{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,6]],"date-time":"2026-07-06T20:17:46Z","timestamp":1783369066983,"version":"3.54.6"},"reference-count":53,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001348","name":"Agency for Science, Technology and Research","doi-asserted-by":"publisher","award":["IAF-ICP (H25-MCP3438)"],"award-info":[{"award-number":["IAF-ICP (H25-MCP3438)"]}],"id":[{"id":"10.13039\/501100001348","id-type":"DOI","asserted-by":"publisher"}]},{"name":"IITP-ITRC Grant"},{"name":"Korea government (MSIT)"},{"name":"National Research Foundation, Singapore, and the Infocomm Media Development Authority under its Communications and Connectivity Bridging Funding Initiative"},{"name":"National Research Foundation of Korea (NRF) Grant"},{"name":"Korea government (MSIT)"},{"name":"National Research Foundation of Korea (NRF) Grant"},{"name":"Korea government (MSIT)"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Commun."],"published-print":{"date-parts":[[2026]]},"DOI":"10.1109\/tcomm.2026.3700463","type":"journal-article","created":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T19:55:15Z","timestamp":1780602915000},"page":"10658-10675","source":"Crossref","is-referenced-by-count":0,"title":["Communication-Efficient Hybrid Language Model via Uncertainty-Aware Opportunistic and Compressed Transmission"],"prefix":"10.1109","volume":"74","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-3508-4588","authenticated-orcid":false,"given":"Seungeun","family":"Oh","sequence":"first","affiliation":[{"name":"School of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jinhyuk","family":"Kim","sequence":"additional","affiliation":[{"name":"School of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7623-6552","authenticated-orcid":false,"given":"Jihong","family":"Park","sequence":"additional","affiliation":[{"name":"Information Systems Technology and Design Pillar, Singapore University of Technology and Design, Tampines, Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8592-7408","authenticated-orcid":false,"given":"Seung-Woo","family":"Ko","sequence":"additional","affiliation":[{"name":"Department of Smart Mobility Engineering, Inha University, Incheon, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4895-6680","authenticated-orcid":false,"given":"Jinho","family":"Choi","sequence":"additional","affiliation":[{"name":"School of Electrical and Mechanical Engineering, Adelaide University, Adelaide, SA, Australia"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-4037-3149","authenticated-orcid":false,"given":"Tony Q. S.","family":"Quek","sequence":"additional","affiliation":[{"name":"Singapore University of Technology and Design, Tampines, Singapore"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5228-9913","authenticated-orcid":false,"given":"Seong-Lyun","family":"Kim","sequence":"additional","affiliation":[{"name":"School of Electrical and Electronic Engineering, Yonsei University, Seoul, South Korea"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.52202\/068431-2176"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-023-02448-8"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1038\/s44159-023-00241-5"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/3649506"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3662006.3662067"},{"key":"ref6","first-page":"19274","article-title":"Fast inference from transformers via speculative decoding","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Leviathan"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.acl-demo.16"},{"key":"ref8","article-title":"CITER: Collaborative inference for efficient large language model decoding with token-level routing","author":"Zheng","year":"2025","journal-title":"arXiv:2502.01976"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/JIOT.2025.3618095"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/WCSP68525.2025.1010651"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1016\/j.jiixd.2025.12.008"},{"key":"ref12","first-page":"1","article-title":"DSSD: Efficient edge-device deployment and collaborative inference via distributed split speculative decoding","volume-title":"Proc. ICML Workshop Mach. Learn. Wireless Commun. Netw. (ML4Wireless)","author":"Ning"},{"key":"ref13","article-title":"FlowSpec: Continuous pipelined speculative decoding for efficient distributed LLM inference","author":"Liu","year":"2025","journal-title":"arXiv:2507.02620"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/ICCD65941.2025.00043"},{"key":"ref15","article-title":"Efficient LLM inference over heterogeneous edge networks with speculative decoding","author":"Zhu","year":"2025","journal-title":"arXiv:2510.11331"},{"key":"ref16","article-title":"GPT-4 technical report","volume-title":"arXiv:2303.08774","author":"Achiam","year":"2023"},{"key":"ref17","article-title":"DeepSeek-r1: Incentivizing reasoning capability in LLMs via reinforcement learning","author":"Guo","year":"2025","journal-title":"arXiv:2501.12948"},{"key":"ref18","article-title":"Fast and slow generating: An empirical study on large and small language models