Towards Energy Impact on AI-Powered 6G IoT Networks: Centralized vs. Decentralized

Andreas Weinand; Anjie Qiu; Donglin Wang; Hans D. Schotten; Sanket Partani

arxiv: 2604.19377 · v1 · submitted 2026-04-21 · 💻 cs.AI

Towards Energy Impact on AI-Powered 6G IoT Networks: Centralized vs. Decentralized

Anjie Qiu , Donglin Wang , Sanket Partani , Andreas Weinand , Hans D. Schotten This is my paper

Pith reviewed 2026-05-10 02:25 UTC · model grok-4.3

classification 💻 cs.AI

keywords 6G IoTdistributed learningenergy efficiencypredictive maintenancecentralized learningrailway sensors

0 comments

The pith

Distributed machine learning cuts electricity use by up to 70% in 6G IoT networks while holding predictive accuracy near 90%.

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

The paper compares centralized and decentralized machine learning for Internet of Things networks in the 6G era, focusing on how each architecture handles the energy costs of training models and moving data. It first builds separate energy models for transmission and computation in each case, then applies both to real sensor data from a German railway testbed used for predictive maintenance. The analysis shows that spreading the work across devices keeps accuracy competitive at roughly 90 percent yet lowers total electricity draw by as much as 70 percent, largely by cutting long-distance data transfers. This result matters because energy limits are a primary barrier to scaling AI-driven IoT systems for infrastructure monitoring and similar tasks.

Core claim

Comparative analysis of distributed versus centralized learning architectures on the railway sensor dataset reveals that distributed models maintain competitive predictive accuracy of approximately 90 percent while reducing overall electricity consumption by up to 70 percent, chiefly by lowering transmission-related energy costs.

What carries the argument

The energy consumption model that quantifies separate transmission and training costs for centralized versus decentralized learning, tested on sensor data from the German railway infrastructure for predictive maintenance.

Load-bearing premise

The modeled energy figures for data movement and computation accurately represent real costs in the railway testbed and the reported 70 percent savings extend to other IoT deployments and data sets.

What would settle it

Side-by-side metering of actual electricity consumption for the same predictive maintenance task run once in centralized form and once in distributed form on comparable hardware and network links.

Figures

Figures reproduced from arXiv: 2604.19377 by Andreas Weinand, Anjie Qiu, Donglin Wang, Hans D. Schotten, Sanket Partani.

**Figure 3.** Figure 3: Comparison of prediction performance between CL [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

**Figure 4.** Figure 4: Comparison of train speed prediction with CNN model [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

**Figure 5.** Figure 5: Comparison of energy consumption for data transmis [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗

read the original abstract

The emergence of sixth-generation (6G) technologies has introduced new challenges and opportunities for machine learning (ML) applications in Internet of Things (IoT) networks, particularly concerning energy efficiency. As model training and data transmission contribute significantly to energy consumption, optimizing these processes has become critical for sustainable system design. This study first conduct analysis on the energy consumption model for both centralized and decentralized architecture and then presents a testbed deployed within the German railway infrastructure, leveraging sensor data for ML-based predictive maintenance. A comparative analysis of distributed versus Centralized Learning (CL) architectures reveals that distributed models maintain competitive predictive accuracy (~90%) while reducing overall electricity consumption by up to 70%. These findings underscore the potential of distributed ML to improve energy efficiency in real-world IoT deployments, particularly by mitigating transmission-related energy costs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper runs a real railway IoT testbed comparison and reports decentralized learning holds ~90% accuracy with up to 70% lower energy use, but the energy model behind that number has no hardware validation.

read the letter

The main point is straightforward: they took the standard centralized versus decentralized learning comparison, put it on sensor data from a German railway infrastructure testbed for predictive maintenance, and measured that the distributed version keeps accuracy near 90 percent while cutting electricity consumption by as much as 70 percent. The testbed deployment itself is the clearest addition here. Working with actual 6G IoT constraints and real transmission costs in a live setting gives the numbers more weight than another simulation study would have. That part is useful for anyone thinking about sustainable sensor networks in transportation or similar domains. The energy model is where it gets thin. The abstract says they analyzed consumption for both architectures, yet it gives no equations, no parameter values for transmission or training costs, and no side-by-side check against measured power draw on the actual nodes or server. Without those, the 70 percent figure rests on whatever assumptions they made about packet sizes, duty cycles, or idle power, and those assumptions could shift the result. The stress-test note on missing hardware-level validation lines up with what is shown. This work is aimed at practitioners and researchers who need concrete examples of distributed ML in deployed IoT rather than new theory. Someone building large-scale sensor systems might pick up the testbed approach or the accuracy-energy tradeoff as a reference point. It is not a foundational result, but the real-world setup makes it worth a look. I would send it to peer review. The testbed is a genuine plus and the core claim is testable; the authors just need to add the model details, parameter values, and at least one direct measurement comparison so referees can judge whether the savings hold up.

