{"id":"https://openalex.org/W4416955988","doi":"https://doi.org/10.1145/3769102.3770608","title":"SLED: A Speculative LLM Decoding Framework for Efficient Edge Serving","display_name":"SLED: A Speculative LLM Decoding Framework for Efficient Edge Serving","publication_year":2025,"publication_date":"2025-12-03","ids":{"openalex":"https://openalex.org/W4416955988","doi":"https://doi.org/10.1145/3769102.3770608"},"language":"en","primary_location":{"id":"doi:10.1145/3769102.3770608","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3769102.3770608","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3769102.3770608","source":null,"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth ACM/IEEE Symposium on Edge Computing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3769102.3770608","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5069413313","display_name":"Xiangchen Li","orcid":"https://orcid.org/0000-0002-5293-5348"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xiangchen Li","raw_affiliation_strings":["Bradley Department of Electrical and Computer Engineering, Virginia Tech, Blacksburg, VA, USA"],"raw_orcid":"https://orcid.org/0000-0002-5293-5348","affiliations":[{"raw_affiliation_string":"Bradley Department of Electrical and Computer Engineering, Virginia Tech, Blacksburg, VA, USA","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005011713","display_name":"Dimitrios Spatharakis","orcid":"https://orcid.org/0000-0002-0157-7173"},"institutions":[{"id":"https://openalex.org/I174458059","display_name":"National Technical University of Athens","ror":"https://ror.org/03cx6bg69","country_code":"GR","type":"education","lineage":["https://openalex.org/I174458059"]}],"countries":["GR"],"is_corresponding":false,"raw_author_name":"Dimitrios Spatharakis","raw_affiliation_strings":["School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece"],"raw_orcid":"https://orcid.org/0000-0002-0157-7173","affiliations":[{"raw_affiliation_string":"School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece","institution_ids":["https://openalex.org/I174458059"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019689955","display_name":"Saeid Ghafouri","orcid":"https://orcid.org/0000-0003-3799-5702"},"institutions":[{"id":"https://openalex.org/I126231945","display_name":"Queen's University Belfast","ror":"https://ror.org/00hswnk62","country_code":"GB","type":"education","lineage":["https://openalex.org/I126231945"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Saeid Ghafouri","raw_affiliation_strings":["School of Electronics, Electrical Engineering and Computer Science, Queen\u2019s University Belfast, Belfast, United Kingdom","School of Electronics, Electrical Engineering and Computer Science, Queen's University Belfast, Belfast, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0003-3799-5702","affiliations":[{"raw_affiliation_string":"School of Electronics, Electrical Engineering and Computer Science, Queen\u2019s University Belfast, Belfast, United Kingdom","institution_ids":["https://openalex.org/I126231945"]},{"raw_affiliation_string":"School of Electronics, Electrical Engineering and Computer Science, Queen's University Belfast, Belfast, United Kingdom","institution_ids":["https://openalex.org/I126231945"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Jiakun Fan","orcid":"https://orcid.org/0009-0004-4563-8697"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jiakun Fan","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, VA, USA"],"raw_orcid":"https://orcid.org/0009-0004-4563-8697","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, VA, USA","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086758775","display_name":"Hans Vandierendonck","orcid":"https://orcid.org/0000-0001-5868-9259"},"institutions":[{"id":"https://openalex.org/I126231945","display_name":"Queen's University Belfast","ror":"https://ror.org/00hswnk62","country_code":"GB","type":"education","lineage":["https://openalex.org/I126231945"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Hans Vandierendonck","raw_affiliation_strings":["School of Electronics, Electrical Engineering and Computer Science, Queens University Belfast, Belfast, United Kingdom"],"raw_orcid":"https://orcid.org/0000-0001-5868-9259","affiliations":[{"raw_affiliation_string":"School of Electronics, Electrical Engineering and Computer Science, Queens University Belfast, Belfast, United Kingdom","institution_ids":["https://openalex.org/I126231945"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029098934","display_name":"Deepu John","orcid":"https://orcid.org/0000-0002-6139-1100"},"institutions":[{"id":"https://openalex.org/I100930933","display_name":"University College Dublin","ror":"https://ror.org/05m7pjf47","country_code":"IE","type":"education","lineage":["https://openalex.org/I100930933"]}],"countries":["IE"],"is_corresponding":false,"raw_author_name":"Deepu John","raw_affiliation_strings":["School of Electrical and Electronics Engineering, University College Dublin, Dublin, Ireland"],"raw_orcid":"https://orcid.org/0000-0002-6139-1100","affiliations":[{"raw_affiliation_string":"School of Electrical