pith. sign in

arxiv: 2512.06493 · v2 · submitted 2025-12-06 · 💻 cs.NI

Programmable and GPU-Accelerated Edge Inference for Real-Time ISAC on NVIDIA Aerial Testbed

Davide Villa , Mauro Belgiovine , Nicholas Hedberg , Michele Polese , Chris Dick , Tommaso Melodia This is my paper

Pith reviewed 2026-05-17 00:27 UTC · model grok-4.3

classification 💻 cs.NI
keywords ISACedge AIGPU acceleration5G localizationOpen RANreal-time inferenceDMRS channel estimatesNVIDIA Aerial
0
0 comments X

The pith

A programmable framework adds real-time GPU AI for sensing to standard 5G RAN hardware with 150 microsecond overhead.

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

The paper establishes that AI applications can run directly on edge RAN infrastructure by processing PHY and MAC signals from existing 5G deployments. It shows this through an open-source framework built on the Open RAN dApp model that connects to a GPU-accelerated gNB on the NVIDIA Aerial Testbed. The framework supports multiple inference backends and adds minimal latency while preserving communication performance. A demonstration called cuSense uses uplink DMRS channel estimates and a neural network to localize a moving person indoors, reaching 77 cm mean error without extra hardware or signal changes. This approach matters because it enables integrated sensing and communication services on commodity cellular equipment.

Core claim

The authors present a framework that interfaces custom AI logic with a GPU-accelerated gNB based on the NVIDIA Aerial Testbed, feeding PHY/MAC data through an Open RAN dApp architecture with 150 us overhead and support for several AI backends. They demonstrate it with cuSense, which removes static multipath from uplink DMRS channel estimates and runs a neural network to infer user position. In a 3GPP-compliant 5G NR indoor deployment, cuSense delivers 77 cm mean localization error, with 75 percent of predictions within 1 meter, using only standard communication signals and no dedicated sensing hardware or RAN modifications.

What carries the argument

The dApp framework that connects the GPU-accelerated gNB to custom AI inference engines, supplying PHY/MAC data for real-time processing.

If this is right

  • AI sensing applications can be deployed on existing 5G infrastructure without hardware additions or signal modifications.
  • Multiple GPU platforms can host the framework while maintaining low latency for primary communication.
  • Open-source release allows other researchers to build additional ISAC dApps on the same edge RAN base.
  • Real-time processing of channel estimates supports sensing tasks that piggyback on normal uplink transmissions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same signal-processing pipeline could support additional tasks such as activity recognition or multi-target tracking if the neural network is adapted.
  • Deployment across commercial Open RAN sites would test whether the 150 us overhead scales under loaded network conditions.
  • Extending the framework to other frequency bands or higher mobility scenarios would reveal limits of the current DMRS-based approach.

Load-bearing premise

The neural network trained on specific indoor scenarios continues to deliver accurate results in other environments and the added inference time stays low enough to leave communication performance unchanged.

What would settle it

Run the cuSense dApp in an outdoor setting with new user paths and measure whether mean localization error remains near 77 cm and 75 percent of estimates stay within 1 meter, while confirming overhead does not exceed 150 us.

Figures

Figures reproduced from arXiv: 2512.06493 by Chris Dick, Davide Villa, Mauro Belgiovine, Michele Polese, Nicholas Hedberg, Tommaso Melodia.