Referee Report

1 major / 1 minor

Summary. The manuscript develops an energy consumption model for centralized and decentralized machine learning in 6G IoT networks and evaluates it using a testbed deployed in the German railway infrastructure for sensor-based predictive maintenance. It claims that distributed learning architectures achieve approximately 90% predictive accuracy while reducing overall electricity consumption by up to 70% compared to centralized approaches, primarily by lowering transmission-related energy costs.

Significance. If the energy model is validated and the results generalize, this work would provide valuable empirical evidence on the energy benefits of decentralized ML in real-world IoT deployments, particularly for 6G networks. The use of an actual railway testbed strengthens the practical relevance, offering insights into mitigating energy consumption in AI-powered IoT systems.

major comments (1)

[Energy consumption model analysis] The central claim of up to 70% electricity reduction relies on the energy consumption model for centralized vs. decentralized architectures. However, the manuscript does not provide the explicit equations, parameter values (such as packet sizes, duty cycles, or transmission costs), or any validation against hardware-level power measurements from the IoT nodes and central server in the testbed. This omission makes it impossible to assess whether the 70% figure accurately reflects real transmission and training costs under the deployed conditions or is sensitive to unstated modeling choices.

minor comments (1)

[Abstract] The phrasing 'This study first conduct analysis' is grammatically incorrect and should be corrected to 'This study first conducts an analysis' or similar for clarity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the detailed and constructive review of our manuscript. The feedback on the energy consumption model is particularly helpful for improving the clarity and reproducibility of our work. We address the major comment point by point below.

read point-by-point responses

Referee: The central claim of up to 70% electricity reduction relies on the energy consumption model for centralized vs. decentralized architectures. However, the manuscript does not provide the explicit equations, parameter values (such as packet sizes, duty cycles, or transmission costs), or any validation against hardware-level power measurements from the IoT nodes and central server in the testbed. This omission makes it impossible to assess whether the 70% figure accurately reflects real transmission and training costs under the deployed conditions or is sensitive to unstated modeling choices.

Authors: We agree that the energy model requires more explicit detail to support the 70% reduction claim. The original manuscript presents an analysis of energy consumption for centralized and decentralized architectures but does not include the full set of equations or parameters. In the revised manuscript, we will add the complete mathematical formulation (e.g., transmission energy E_tx = P_tx * duration, computation energy per training round, and aggregate system energy), along with all parameter values used such as packet sizes, duty cycles, transmission power, and per-operation costs drawn from the testbed hardware. We will also add a dedicated subsection describing how the model was cross-checked against available power data from the IoT nodes and server during the railway deployment. These additions will allow readers to evaluate the 70% figure and its sensitivity to modeling choices. revision: yes

Circularity Check

0 steps flagged

No circularity: central claims derive from empirical testbed comparison

full rationale

The paper conducts an analysis of an energy consumption model for centralized versus decentralized architectures and then reports results from a real German railway testbed deployment using sensor data for ML-based predictive maintenance. The key finding (~90% accuracy with up to 70% lower electricity use for distributed models) is presented as the outcome of this comparative experiment rather than any derivation that reduces by construction to fitted parameters, self-definitions, or self-citation chains. No equations or steps in the provided abstract or description exhibit the enumerated circular patterns; the work remains self-contained against the external benchmark of the deployed testbed.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based solely on the abstract, no explicit free parameters, ad-hoc axioms, or invented entities are identifiable. The energy consumption model is referenced but not detailed, so standard domain assumptions about ML training and wireless transmission costs are presumed without further specification.

pith-pipeline@v0.9.0 · 5449 in / 1118 out tokens · 40968 ms · 2026-05-10T02:25:43.938151+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

[1]

Energy-efficient architecture for optimized iot data transmission from edge to cloud,