and Electronics Engineering, University College Dublin, Dublin, Ireland","institution_ids":["https://openalex.org/I100930933"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040279482","display_name":"Bo Ji","orcid":"https://orcid.org/0000-0003-0149-7509"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Bo Ji","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, VA, USA"],"raw_orcid":"https://orcid.org/0000-0003-0149-7509","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, VA, USA","institution_ids":["https://openalex.org/I859038795"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005410613","display_name":"Dimitrios S. Nikolopoulos","orcid":"https://orcid.org/0000-0003-0217-8307"},"institutions":[{"id":"https://openalex.org/I859038795","display_name":"Virginia Tech","ror":"https://ror.org/02smfhw86","country_code":"US","type":"education","lineage":["https://openalex.org/I859038795"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dimitrios S. Nikolopoulos","raw_affiliation_strings":["Department of Computer Science, Virginia Tech, Blacksburg, VA, USA"],"raw_orcid":"https://orcid.org/0000-0003-0217-8307","affiliations":[{"raw_affiliation_string":"Department of Computer Science, Virginia Tech, Blacksburg, VA, USA","institution_ids":["https://openalex.org/I859038795"]}]}],"institutions":[],"countries_distinct_count":4,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":4.7447,"has_fulltext":true,"cited_by_count":6,"citation_normalized_percentile":{"value":0.95609326,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":91,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"8"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.28119999170303345,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.28119999170303345,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.1137000024318695,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10273","display_name":"IoT and Edge/Fog Computing","score":0.07959999889135361,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.7128000259399414},{"id":"https://openalex.org/keywords/edge-computing","display_name":"Edge computing","score":0.6123999953269958},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.5947999954223633},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5638999938964844},{"id":"https://openalex.org/keywords/upstream","display_name":"Upstream (networking)","score":0.5595999956130981},{"id":"https://openalex.org/keywords/memory-footprint","display_name":"Memory footprint","score":0.5164999961853027},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.4603999853134155}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7990000247955322},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.7128000259399414},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.6123999953269958},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.5947999954223633},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5638999938964844},{"id":"https://openalex.org/C191172861","wikidata":"https://www.wikidata.org/wiki/Q7899321","display_name":"Upstream (networking)","level":2,"score":0.5595999956130981},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.5164999961853027},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.4603999853134155},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.44110000133514404},{"id":"https://openalex.org/C132943942","wikidata":"https://www.wikidata.org/wiki/Q2562511","display_name":"Footprint","level":2,"score":0.4129999876022339},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.40209999680519104},{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.38830000162124634},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.33889999985694885},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3084999918937683},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.29989999532699585},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.2919999957084656},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.29089999198913574},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.27129998803138733},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.267300009727478},{"id":"https://openalex.org/C7366592","wikidata":"https://www.wikidata.org/wiki/Q1255620","display_name":"Dram","level":2,"score":0.26669999957084656},{"id":"https://openalex.org/C70388272","wikidata":"https://www.wikidata.org/wiki/Q5968558","display_name":"IBM","level":2,"score":0.25459998846054077},{"id":"https://openalex.org/C130795937","wikidata":"https://www.wikidata.org/wiki/Q2561570","display_name":"Remote direct memory access","level":2,"score":0.2513999938964844}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1145/3769102.3770608","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3769102.3770608","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3769102.3770608","source":null,"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth ACM/IEEE Symposium on Edge Computing","raw_type":"proceedings-article"},{"id":"pmh:oai:vtechworks.lib.vt.edu:10919/140714","is_oa":false,"landing_page_url":"https://hdl.handle.net/10919/140714","pdf_url":null,"source":{"id":"https://openalex.org/S4306400248","display_name":"VTechWorks (Virginia Tech)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I859038795","host_organization_name":"Virginia Tech","host_organization_lineage":["https://openalex.org/I859038795"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":null,"raw_type":"Article - Refereed"},{"id":"pmh:oai:pure.qub.ac.uk/portal:openaire/7b09a9e2-96ef-490b-87a9-2139a89aa9a9","is_oa":true,"landing_page_url":"https://pure.qub.ac.uk/en/publications/7b09a9e2-96ef-490b-87a9-2139a89aa9a9","pdf_url":null,"source":{"id":"https://openalex.org/S4306402319","display_name":"Research Portal (Queen's University Belfast)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I126231945","host_organization_name":"Queen's University Belfast","host_organization_lineage":["https://openalex.org/I126231945"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Li, X, Spatharakis, D, Ghafouri, S, Fan, J, Vandierendonck, H, John, D, Ji, B & Nikolopoulos, D S 2025, SLED: A speculative LLM decoding framework for efficient edge serving. in SEC '25: Proceedings of the Tenth ACM/IEEE Symposium on Edge Computing., 26, Association for Computing Machinery. https://doi.org/10.1145/3769102.3770608","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.1145/3769102.3770608","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3769102.3770608","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3769102.3770608","source":null,"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Tenth ACM/IEEE Symposium on Edge Computing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3294571139","display_name":"U.S.-Ireland R&D Partnership:CNS:Small:SWEET: Hardware and Software for Sustainable Wearable Edge Intelligence","funder_award_id":"2315851","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8515454090","display_name":"Collaborative Research: CNS Core: Medium:HardLambda: A new FaaS Abstraction for Cross-Stack Resource Management in Disaggregated Datacenters","funder_award_id":"2106634","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416955988.pdf","grobid_xml":"https://content.openalex.org/works/W4416955988.grobid-xml"},"referenced_works_count":8,"referenced_works":["https://openalex.org/W2012757436","https://openalex.org/W2960833983","https://openalex.org/W4389519530","https://openalex.org/W4405717632","https://openalex.org/W4405934565","https://openalex.org/W4409362928","https://openalex.org/W4415196431","https://openalex.org/W4416557185"],"related_works":[],"abstract_inverted_index":{"The":[0],"growing":[1],"gap":[2],"between":[3],"the":[4,13,120,131,135,139],"increasing":[5],"complexity":[6],"of":[7,17,76,133],"large":[8],"language":[9],"models":[10],"(LLMs)":[11],"and":[12,150,174,195],"limited":[14],"computational":[15],"budgets":[16],"edge":[18,85,101,117,136,176],"devices":[19,102,149],"poses":[20],"a":[21,58,69,79,96,114,123,157],"key":[22],"challenge":[23],"for":[24,46,73,84],"efficient":[25],"on-device":[26],"inference,":[27,43],"despite":[28],"gradual":[29],"improvements":[30],"in":[31],"hardware":[32],"capabilities.":[33],"Existing":[34],"strategies,":[35],"such":[36],"as":[37,68,78],"aggressive":[38],"quantization,":[39],"pruning,":[40],"or":[41,48],"remote":[42],"trade":[44],"accuracy":[45],"efficiency":[47,132],"lead":[49],"to":[50,103],"substantial":[51,185],"cost":[52,197],"burdens.":[53],"This":[54,145],"position":[55],"paper":[56],"introduces":[57],"new":[59],"framework":[60,97],"that":[61,98],"leverages":[62],"speculative":[63],"decoding,":[64],"previously":[65],"viewed":[66],"primarily":[67],"decoding":[70],"acceleration":[71],"technique":[72],"autoregressive":[74],"generation":[75],"LLMs,":[77],"promising":[80],"approach":[81,146],"specifically":[82],"adapted":[83],"computing":[86],"by":[87,155],"orchestrating":[88],"computation":[89],"across":[90,162],"heterogeneous":[91,148],"devices.":[92,144,163],"We":[93],"propose":[94],"SLED,":[95],"allows":[99],"lightweight":[100],"draft":[104,111],"multiple":[105],"candidate":[106],"tokens":[107,121],"locally":[108],"using":[109],"diverse":[110,140],"models,":[112],"while":[113],"single,":[115],"shared":[116],"server":[118,137,177],"verifies":[119],"utilizing":[122],"more":[124],"precise":[125],"target":[126,160],"model.":[127],"To":[128],"further":[129],"increase":[130],"verification,":[134],"batches":[138],"verification":[141],"requests":[142],"from":[143],"supports":[147],"reduces":[151],"server-side":[152],"memory":[153],"footprint":[154],"sharing":[156],"single":[158],"upstream":[159],"model":[161,202],"Our":[164],"initial":[165],"experiments":[166],"with":[167,179],"Jetson":[168],"Orin":[169],"Nano,":[170],"Raspberry":[171],"Pi":[172],"4B/5,":[173],"an":[175],"equipped":[178],"4":[180],"Nvidia":[181],"A100":[182],"GPUs":[183],"indicate":[184],"benefits:":[186],"\u00d72.2":[187],"higher":[188,192],"system":[189,193],"throughput,":[190],"\u00d72.8":[191],"capacity,":[194],"better":[196],"efficiency,":[198],"all":[199],"without":[200],"sacrificing":[201],"accuracy.":[203]},"counts_by_year":[{"year":2026,"cited_by_count":5},{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-03T08:13:44.112507","created_date":"2025-12-03T00:00:00"}