Figure 1
Figure 1. Figure 1: Overview of our work at the intersection of [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: NVIDIA ARC-OTA dApp Integration Architecture. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Data path integration between Aerial L1, [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Support for multiple RAN nodes, E3 Agents, [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: dApp container reference architecture with [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Client latency (Operations 5-8 of Table 1) and [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Overview of the cuSense UL DMRS-based ISAC dApp for person localization. cuSense targets indoor location estimation in a single-cell CBRS deployment using CSI derived from UL PUSCH trans￾missions. In particular, it leverages DeModulation Reference Signal (DMRS) signals transmitted by a UE and exposed by the E3 Agent, in what is known as UL-collaborative ISAC [28] [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 8
Figure 8. Figure 8: The goal is to estimate the 2D position (𝑥𝑡 , 𝑦𝑡) of a person walking in the space through a probability map over the area of interest. The system addresses the following key challenges in UL-CSI-based sensing: (i) extracting real-time channel perturbations from static multipath; (ii) learning spatial mapping from high-dimensional CSI data; and (iii) achieving real-time inference. Finally, cuSense is desig… view at source ↗
Figure 9
Figure 9. Figure 9: cuSense dApp processing pipeline overview. [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Proposed NN architecture of cuSense dApp for CSI-based location estimation. [PITH_FULL_IMAGE:figures/full_fig_p010_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: cuSense experimental environment. ground-truth 2D trajectories of the person walking. Synchro￾nizing CSI data and video frames requires handling both time-reference differences and clock skew between devices. CSI records from ADL use International Atomic Time (TAI) timestamps, while video frames from the camera (an iPhone in our experiments, though any commercial 5G camera could be used) employ standard C… view at source ↗
Figure 12
Figure 12. Figure 12: Temporal-random split strategy with example of contiguous validation/test blocks within training runs and fully unseen runs. cuSense generalization across two independent unseen runs (Unseen-Run 1 and Unseen-Run 2) collected separately from all training, testing, or validation data. We measure localiza￾tion performance using standard Root Mean Squared Error (RMSE), median, and success rate within differen… view at source ↗
Figure 13
Figure 13. Figure 13: cuSense localization accuracy on test set (a, [PITH_FULL_IMAGE:figures/full_fig_p011_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: Trajectory tracking comparison of an unseen run segment for the X and Y axis [PITH_FULL_IMAGE:figures/full_fig_p011_14.png] view at source ↗
read the original abstract

The transition of cellular networks to (i) software-based systems on commodity hardware and (ii) platforms for services beyond connectivity introduces critical system-level challenges. As sensing emerges as a key feature toward 6G standardization, supporting Integrated Sensing and Communication (ISAC) with limited bandwidth and piggybacking on communication signals, while maintaining high reliability and performance, remains a fundamental challenge. In this paper, we provide two key contributions. First, we present a programmable, open-source framework for processing PHY/MAC signals through real-time, GPU-accelerated Artificial Intelligence (AI) applications on the edge Radio Access Network (RAN) infrastructure. Building on the Open RAN dApp architecture, the framework interfaces with a GPU-accelerated gNB based on NVIDIA Aerial Testbed (ATB), feeding PHY/MAC data to custom AI logic with a framework overhead of 150 us, multiple inference engines, and support for several AI backends. We evaluate the framework on multiple GPU platforms with and without hardware-level GPU isolation. Second, we demonstrate the framework capabilities through cuSense, an indoor localization dApp that consumes uplink DMRS channel estimates, removes static multipath components, and runs a neural network to infer the position of a moving person. Evaluated on a 3GPP-compliant 5G NR deployment, cuSense achieves a mean localization error of 77 cm, with 75% of predictions falling within 1 meter, without dedicated sensing hardware or modifications to the RAN stack or signals. The framework is released as open source, providing a reference design for future AI-native RANs and ISAC applications.

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.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a programmable open-source framework for real-time GPU-accelerated AI processing of PHY/MAC signals on the NVIDIA Aerial Testbed gNB, with a reported framework overhead of 150 μs and support for multiple inference engines and backends. It demonstrates the framework via cuSense, an ISAC dApp that uses uplink DMRS channel estimates, removes static multipath, and applies a neural network for indoor localization of a moving person, achieving a mean error of 77 cm with 75% of predictions within 1 m on a 3GPP-compliant 5G NR deployment without dedicated sensing hardware or RAN stack modifications.