M. Zeba, M. Mashira, M. Mim, M. Islam, PhD, M. Hassan, and M. Has- san, “Energy-efficient architecture for optimized iot data transmission from edge to cloud,” 03 2024

work page 2024
[2]

Energy efficient distributed learning in integrated fog-cloud computing enabled iot networks,

M. S. Al-Abiad, M. Z. Hassan, and M. J. Hossain, “Energy efficient distributed learning in integrated fog-cloud computing enabled iot networks,” in2022 IEEE International Conference on Communications Workshops (ICC Workshops), 2022, pp. 872–877

work page 2022
[3]

Power consumption reduction for iot devices thanks to edge-ai: Application to human activity recognition,

A. C. Muhoza, E. Bergeret, C. Brdys, and F. Gary, “Power consumption reduction for iot devices thanks to edge-ai: Application to human activity recognition,” vol. 24, 09 2023

work page 2023
[4]

Energy-efficient edge com- puting architectures for ai workloads: A comparative analysis in cloud- driven environments,

A. Anand, S. Goel, and G. Singh Panesar, “Energy-efficient edge com- puting architectures for ai workloads: A comparative analysis in cloud- driven environments,” in2024 International Conference on Emerging Innovations and Advanced Computing (INNOCOMP), 2024, pp. 614– 620

work page 2024
[5]

Energy-efficient distributed machine learning in cloud fog networks,

M. M. Alenazi, B. A. Yosuf, S. H. Mohamed, T. E. El-Gorashi, and J. M. H. Elmirghani, “Energy-efficient distributed machine learning in cloud fog networks,” in2021 IEEE 7th World Forum on Internet of Things (WF-IoT), 2021, pp. 935–941

work page 2021
[6]

The energy cost of artificial intelligence of things lifecycle,

S.-K. Chou, J. Hribar, M. Mohor ˇciˇc, and C. Fortuna, “The energy cost of artificial intelligence of things lifecycle,” 2024. [Online]. Available: https://arxiv.org/pdf/2408.00540

work page arXiv 2024
[7]

Energy-efficient edge learning via joint data deepening-and-prefetching,

S. Kook, W.-Y . Shin, S.-L. Kim, and S.-W. Ko, “Energy-efficient edge learning via joint data deepening-and-prefetching,” 2024. [Online]. Available: https://arxiv.org/pdf/2402.11925

work page arXiv 2024
[8]

Recent developments in ai and ml for iot: A systematic literature review on lorawan energy efficiency and performance optimization,

M. A. Alkhayyal and A. Mostafa, “Recent developments in ai and ml for iot: A systematic literature review on lorawan energy efficiency and performance optimization,”Sensors, vol. 24, p. 4482, 2024

work page 2024
[9]

Machine learning and artificial intelligence algorithms that minimize power consumption in internet of things gadgets,

B. Shalu and V . R. Sarobin, “Machine learning and artificial intelligence algorithms that minimize power consumption in internet of things gadgets,” pp. 1 – 6, 2023

work page 2023
[10]

An integrated deep learning and edge computing framework for intelligent energy man- agement in iot-based smart cities,

R. Udayakumar, B. Mahesh, R. Sathiyakala, T. Kavitha, A. Choubey, A. Khurramov, L. Alzubaidi, and J. Sravanthi, “An integrated deep learning and edge computing framework for intelligent energy man- agement in iot-based smart cities,” 2023

work page 2023
[11]

Energy-efficient cloud infrastructure for iot device management: A comprehensive analysis of edge-cloud workload dis- tribution strategies,

K. Pothakanoori, “Energy-efficient cloud infrastructure for iot device management: A comprehensive analysis of edge-cloud workload dis- tribution strategies,”International journal of scientific research in computer science, engineering and information technology, vol. 10, pp. 1439 – 1449, 2024

work page 2024
[12]

Green hero: Energy-efficient hierarchical federated learning with client association in industrial iot,

C. Gong, M. Ma, L. Zeng, Y . Yang, and L. Wu, “Green hero: Energy-efficient hierarchical federated learning with client association in industrial iot,”IEEE internet of things magazine, vol. 7, pp. 30 – 36, 2024

work page 2024
[13]

A hybrid model for energy-efficient green internet of things enabled intelligent transportation systems using federated learning,

S. Kaleem, A. Sohail, M. Babar, A. Ahmad, and M. U. Tariq, “A hybrid model for energy-efficient green internet of things enabled intelligent transportation systems using federated learning,”Internet of things, 2023

work page 2023
[14]

