{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,16]],"date-time":"2026-01-16T18:08:53Z","timestamp":1768586933557,"version":"3.49.0"},"reference-count":18,"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\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62571477"],"award-info":[{"award-number":["62571477"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Zhejiang Provincial Key Laboratory of Multi-Modal Communication Networks and Intelligent Information Processing, Hangzhou, China","award":["310027"],"award-info":[{"award-number":["310027"]}]},{"DOI":"10.13039\/100006639","name":"Velux Foundation, Denmark, through the Villum Investigator Grant WATER, nr 37793","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100006639","id-type":"DOI","asserted-by":"publisher"}]},{"name":"European Research Council (ERC) under the European Union\u2019s Horizon Europe Programme","award":["101198347"],"award-info":[{"award-number":["101198347"]}]},{"name":"Open Fellowship of the EPSRC","award":["EP\/W024101\/1"],"award-info":[{"award-number":["EP\/W024101\/1"]}]},{"name":"EPSRC project","award":["EP\/X011852\/1"],"award-info":[{"award-number":["EP\/X011852\/1"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Commun. Lett."],"published-print":{"date-parts":[[2026]]},"DOI":"10.1109\/lcomm.2026.3651580","type":"journal-article","created":{"date-parts":[[2026,1,6]],"date-time":"2026-01-06T18:35:43Z","timestamp":1767724543000},"page":"852-856","source":"Crossref","is-referenced-by-count":0,"title":["Quantize-Sample-and-Verify: LLM Acceleration via Adaptive Edge-Cloud Speculative Decoding"],"prefix":"10.1109","volume":"30","author":[{"ORCID":"https:\/\/orcid.org\/0009-0002-2169-311X","authenticated-orcid":false,"given":"Guangyi","family":"Zhang","sequence":"first","affiliation":[{"name":"College of Information Science and Electronic Engineering and Zhejiang Provincial Key Laboratory of Multi-Modal Communication Networks and Intelligent Information Processing, Zhejiang University, Hangzhou, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9418-1700","authenticated-orcid":false,"given":"Yunlong","family":"Cai","sequence":"additional","affiliation":[{"name":"College of Information Science and Electronic Engineering and Zhejiang Provincial Key Laboratory of Multi-Modal Communication Networks and Intelligent Information Processing, Zhejiang University, Hangzhou, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1354-2557","authenticated-orcid":false,"given":"Guanding","family":"Yu","sequence":"additional","affiliation":[{"name":"College of Information Science and Electronic Engineering and Zhejiang Provincial Key Laboratory of Multi-Modal Communication Networks and Intelligent Information Processing, Zhejiang University, Hangzhou, China"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6195-4797","authenticated-orcid":false,"given":"Petar","family":"Popovski","sequence":"additional","affiliation":[{"name":"Connectivity Section, Aalborg University, Aalborg, Denmark"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9898-3209","authenticated-orcid":false,"given":"Osvaldo","family":"Simeone","sequence":"additional","affiliation":[{"name":"Intelligent Networked Systems Institute (INSI), Northeastern University London, London, U.K."}]}],"member":"263","reference":[{"key":"ref1","article-title":"Accelerating large language model decoding with speculative sampling","author":"Chen","year":"2023","journal-title":"arXiv:2302.01318"},{"key":"ref2","article-title":"AI flow at the network edge","author":"Shao","year":"2024","journal-title":"arXiv:2411.12469"},{"key":"ref3","article-title":"Fast and cost-effective speculative edge-cloud decoding with early exits","author":"Venkatesha","year":"2025","journal-title":"arXiv:2505.21594"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/3662006.3662067"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/ICMLCN64995.2025.11140540"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.3390\/s23062931"},{"key":"ref7","first-page":"19274","article-title":"Fast inference from transformers via speculative decoding","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Leviathan"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/LCOMM.2020.3000588"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/DCC.2011.40"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.52202\/079017-1072"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72655-2_4"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.52202\/079017-2181"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/TCOMM.2024.3484939"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v30i1.10295"},{"key":"ref15","article-title":"OPT: Open pre-trained transformer language models","author":"Zhang","year":"2022","journal-title":"arXiv:2205.01068"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/K16-1028"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/MCOM.008.2100382"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/MWC.001.2400355"}],"container-title":["IEEE Communications Letters"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/4234\/11301808\/11333273.pdf?arnumber=11333273","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,16]],"date-time":"2026-01-16T05:34:20Z","timestamp":1768541660000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11333273\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":18,"URL":"https:\/\/doi.org\/10.1109\/lcomm.2026.3651580","relation":{},"ISSN":["1089-7798","1558-2558","2373-7891"],"issn-type":[{"value":"1089-7798","type":"print"},{"value":"1558-2558","type":"electronic"},{"value":"2373-7891","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026]]}}}