Significance. If the integration claims hold, the work supplies a concrete, open-source reference design for AI-native edge RAN and ISAC applications on commodity hardware. The specific measured metrics (150 μs overhead, localization accuracy from a real 5G NR testbed) and evaluation across GPU platforms with hardware isolation provide a practical foundation for future 6G sensing-communication systems. The open-source release strengthens reproducibility and utility as a starting point for additional dApps.

major comments (2)
  1. Abstract: The central claim that the framework 'adds only 150 us overhead without degrading primary communication performance' and requires 'no modifications to the RAN stack or signals' is load-bearing for the engineering contribution, yet the manuscript provides no comparative measurements (e.g., throughput, BLER, or scheduling latency) on the same Aerial Testbed configuration with the dApp active versus baseline.
  2. Evaluation (cuSense results): Details on the neural network training dataset size, diversity, cross-validation, and handling of potential indoor environmental biases are limited, which directly affects confidence in the reported 77 cm mean error and generalization beyond the specific test scenarios.
minor comments (2)
  1. The abstract and introduction could more explicitly state the number of inference engines supported and the exact GPU isolation mechanisms tested.
  2. Figure and table captions should include additional context on the 3GPP compliance parameters and testbed configuration to aid readers.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive review and the opportunity to clarify aspects of our work. We address each major comment below and indicate the changes planned for the revised manuscript.

read point-by-point responses

  1. Referee: Abstract: The central claim that the framework 'adds only 150 us overhead without degrading primary communication performance' and requires 'no modifications to the RAN stack or signals' is load-bearing for the engineering contribution, yet the manuscript provides no comparative measurements (e.g., throughput, BLER, or scheduling latency) on the same Aerial Testbed configuration with the dApp active versus baseline.

    Authors: We agree that direct comparative measurements would strengthen the claim. The manuscript reports the measured 150 μs framework overhead and confirms that cuSense operates on existing uplink DMRS without RAN stack or signal modifications, but it does not include side-by-side throughput, BLER, or scheduling latency results with the dApp enabled versus baseline on the same testbed. In the revision we will add these measurements, collected under identical Aerial Testbed configurations, to demonstrate that primary communication performance remains unaffected. revision: yes

  2. Referee: Evaluation (cuSense results): Details on the neural network training dataset size, diversity, cross-validation, and handling of potential indoor environmental biases are limited, which directly affects confidence in the reported 77 cm mean error and generalization beyond the specific test scenarios.

    Authors: We acknowledge that the current manuscript provides only high-level information on the neural network training. To increase confidence in the reported localization accuracy and to better support claims of generalization, the revised version will expand the evaluation section with explicit details on dataset size, scenario diversity (including multiple indoor layouts and movement patterns), the cross-validation procedure employed, and steps taken to mitigate environmental biases such as multi-day data collection and controlled variation of room conditions. revision: yes

Circularity Check

0 steps flagged

No circularity; claims rest on direct testbed measurements

full rationale

The manuscript describes an implemented open-source framework integrated with the NVIDIA Aerial Testbed gNB and reports concrete performance figures (150 μs overhead, 77 cm mean localization error) obtained from experimental runs on external hardware. No equations, fitted parameters, or self-citations are invoked to derive these quantities; they are presented as outcomes of direct measurement rather than reductions to prior inputs. The derivation chain is therefore self-contained against external benchmarks and does not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The contribution is primarily systems implementation and experimental demonstration rather than new theoretical constructs; relies on standard 5G NR signal models and neural network assumptions.

axioms (2)
  • domain assumption Uplink DMRS channel estimates contain sufficient information for position inference after static multipath removal.
    Invoked in the cuSense description as the basis for the neural network input.
  • domain assumption GPU-accelerated inference can be inserted into the RAN pipeline with bounded latency on the Aerial Testbed.
    Central to the framework overhead claim of 150 us.

pith-pipeline@v0.9.0 · 5607 in / 1406 out tokens · 41629 ms · 2026-05-17T00:27:43.578515+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ARCHES: Adaptive Real-Time Switching of AI Models for the RAN

    cs.NI 2026-04 conditional novelty 6.0

    ARCHES dynamically switches between AI-based and conventional experts in RAN PHY pipelines at sub-millisecond granularity, delivering 5-7% throughput gains and power savings on uplink channel estimation under varying ...