Energy-efficient federated learning-based device selection in iot environments,

A. AlZailaa, H. Ran, A. Radwan, and R. L. Aguiar, “Energy-efficient federated learning-based device selection in iot environments,” 2023

work page 2023
[15]

Joint data deepening- and-prefetching for energy-efficient edge learning,

S. Kook, W.-Y . Shin, S.-L. Kim, and S.-W. Ko, “Joint data deepening- and-prefetching for energy-efficient edge learning,”arXiv.org, vol. abs/2211.07146, 2022

work page arXiv 2022
[16]

The tradeoffs of large scale learning

L. Bottou and O. Bousquet, “The tradeoffs of large scale learning.” vol. 20, 01 2007

work page 2007
[17]

Characteristics of low-carbon data centres,

E. Masanet, A. Shehabi, and J. Koomey, “Characteristics of low-carbon data centres,”Nature Climate Change, vol. 3, pp. 627–630, 06 2013

work page 2013
[18]

Cooling systems in data centers: State of art and emerging technologies,

A. Capozzoli and G. Primiceri, “Cooling systems in data centers: State of art and emerging technologies,”Energy Procedia, vol. 83, pp. 484– 493, 12 2015

work page 2015
[19]

On global electricity usage of communication technology: Trends to 2030,

A. Andrae and T. Edler, “On global electricity usage of communication technology: Trends to 2030,”Challenges, vol. 6, pp. 117–157, 04 2015

work page 2030
[20]

Kairouz, H

P. Kairouz, H. B. McMahan, B. Avent, A. Bellet, M. Bennis, A. N. Bhagoji, K. Bonawit, Z. Charles, G. Cormode, R. Cummings, R. G. L. D’Oliveira, H. Eichner, S. El Rouayheb, D. Evans, J. Gardner, Z. Gar- rett, A. Gascón, B. Ghazi, P. B. Gibbons, M. Gruteser, Z. Harchaoui, C. He, L. He, Z. Huo, B. Hutchinson, J. Hsu, M. Jaggi, T. Ja- vidi, G. Joshi, M. Khoda...

work page 2021
[21]

Communication-efficient decentralized machine learning over hetero- geneous networks,

P. Zhou, Q. Lin, D. Loghin, B. C. Ooi, Y . Wu, and H. Yu, “Communication-efficient decentralized machine learning over hetero- geneous networks,” in2021 IEEE 37th International Conference on Data Engineering (ICDE), 2021, pp. 384–395

work page 2021
[22]

i.mx 8m plus – arm® cortex®-a53, machine learning, vision, multimedia and industrial iot,

NXP, “i.mx 8m plus – arm® cortex®-a53, machine learning, vision, multimedia and industrial iot,” https://www.nxp.com/products, 2025, accessed: May 19, 2025

work page 2025
[23]

Slki – second level ki in weichen,

B. für Verkehr, “Slki – second level ki in weichen,” https://www.bmv.de/SharedDocs/DE/Artikel/DG/InnoNT- Projekte/slki.html, 2025, accessed: May 23, 2025

work page 2025
[24]

Nxp website,

N. Cooperation, “Nxp website,” https://www.nxp.com/, 2025, accessed: May 19, 2025

work page 2025
[25]

Arm cortex proccessors,

Arm, “Arm cortex proccessors,” https://www.arm.com/products/silicon- ip-cpu, 2025, accessed: May 19, 2025

work page 2025
[26]

mlco2/codecarbon: v2.4.1,

B. Courty, V . Schmidt, S. Luccioni, Goyal-Kamal, MarionCoutarel, B. Feld, J. Lecourt, LiamConnell, A. Saboni, Inimaz, supatomic, M. Léval, L. Blanche, A. Cruveiller, ouminasara, F. Zhao, A. Joshi, A. Bogroff, H. de Lavoreille, N. Laskaris, E. Abati, D. Blank, Z. Wang, A. Catovic, M. Alencon, Michał St˛ echły, C. Bauer, L. O. N. de Araújo, JPW, and Minerv...

work page
[27]

2024.mlco2/codecarbon: v2.4.1

[Online]. Available: https://doi.org/10.5281/zenodo.11171501

work page doi:10.5281/zenodo.11171501

[1] [1]

Energy-efficient architecture for optimized iot data transmission from edge to cloud,