collaborative decoding","author":"Zhang","year":"2024","journal-title":"arXiv:2406.12295"},{"key":"ref19","article-title":"TinyLLM: A framework for training and deploying language models at the edge computers","author":"Kandala","year":"2024","journal-title":"arXiv:2412.15304"},{"key":"ref20","article-title":"TinyLlama: An open-source small language model","author":"Zhang","year":"2024","journal-title":"arXiv:2401.02385"},{"issue":"3","key":"ref21","first-page":"3","article-title":"Phi-2: The surprising power of small language models","volume":"1","author":"Javaheripi","year":"2023","journal-title":"Microsoft Res. Blog"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/ICICNIS64247.2024.10823332"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0950"},{"key":"ref24","article-title":"LoftQ: LoRA-fine-tuning-aware quantization for large language models","author":"Li","year":"2023","journal-title":"arXiv:2310.08659"},{"key":"ref25","article-title":"DistillSpec: Improving speculative decoding via knowledge distillation","author":"Zhou","year":"2023","journal-title":"arXiv:2310.08461"},{"key":"ref26","article-title":"Model compression and efficient inference for large language models: A survey","author":"Wang","year":"2024","journal-title":"arXiv:2402.09748"},{"key":"ref27","article-title":"LLMCad: Fast and scalable on-device large language model inference","author":"Xu","year":"2023","journal-title":"arXiv:2309.04255"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/TMC.2024.3513457"},{"key":"ref29","article-title":"Accelerating large language model decoding with speculative sampling","author":"Chen","year":"2023","journal-title":"arXiv:2302.01318"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1145\/3649329.3658473"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/tcsi.2025.3546256"},{"key":"ref32","article-title":"Inferflow: An efficient and highly configurable inference engine for large language models","author":"Shi","year":"2024","journal-title":"arXiv:2401.08294"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1145\/3552326.3587438"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/ICMLCN64995.2025.11140540"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.eacl-long.143"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1542\/peds.88.3.a78a"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.5555\/3045390.3045502"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1145\/219717.219748"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P18-1082"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.2307\/1403865"},{"key":"ref41","article-title":"A short note on an inequality between KL and TV","author":"Canonne","year":"2022","journal-title":"arXiv:2202.07198"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.3390\/e27040373"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.5555\/3104322.3104425"},{"key":"ref44","article-title":"The curious case of neural text degeneration","author":"Holtzman","year":"2019","journal-title":"arXiv preprint arXiv:1904.09751"},{"key":"ref45","article-title":"Turning up the heat: Min-p sampling for creative and coherent LLM outputs","author":"Nguyen","year":"2024","journal-title":"arXiv:2407.01082"},{"key":"ref46","article-title":"Top-n\u03c3: Not all logits are you need","author":"Tang","year":"2024","journal-title":"arXiv:2411.07641"},{"key":"ref47","article-title":"EAGLE-3: Scaling up inference acceleration of large language models via training-time test","author":"Li","year":"2025","journal-title":"arXiv:2503.01840"},{"key":"ref48","volume-title":"Stanford Alpaca: An Instruction-Following LLaMA Model","author":"Taori","year":"2023"},{"key":"ref49","article-title":"Bidirectional encoder representations from transformers (BERT): A sentiment analysis Odyssey","author":"Alaparthi","year":"2020","journal-title":"arXiv:2007.01127"},{"key":"ref50","first-page":"22631","article-title":"The FLAN collection: Designing data and methods for effective instruction tuning","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Longpre"},{"key":"ref51","article-title":"Pre-training distillation for large language models: A design space exploration","author":"Peng","year":"2024","journal-title":"arXiv:2410.16215"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2025.acl-long.885"},{"key":"ref53","article-title":"MCP safety audit: LLMs with the model context protocol allow major security exploits","author":"Radosevich","year":"2025","journal-title":"arXiv:2504.03767"}],"container-title":["IEEE Transactions on Communications"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/26\/11320979\/11551572.pdf?arnumber=11551572","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,6]],"date-time":"2026-07-06T19:45:03Z","timestamp":1783367103000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11551572\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":53,"URL":"https:\/\/doi.org\/10.1109\/tcomm.2026.3700463","relation":{},"ISSN":["0090-6778","1558-0857"],"issn-type":[{"value":"0090-6778","type":"print"},{"value":"1558-0857","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026]]}}}