Reference graph

Works this paper leans on

44 extracted references · 44 canonical work pages · cited by 1 Pith paper

  1. [1]

    2024.Study on Integrated Sensing and Communication

    3GPP. 2024.Study on Integrated Sensing and Communication. Technical Specification 22.137. 3rd Generation Partnership Project (3GPP). https: //www.3gpp.org/ftp/Specs/archive/22_series/22.137/

    work page 2024

  2. [2]

    Fadel Adib, Zach Kabelac, Dina Katabi, and Robert C. Miller. 2014. 3D Tracking via Body Radio Reflections. In11th USENIX Symposium on Networked Systems Design and Implementation (NSDI 14). USENIX Association, Seattle, WA, 317–329

    work page 2014

  3. [3]

    Fadel Adib and Dina Katabi. 2013. See through walls with WiFi! SIGCOMM Comput. Commun. Rev.43, 4 (August 2013), 75–86

    work page 2013

  4. [4]

    AI-RAN Alliance. 2024. AI-RAN Alliance: Vision and Mission White Paper. https://ai-ran.org/wp-content/uploads/2024/12/AI- RAN_Alliance_Whitepaper.pdf

    work page 2024

  5. [5]

    Chowdhury, Stefano Basagni, and Tommaso Melodia

    Leonardo Bonati, Pedram Johari, Michele Polese, Salvatore D’Oro, Subhramoy Mohanti, Miead Tehrani-Moayyed, Davide Villa, Shweta Shrivastava, Chinenye Tassie, Kurt Yoder, Ajeet Bagga, Paresh Patel, Ventz Petkov, Michael Seltser, Francesco Restuccia, Abhimanyu Gosain, Kaushik R. Chowdhury, Stefano Basagni, and Tommaso Melodia. 2021. Colosseum: Large-Scale W...

    work page 2021

  6. [6]

    Nada Bouknana, Mohsen Ahadi, Florian Kaltenberger, and Robert Schmidt. 2025. An O-RAN Framework for AI/ML-Based Localization with OpenAirInterface and FlexRIC. arXiv:2511.19233 [cs.NI]

    work page arXiv 2025

  7. [7]

    Stoller, Jacobus Van der Merwe, Kirk Webb, and Gary Wong

    Joe Breen, Andrew Buffmire, Jonathon Duerig, Kevin Dutt, Eric Eide, Mike Hibler, David Johnson, Sneha Kumar Kasera, Earl Lewis, Dustin Maas, Alex Orange, Neal Patwari, Daniel Reading, Robert Ricci, David Schurig, Leigh B. Stoller, Jacobus Van der Merwe, Kirk Webb, and Gary Wong. 2020. POWDER: Platform for Open Wireless Data-driven Experimental Research. I...

    work page 2020

  8. [8]

    Tingjun Chen, Prasanthi Maddala, Panagiotis Skrimponis, Jakub Kolodziejski, Abhishek Adhikari, Hang Hu, Zhihui Gao, Arun Paidi- marri, Alberto Valdes-Garcia, Myung Lee, Sundeep Rangan, Gil Zuss- man, and Ivan Seskar. 2023. Open-access millimeter-wave software- defined radios in the PAWR COSMOS testbed: Design, deployment, and experimentation.Computer Netw...

    work page 2023

  9. [9]

    Francisco Crespo, Javier Villegas, Carlos Baena, Eduardo Baena, Sergio Fortes, and Raquel Barco. 2025. Energy-Aware CPU Orchestration in O- RAN: A dApp-Driven Lightweight Approach. arXiv:2508.00629 [cs.NI]

    work page arXiv 2025

  10. [10]