M. Zeba, M. Mashira, M. Mim, M. Islam, PhD, M. Hassan, and M. Has- san, “Energy-efficient architecture for optimized iot data transmission from edge to cloud,” 03 2024

work page 2024

[2] [2]

Energy efficient distributed learning in integrated fog-cloud computing enabled iot networks,

M. S. Al-Abiad, M. Z. Hassan, and M. J. Hossain, “Energy efficient distributed learning in integrated fog-cloud computing enabled iot networks,” in2022 IEEE International Conference on Communications Workshops (ICC Workshops), 2022, pp. 872–877

work page 2022

[3] [3]

Power consumption reduction for iot devices thanks to edge-ai: Application to human activity recognition,

A. C. Muhoza, E. Bergeret, C. Brdys, and F. Gary, “Power consumption reduction for iot devices thanks to edge-ai: Application to human activity recognition,” vol. 24, 09 2023

work page 2023

[4] [4]

Energy-efficient edge com- puting architectures for ai workloads: A comparative analysis in cloud- driven environments,

A. Anand, S. Goel, and G. Singh Panesar, “Energy-efficient edge com- puting architectures for ai workloads: A comparative analysis in cloud- driven environments,” in2024 International Conference on Emerging Innovations and Advanced Computing (INNOCOMP), 2024, pp. 614– 620

work page 2024

[5] [5]

Energy-efficient distributed machine learning in cloud fog networks,

M. M. Alenazi, B. A. Yosuf, S. H. Mohamed, T. E. El-Gorashi, and J. M. H. Elmirghani, “Energy-efficient distributed machine learning in cloud fog networks,” in2021 IEEE 7th World Forum on Internet of Things (WF-IoT), 2021, pp. 935–941

work page 2021

[6] [6]

The energy cost of artificial intelligence of things lifecycle,

S.-K. Chou, J. Hribar, M. Mohor ˇciˇc, and C. Fortuna, “The energy cost of artificial intelligence of things lifecycle,” 2024. [Online]. Available: https://arxiv.org/pdf/2408.00540

work page arXiv 2024

[7] [7]

Energy-efficient edge learning via joint data deepening-and-prefetching,

S. Kook, W.-Y . Shin, S.-L. Kim, and S.-W. Ko, “Energy-efficient edge learning via joint data deepening-and-prefetching,” 2024. [Online]. Available: https://arxiv.org/pdf/2402.11925

work page arXiv 2024

[8] [8]

Recent developments in ai and ml for iot: A systematic literature review on lorawan energy efficiency and performance optimization,

M. A. Alkhayyal and A. Mostafa, “Recent developments in ai and ml for iot: A systematic literature review on lorawan energy efficiency and performance optimization,”Sensors, vol. 24, p. 4482, 2024

work page 2024

[9] [9]

Machine learning and artificial intelligence algorithms that minimize power consumption in internet of things gadgets,

B. Shalu and V . R. Sarobin, “Machine learning and artificial intelligence algorithms that minimize power consumption in internet of things gadgets,” pp. 1 – 6, 2023

work page 2023

[10] [10]

An integrated deep learning and edge computing framework for intelligent energy man- agement in iot-based smart cities,

R. Udayakumar, B. Mahesh, R. Sathiyakala, T. Kavitha, A. Choubey, A. Khurramov, L. Alzubaidi, and J. Sravanthi, “An integrated deep learning and edge computing framework for intelligent energy man- agement in iot-based smart cities,” 2023

work page 2023

[11] [11]

Energy-efficient cloud infrastructure for iot device management: A comprehensive analysis of edge-cloud workload dis- tribution strategies,

K. Pothakanoori, “Energy-efficient cloud infrastructure for iot device management: A comprehensive analysis of edge-cloud workload dis- tribution strategies,”International journal of scientific research in computer science, engineering and information technology, vol. 10, pp. 1439 – 1449, 2024

work page 2024

[12] [12]

Green hero: Energy-efficient hierarchical federated learning with client association in industrial iot,

C. Gong, M. Ma, L. Zeng, Y . Yang, and L. Wu, “Green hero: Energy-efficient hierarchical federated learning with client association in industrial iot,”IEEE internet of things magazine, vol. 7, pp. 30 – 36, 2024

work page 2024

[13] [13]

A hybrid model for energy-efficient green internet of things enabled intelligent transportation systems using federated learning,

S. Kaleem, A. Sohail, M. Babar, A. Ahmad, and M. U. Tariq, “A hybrid model for energy-efficient green internet of things enabled intelligent transportation systems using federated learning,”Internet of things, 2023

work page 2023

[14] [14]

