{"id":"https://openalex.org/W7131314377","doi":"https://doi.org/10.1109/tmc.2026.3667591","title":"EdgeSD: Efficient Speculative Decoding With Vision-Decoding Disaggregation for MLLM Inference in Edge-Cloud Networks","display_name":"EdgeSD: Efficient Speculative Decoding With Vision-Decoding Disaggregation for MLLM Inference in Edge-Cloud Networks","publication_year":2026,"publication_date":"2026-02-24","ids":{"openalex":"https://openalex.org/W7131314377","doi":"https://doi.org/10.1109/tmc.2026.3667591"},"language":null,"primary_location":{"id":"doi:10.1109/tmc.2026.3667591","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tmc.2026.3667591","pdf_url":null,"source":{"id":"https://openalex.org/S69141925","display_name":"IEEE Transactions on Mobile Computing","issn_l":"1536-1233","issn":["1536-1233","1558-0660","2161-9875"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Mobile Computing","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5002941898","display_name":"Hualong Huang","orcid":"https://orcid.org/0000-0003-2176-3931"},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hualong Huang","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0000-0003-2176-3931","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043425375","display_name":"Wenhan Zhan","orcid":"https://orcid.org/0000-0002-1851-7185"},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenhan Zhan","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0000-0002-1851-7185","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073225822","display_name":"Hancong Duan","orcid":"https://orcid.org/0000-0003-2656-2661"},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hancong Duan","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0000-0002-7721-7422","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126711045","display_name":"Kai Peng","orcid":null},"institutions":[{"id":"https://openalex.org/I119045251","display_name":"Huaqiao University","ror":"https://ror.org/03frdh605","country_code":"CN","type":"education","lineage":["https://openalex.org/I119045251"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kai Peng","raw_affiliation_strings":["College of Engineering, Huaqiao University, Quanzhou, China"],"raw_orcid":"https://orcid.org/0000-0003-4809-2234","affiliations":[{"raw_affiliation_string":"College of Engineering, Huaqiao University, Quanzhou, China","institution_ids":["https://openalex.org/I119045251"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126700585","display_name":"Geyong Min","orcid":null},"institutions":[{"id":"https://openalex.org/I23923803","display_name":"University of Exeter","ror":"https://ror.org/03yghzc09","country_code":"GB","type":"education","lineage":["https://openalex.org/I23923803"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Geyong Min","raw_affiliation_strings":["Department of Computer Science, University of Exeter, Exeter, U.K"],"raw_orcid":"https://orcid.org/0000-0003-1395-7314","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Exeter, Exeter, U.K","institution_ids":["https://openalex.org/I23923803"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126714301","display_name":"Zijia Zhao","orcid":null},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zijia Zhao","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0009-0005-8242-6092","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084729575","display_name":"Zitian Zhao","orcid":"https://orcid.org/0000-0001-8605-0938"},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zitian Zhao","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":"https://orcid.org/0000-0001-8605-0938","affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5023722190","display_name":"YaLan Ye","orcid":null},"institutions":[{"id":"https://openalex.org/I150229711","display_name":"University of Electronic Science and Technology of China","ror":"https://ror.org/04qr3zq92","country_code":"CN","type":"education","lineage":["https://openalex.org/I150229711"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yalan Ye","raw_affiliation_strings":["School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, China","institution_ids":["https://openalex.org/I150229711"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.17955512,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"25","issue":"8","first_page":"11537","last_page":"11553"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.15929999947547913,"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":0.15929999947547913,"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":0.08869999647140503,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.0723000019788742,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"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/security-token","display_name":"Security token","score":0.6345999836921692},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5454000234603882},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5428000092506409},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5188999772071838},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.5135999917984009},{"id":"https://openalex.org/keywords/server","display_name":"Server","score":0.49559998512268066},{"id":"https://openalex.org/keywords/edge-computing","display_name":"Edge