    Yuanhao Cui, Xiaojun Jing, and Junsheng Mu. 2022. Integrated Sensing and Communications Via 5G NR Waveform: Performance Analysis. In ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). 8747–8751

    work page 2022

  11. [11]

    Yuanhao Cui, Fan Liu, Xiaojun Jing, and Junsheng Mu. 2021. Integrat- ing Sensing and Communications for Ubiquitous IoT: Applications, Trends, and Challenges.IEEE Network35, 5 (2021), 158–167

    work page 2021

  12. [12]

    Hulede, Oscar Venegas, Jonathan Ashdown, and Amitav Mukherjee

    Manu Dwivedi, Ian Ellis L. Hulede, Oscar Venegas, Jonathan Ashdown, and Amitav Mukherjee. 2024. 5G-Based Passive Radar Sensing for Human Activity Recognition Using Deep Learning. In2024 IEEE Radar Conference (RadarConf24). 1–6

    work page 2024

  13. [13]

    Salvatore D’Oro, Michele Polese, Leonardo Bonati, Hai Cheng, and Tommaso Melodia. 2022. dApps: Distributed Applications for Real- Time Inference and Control in O-RAN.IEEE Communications Magazine 60, 11 (Nov. 2022), 52–58

    work page 2022

  14. [14]

    Russell Ford, Hao Chen, Pranav Madadi, Mandar Kulkarni, Xiaochuan Ma, Daoud Burghal, Guanbo Chen, Yeqing Hu, Chance Tarver, Panagi- otis Skrimponis, Vitali Loseu, Yu Zhang, Yan Xin, Yang Li, Jianzhong Zhang, Shubham Khunteta, Yeswanth Guddeti Reddy, Ashok Ku- mar Reddy Chavva, Mahantesh Kothiwale, and Davide Villa. 2025. Sim2Field: End-to-End Development ...

    work page 2025

  15. [15]

    Xenofon Foukas, Bozidar Radunovic, Matthew Balkwill, and Zhihua Lai. 2023. Taking 5G RAN Analytics and Control to a New Level. In Proceedings of ACM MobiCom ’23(Madrid, Spain). ACM, New York, NY, USA

    work page 2023

  16. [16]

    Marco Giordani, Michele Polese, Marco Mezzavilla, Sundeep Rangan, and Michele Zorzi. 2020. Toward 6G Networks: Use Cases and Tech- nologies.IEEE Communications Magazine58, 3 (2020), 55–61

    work page 2020

  17. [17]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep Residual Learning for Image Recognition. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

    work page 2016

  18. [18]

    Peiwen Huang, Fan Liu, Fuwang Dong, and Zhenkun Wang. 2025. Uplink Collaborative Sensing with OFDM Communication Signals. In ICC 2025 - IEEE International Conference on Communications. 1–6

    work page 2025

  19. [19]

    Wei Jiang, Bin Han, Mohammad Asif Habibi, and Hans Dieter Schotten

    work page

  20. [20]

    The Road Towards 6G: A Comprehensive Survey.IEEE Open Journal of the Communications Society2 (2021), 334–366

    work page 2021

  21. [21]

    Glenn Jocher, Jing Qiu, and Ayush Chaurasia. 2023. Ultralytics YOLO. https://github.com/ultralytics/ultralytics

    work page 2023

  22. [22]

    Florian Kaltenberger, Tommaso Melodia, Irfan Ghauri, Michele Polese, Raymond Knopp, Tien Thinh Nguyen, Sakthivel Velumani, Davide Villa, Leonardo Bonati, Robert Schmidt, Sagar Arora, Mikel Irazabal, and Navid Nikaein. 2025. Driving Innovation in 6G Wireless Tech- nologies: The OpenAirInterface Approach.Computer Networks269 (September 2025), 111410

    work page 2025

  23. [23]