Energy-efficient federated learning-based device selection in iot environments,

A. AlZailaa, H. Ran, A. Radwan, and R. L. Aguiar, “Energy-efficient federated learning-based device selection in iot environments,” 2023

work page 2023

[15] [15]

Joint data deepening- and-prefetching for energy-efficient edge learning,

S. Kook, W.-Y . Shin, S.-L. Kim, and S.-W. Ko, “Joint data deepening- and-prefetching for energy-efficient edge learning,”arXiv.org, vol. abs/2211.07146, 2022

work page arXiv 2022

[16] [16]

The tradeoffs of large scale learning

L. Bottou and O. Bousquet, “The tradeoffs of large scale learning.” vol. 20, 01 2007

work page 2007

[17] [17]

Characteristics of low-carbon data centres,

E. Masanet, A. Shehabi, and J. Koomey, “Characteristics of low-carbon data centres,”Nature Climate Change, vol. 3, pp. 627–630, 06 2013

work page 2013

[18] [18]

Cooling systems in data centers: State of art and emerging technologies,

A. Capozzoli and G. Primiceri, “Cooling systems in data centers: State of art and emerging technologies,”Energy Procedia, vol. 83, pp. 484– 493, 12 2015

work page 2015

[19] [19]

On global electricity usage of communication technology: Trends to 2030,

A. Andrae and T. Edler, “On global electricity usage of communication technology: Trends to 2030,”Challenges, vol. 6, pp. 117–157, 04 2015

work page 2030

[20] [20]

Kairouz, H

P. Kairouz, H. B. McMahan, B. Avent, A. Bellet, M. Bennis, A. N. Bhagoji, K. Bonawit, Z. Charles, G. Cormode, R. Cummings, R. G. L. D’Oliveira, H. Eichner, S. El Rouayheb, D. Evans, J. Gardner, Z. Gar- rett, A. Gascón, B. Ghazi, P. B. Gibbons, M. Gruteser, Z. Harchaoui, C. He, L. He, Z. Huo, B. Hutchinson, J. Hsu, M. Jaggi, T. Ja- vidi, G. Joshi, M. Khoda...

work page 2021

[21] [21]

Communication-efficient decentralized machine learning over hetero- geneous networks,

P. Zhou, Q. Lin, D. Loghin, B. C. Ooi, Y . Wu, and H. Yu, “Communication-efficient decentralized machine learning over hetero- geneous networks,” in2021 IEEE 37th International Conference on Data Engineering (ICDE), 2021, pp. 384–395

work page 2021

[22] [22]

i.mx 8m plus – arm® cortex®-a53, machine learning, vision, multimedia and industrial iot,

NXP, “i.mx 8m plus – arm® cortex®-a53, machine learning, vision, multimedia and industrial iot,” https://www.nxp.com/products, 2025, accessed: May 19, 2025

work page 2025

[23] [23]

Slki – second level ki in weichen,

B. für Verkehr, “Slki – second level ki in weichen,” https://www.bmv.de/SharedDocs/DE/Artikel/DG/InnoNT- Projekte/slki.html, 2025, accessed: May 23, 2025

work page 2025

[24] [24]

Nxp website,

N. Cooperation, “Nxp website,” https://www.nxp.com/, 2025, accessed: May 19, 2025

work page 2025

[25] [25]

Arm cortex proccessors,

Arm, “Arm cortex proccessors,” https://www.arm.com/products/silicon- ip-cpu, 2025, accessed: May 19, 2025

work page 2025

[26] [26]

mlco2/codecarbon: v2.4.1,

B. Courty, V . Schmidt, S. Luccioni, Goyal-Kamal, MarionCoutarel, B. Feld, J. Lecourt, LiamConnell, A. Saboni, Inimaz, supatomic, M. Léval, L. Blanche, A. Cruveiller, ouminasara, F. Zhao, A. Joshi, A. Bogroff, H. de Lavoreille, N. Laskaris, E. Abati, D. Blank, Z. Wang, A. Catovic, M. Alencon, Michał St˛ echły, C. Bauer, L. O. N. de Araújo, JPW, and Minerv...

work page

[27] [27]

2024.mlco2/codecarbon: v2.4.1

[Online]. Available: https://doi.org/10.5281/zenodo.11171501

work page doi:10.5281/zenodo.11171501