A Deep Incremental Framework for Multi-Service Multi-Modal Devices in NextG AI-RAN Systems
IEEE Transactions on Mobile Computing · 2026-12 · 25 (12) · 22712-22729
DOI: 10.1109/tmc.2026.3726787 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-12. Verfasst von Mrityunjoy Gain und 7 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 50 Literaturangaben. Gefördert von: Institute of Information & communications Technology Planning & Evaluation, Korea Government, Evolvable Deep Learning Model Generation Platform for Edge Computing.
Abstract (Originalsprache)
The artificial intelligence radio access network (AI-RAN) paradigm integrates AI and ML into RAN operations to enable dynamic optimization, real-time adaptability with unified support for eMBB, uRLLC, mMTC, and emerging AI-native applications. While these services traditionally required dedicated devices, next-generation networks aim to enable a single user equipment (UE) to support multiple services concurrently, critical for applications like the metaverse. In this paper, we propose a deep incremental framework for efficient RAN management, introducing the Multi-Service-Modal UE (MSMU) system, which enables a single UE to handle eMBB and uRLLC services simultaneously. We formulate an optimization problem integrating traffic demand prediction, route optimization, RAN slicing, service identification, and radio resource management under uncertainty. We decompose it into long-term (L-SP) and short-term (S-SP) subproblems, and then propose a reversible instance normalization (ReVIN) enhanced Mamba model for L-SP optimization, which predicts non-stationary eMBB and uRLLC traffic demands and optimizes routes for RAN slicing. For the S-SP, we propose another Mamba model enabling real-time service type identification and resource management, utilizing L-SP predictions. We incorporate continual learning into the S-SP framework to adapt to new service types while preserving prior knowledge. Experimental results demonstrate that the proposed framework outperforms the second-best baseline, achieving a 46.86% reduction in traffic demand prediction error, 76.9% throughput improvement with a 5.1% optimality gap, 99.4%/99.3% accuracy in service identification and route selection, 26.70% PRB and power estimation improvement, while E2E latency reduction by 90%.
Förderung
- Institute of Information & communications Technology Planning & Evaluation
- Korea Government
- Evolvable Deep Learning Model Generation Platform for Edge Computing
- Ministry of Science and ICT
- Institute for Information & Communications Technology Planning & Evaluation
- Information Technology Research Center
- IITP
- Global AI Frontier Lab
Zitieren
APA
Gain, M., Kim, K., Raha, A. D., Adhikary, A., Huh, E., Saad, W., Han, Z., & Hong, C. S. (2026). A Deep Incremental Framework for Multi-Service Multi-Modal Devices in NextG AI-RAN Systems. IEEE Transactions on Mobile Computing, 25(12), 22712-22729. https://doi.org/10.1109/tmc.2026.3726787
Vancouver
Gain M, Kim K, Raha AD, Adhikary A, Huh E, Saad W, et al. A Deep Incremental Framework for Multi-Service Multi-Modal Devices in NextG AI-RAN Systems. IEEE Transactions on Mobile Computing. 2026;25(12):22712-22729. doi:10.1109/tmc.2026.3726787
BibTeX
@article{Gain2026,
author = {Gain, Mrityunjoy and Kim, Kitae and Raha, Avi Deb and Adhikary, Apurba and Huh, Eui-Nam and Saad, Walid and Han, Zhu and Hong, Choong Seon},
title = {A Deep Incremental Framework for Multi-Service Multi-Modal Devices in NextG AI-RAN Systems},
journal = {IEEE Transactions on Mobile Computing},
year = {2026},
volume = {25},
number = {12},
pages = {22712--22729},
doi = {10.1109/tmc.2026.3726787}
}
EUVIMED