing and Communication Systems - Volume 1: PECCS, ISB...

    2019 doi

  9. [18]

    A random forest guided tour,

    G. Biau and E. Scornet, “A random forest guided tour,” Test, vol. 25, pp. 197-227, 2016

    2016

  10. [19]

    Review on convolutional neural networks (cnn) in vegetation remote sensing,

    T. Kattenborn, J. Leitloff, F. Schiefer, and S. Hinz, “Review on convolutional neural networks (cnn) in vegetation remote sensing,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 173, pp. 24–49, 2021

    2021

  11. [20]

    Lstm time and frequency recurrence for automatic speech recognition,

    J. Li, A. Mohamed, G. Zweig, and Y. Gong, “Lstm time and frequency recurrence for automatic speech recognition,” in 2015 IEEE Workshop on Automatic Speech Recog- nition and Understanding (ASRU) . IEEE, 2015, pp. 187–191

    2015

  12. [21]

    The one simulator for dtn protocol eval- uation,

    A. Ker¨anen, J. Ott, and T. K¨arkk¨ainen, “The one simulator for dtn protocol eval- uation,” in Proceedings of the 2nd international conference on simulation tools and techniques. ICST (Institute for Computer Sciences, Social-Informatics and Telecom- munications Engineering), ...

    2009

  13. [22]

    Fibonary spray and wait routing in de- lay tolerant networks,

    P. Das, P. Chowdhury, B. Poudel, and T. De, “Fibonary spray and wait routing in de- lay tolerant networks,” International Journal of Electrical and Computer Engineering, vol. 6, no. 6, p. 3205, 2016

    2016

  14. [23]

    A machine learning concept for dtn routing,

    R. Dudukovich, A. Hylton, and C. Papachristou, “A machine learning concept for dtn routing,” NASA Glenn Research Center / Case Western Reserve University, Tech. Rep., 2017

    2017

  15. [24]

    Application of machine learning techniques to delay-tolerant networking,

    NASA, “ Application of machine learning techniques to delay-tolerant networking,” Tech. Rep., 2019

    2019

  16. [25]

    Carl-dtn: Context adaptive reinforcement learning based routing algorithm in delay tolerant network,

    F. Y. Yesuf and M. Prathap, “Carl-dtn: Context adaptive reinforcement learning based routing algorithm in delay tolerant network,” arXiv preprint arXiv:2105.00544 , 2021

    2021 arXiv

  17. [26]

    Socialized proficient routing in opportunistic mobile networks using machine learning techniques,

    V. Rajendran and G. K. Thonnuthodi, “Socialized proficient routing in opportunistic mobile networks using machine learning techniques,” International Journal of Intelli- gent Engineering and Systems, vol. 13, no. 4, 2020

    2020

  18. [27]

    A survey on support vector machine techniques in delay tolerant networks,

    R. S. Sutton, A. Ghosh, and P. Ghosal, “A survey on support vector machine techniques in delay tolerant networks,” International Journal of Computer Applications, vol. 179, no. 41, pp. 1–6, 2018

    2018

  19. [28]

    Learning to fairly classify the quality of wireless links,

    G. Cerar, H. Yetgin, M. Mohorˇciˇc, and C. Fortuna, “Learning to fairly classify the quality of wireless links,” ArXiv preprint arXiv:2102.11655 , 2021

    2021 arXiv

  20. [29]

    A review on machine learning based routing protocols for delay tolerant networks,

    A. Bhavani and A. V. Ramana, “A review on machine learning based routing protocols for delay tolerant networks,” International Journal of Digital Technology, 2022

    2022

  21. [30]

    Clip- per: A low-latency online prediction serving system,

    D. Crankshaw, X. Wang, G. Zhou, M. J. Franklin, J. E. Gonzalez, and I. Stoica, “Clip- per: A low-latency online prediction serving system,” arXiv preprint arXiv:1612.03079 , 2016

    2016 arXiv

  22. [31]

    A survey on smart cities infrastructure: Challenges and solutions,

    Z. Zhang, L. Xu, W. Liu, and Y. Wang, “A survey on smart cities infrastructure: Challenges and solutions,” Computer Networks, vol. 196, p. 108153, 2021

    2021

  23. [32]

    Optimal routing with node classification (ornc) for dtn over bluetooth/manet,

    A. Tripathia and V. Mekathotia, “Optimal routing with node classification (ornc) for dtn over bluetooth/manet,” in SBRC 2024 , 2024

    2024

  24. [33]

    Delay tolerant network routing as a machine learning classification problem,

    R. Dudukovich and C. Papachristou, “Delay tolerant network routing as a machine learning classification problem,” 2018

    2018

  25. [34]

    A modular dtn testbed for opportunistic networks: Modi- tones,

    F. Author and S. Author, “A modular dtn testbed for opportunistic networks: Modi- tones,” in IEEE Transactions on Network and Service Management, 2016

    2016

  26. [2024]

    Available: https:// arxiv.org/abs/2411.06455

    [Online]. Available: https:// arxiv.org/abs/2411.06455

    arXiv

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.