    Anupa Kelkar and Chris Dick. 2021. NVIDIA Aerial GPU Hosted AI-on-5G. In2021 IEEE 4th 5G World Forum (5GWF). 64–69

    work page 2021

  24. [24]

    Andrea Lacava, Leonardo Bonati, Niloofar Mohamadi, Rajeev Gan- gula, Florian Kaltenberger, Pedram Johari, Salvatore D’Oro, Francesca Cuomo, Michele Polese, and Tommaso Melodia. 2025. dApps: En- abling Real-Time AI-Based Open RAN Control.Computer Networks 269 (September 2025), 111342

    work page 2025

  25. [25]

    An Liu, Zhe Huang, Min Li, Yubo Wan, Wenrui Li, Tony Xiao Han, Chenchen Liu, Rui Du, Danny Kai Pin Tan, Jianmin Lu, Yuan Shen, Fabiola Colone, and Kevin Chetty. 2022. A Survey on Fundamental Lim- its of Integrated Sensing and Communication.IEEE Communications Surveys & Tutorials24, 2 (February 2022), 994–1034

    work page 2022

  26. [26]

    Eldar, and Stefano Buzzi

    Fan Liu, Yuanhao Cui, Christos Masouros, Jie Xu, Tony Xiao Han, Yonina C. Eldar, and Stefano Buzzi. 2022. Integrated Sensing and Communications: Toward Dual-Functional Wireless Networks for 6G and Beyond.IEEE Journal on Selected Areas in Communications40, 6 (2022), 1728–1767

    work page 2022

  27. [27]

    Ruiqi Liu, Mengnan Jian, Dawei Chen, Xu Lin, Yichao Cheng, Wei Cheng, and Shijun Chen. 2023. Integrated sensing and communication based outdoor multi-target detection, tracking, and localization in practical 5G Networks.Intelligent and Converged Networks4, 3 (2023), 261–272

    work page 2023

  28. [28]

    Shihang Lu, Fan Liu, Yunxin Li, Kecheng Zhang, Hongjia Huang, Jiaqi Zou, Xinyu Li, Yuxiang Dong, Fuwang Dong, Jia Zhu, Yifeng Xiong, Weijie Yuan, Yuanhao Cui, and Lajos Hanzo. 2024. Integrated Sensing and Communications: Recent Advances and Ten Open Challenges. IEEE Internet of Things Journal11, 11 (June 2024), 19094–19120

    work page 2024

  29. [29]

    Petropulu, and Lajos Hanzo

    Kaitao Meng, Christos Masouros, Athina P. Petropulu, and Lajos Hanzo

    work page

  30. [30]

    Cooperative ISAC Networks: Opportunities and Challenges.IEEE Wireless Communications32, 3 (June 2025), 212–219

    work page 2025

  31. [31]

    Neasamoni Santhi, Davide Villa, Michele Polese, and Tommaso Melodia

    N. Neasamoni Santhi, Davide Villa, Michele Polese, and Tommaso Melodia. 2025. InterfO-RAN: Real-Time In-band Cellular Uplink In- terference Detection with GPU-Accelerated dApps.Proc. of ACM International Symposium on Theory, Algorithmic Foundations, and Pro- tocol Design for Mobile Networks and Mobile Computing (MobiHoc) (October 2025). MobiSys ’26, June ...

    work page 2025

  32. [32]

    Northeastern University, NVIDIA, Mavenir, MITRE, and Qualcomm

    work page

  33. [33]

    https://mediastorage.o-ran.org/ngrg-rr/nGRG-RR-2024-10- dApp%20use%20cases%20and%20requirements.pdf

    dApps for Real-Time RAN Control: Use Cases and Require- ment. https://mediastorage.o-ran.org/ngrg-rr/nGRG-RR-2024-10- dApp%20use%20cases%20and%20requirements.pdf

    work page 2024

  34. [34]

    NVIDIA. 2025. Aerial RAN co-lab over-the-air (ARC-OTA). https: //docs.nvidia.com/aerial/aerial-ran-colab-ota/current/index.html

    work page 2025

  35. [35]

    NVIDIA Corporation. 2018. Triton Inference Server: An Optimized Cloud and Edge Inferencing Solution. https://docs.nvidia.com/ deeplearning/triton-inference-server/user-guide/docs/index.html Ac- cessed December 2025

    work page 2018

  36. [36]

    Filippo Olimpieri, Noemi Giustini, Andrea Lacava, Salvatore D’Oro, Tommaso Melodia, and Francesca Cuomo. 2025. LibIQ: Toward Real- Time Spectrum Classification in O-RAN dApps. InProceedings of IEEE Mediterranean Communication and Computer Networking Conference (MedComNet)(Cagliari, Italy)

    work page 2025

  37. [37]

    Michele Polese, Leonardo Bonati, Salvatore D’Oro, Stefano Basagni, and Tommaso Melodia. 2023. Understanding O-RAN: Architecture, Interfaces, Algorithms, Security, and Research Challenges.IEEE Com- munications Surveys & Tutorials25, 2 (Q2 2023), 1376–1411

    work page 2023

  38. [38]

    Yanlin Ruan, Liang Chen, Xin Zhou, Guangyi Guo, and Ruizhi Chen

    work page

  39. [39]

    IEEE Transactions on Instrumentation and Measurement71 (August 2022), 1–15

    Hi-Loc: Hybrid Indoor Localization via Enhanced 5G NR CSI. IEEE Transactions on Instrumentation and Measurement71 (August 2022), 1–15

    work page 2022

  40. [40]

    Upadhyaya, Nishith Tripathi, Joseph Gaeddert, and Jef- frey H

    Pratheek S. Upadhyaya, Nishith Tripathi, Joseph Gaeddert, and Jef- frey H. Reed. 2023. Open AI Cellular (OAIC): An Open Source 5G O-RAN Testbed for Design and Testing of AI-Based RAN Management Algorithms.IEEE Network37, 5 (September 2023), 7–15

    work page 2023

  41. [41]

    Jornet, Tommaso Melodia, Michele Polese, and Dimitrios Koutsonikolas

    Davide Villa, Imran Khan, Florian Kaltenberger, Nicholas Hedberg, Rúben Soares da Silva, Stefano Maxenti, Leonardo Bonati, Anupa Kelkar, Chris Dick, Eduardo Baena, Josep M. Jornet, Tommaso Melodia, Michele Polese, and Dimitrios Koutsonikolas. 2025. X5G: An Open, Programmable, Multi-Vendor, End-to-End, Private 5G O-RAN Testbed With NVIDIA ARC and OpenAirIn...

    work page 2025

  42. [42]

    Fei Wang, Sanping Zhou, Stanislav Panev, Jinsong Han, and Dong Huang. 2019. Person-in-WiFi: Fine-Grained Person Perception Us- ing WiFi. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    work page 2019

  43. [43]

    Zhiqing Wei, Fengyun Li, Haotian Liu, Xu Chen, Huici Wu, Kaifeng Han, and Zhiyong Feng. 2024. Multiple Reference Signals Collaborative Sensing for Integrated Sensing and Communication System Towards 5G-A and 6G.IEEE Transactions on Vehicular Technology73, 10 (2024)

    work page 2024

  44. [44]

    Zhenkun Zhang, Hong Ren, Cunhua Pan, Sheng Hong, Dongming Wang, Jiangzhou Wang, and Xiaohu You. 2025. Target Localization in Cooperative ISAC Systems: A Scheme Based on 5G NR OFDM Signals. IEEE Transactions on Communications73, 5 (May 2025), 3562–3578

    work page 2025