computing","score":0.44589999318122864},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.43959999084472656},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.4374000132083893},{"id":"https://openalex.org/keywords/subnetwork","display_name":"Subnetwork","score":0.41999998688697815}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.9196000099182129},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6345999836921692},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.5559999942779541},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5454000234603882},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5428000092506409},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5188999772071838},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5135999917984009},{"id":"https://openalex.org/C93996380","wikidata":"https://www.wikidata.org/wiki/Q44127","display_name":"Server","level":2,"score":0.49559998512268066},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.44589999318122864},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.43959999084472656},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4374000132083893},{"id":"https://openalex.org/C2780186347","wikidata":"https://www.wikidata.org/wiki/Q11414","display_name":"Subnetwork","level":2,"score":0.41999998688697815},{"id":"https://openalex.org/C150202949","wikidata":"https://www.wikidata.org/wiki/Q107602","display_name":"Pointer (user interface)","level":2,"score":0.39079999923706055},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.37439998984336853},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.35440000891685486},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.35339999198913574},{"id":"https://openalex.org/C2781357197","wikidata":"https://www.wikidata.org/wiki/Q5757597","display_name":"High memory","level":2,"score":0.34689998626708984},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.34130001068115234},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.3296000063419342},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.3203999996185303},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.31520000100135803},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.31209999322891235},{"id":"https://openalex.org/C179799912","wikidata":"https://www.wikidata.org/wiki/Q205084","display_name":"Computational complexity theory","level":2,"score":0.30959999561309814},{"id":"https://openalex.org/C163797641","wikidata":"https://www.wikidata.org/wiki/Q2067937","display_name":"Tree structure","level":3,"score":0.30869999527931213},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.3061000108718872},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3046000003814697},{"id":"https://openalex.org/C115067241","wikidata":"https://www.wikidata.org/wiki/Q1639854","display_name":"Token passing","level":3,"score":0.2978000044822693},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.2937999963760376},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.2903999984264374},{"id":"https://openalex.org/C2780601250","wikidata":"https://www.wikidata.org/wiki/Q1863181","display_name":"Myrinet","level":3,"score":0.2838999927043915},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.28220000863075256},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.28049999475479126},{"id":"https://openalex.org/C130795937","wikidata":"https://www.wikidata.org/wiki/Q2561570","display_name":"Remote direct memory access","level":2,"score":0.27950000762939453},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C127964446","wikidata":"https://www.wikidata.org/wiki/Q1092142","display_name":"Computational resource","level":3,"score":0.2660999894142151},{"id":"https://openalex.org/C140745168","wikidata":"https://www.wikidata.org/wiki/Q1210082","display_name":"Tree traversal","level":2,"score":0.26600000262260437},{"id":"https://openalex.org/C55282118","wikidata":"https://www.wikidata.org/wiki/Q252683","display_name":"Snapshot (computer storage)","level":2,"score":0.2621000111103058},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.25699999928474426},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tmc.2026.3667591","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tmc.2026.3667591","pdf_url":null,"source":{"id":"https://openalex.org/S69141925","display_name":"IEEE Transactions on Mobile Computing","issn_l":"1536-1233","issn":["1536-1233","1558-0660","2161-9875"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Mobile Computing","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2568803418","display_name":null,"funder_award_id":"U2333211","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"deployment":[1],"of":[2,54,151],"multimodal":[3,163],"large":[4],"language":[5],"models":[6],"(MLLMs)":[7],"in":[8,88,181],"edge-cloud":[9,43,182],"networks":[10],"faces":[11],"critical":[12],"challenges,":[13],"including":[14],"computational":[15,85],"resource":[16,72],"heterogeneity,":[17],"memory":[18,68],"bottlenecks,":[19],"and":[20,51,74,84,135,165,178],"bandwidth":[21],"constraints.":[22,158],"To":[23],"address":[24],"these":[25],"issues,":[26],"we":[27],"propose":[28],"EdgeSD,":[29],"a":[30,94,142,186],"novel":[31],"framework":[32],"that":[33,173],"accelerates":[34],"MLLM":[35,57,170],"inference":[36],"by":[37],"integrating":[38],"speculative":[39],"decoding":[40,52],"(SD)":[41],"with":[42,168],"collaboration.":[44],"First,":[45],"EdgeSD":[46,92,131,174],"decouples":[47],"the":[48,55,81,149],"vision":[49],"encoding":[50],"processes":[53],"draft":[56,121,188],"across":[58],"heterogeneous":[59],"edge":[60,156],"servers":[61],"(ESs).":[62],"This":[63,146],"disaggregation":[64,118],"architecture":[65],"overcomes":[66],"single-node":[67],"constraints,":[69],"enabling":[70],"optimized":[71],"utilization":[73],"high-resolution":[75],"input":[76],"processing.":[77],"Second,":[78],"to":[79,124,194],"resolve":[80],"communication":[82],"bottleneck":[83],"burden":[86],"inherent":[87],"this":[89,106],"distributed":[90],"architecture,":[91],"integrates":[93],"bandwidth-aware":[95],"dynamic":[96],"image":[97],"token":[98,137],"merging":[99],"(ITM)":[100],"method.":[101],"Unlike":[102],"general":[103],"pruning":[104],"techniques,":[105],"EdgeSD-specific":[107],"ITM":[108],"method":[109],"focuses":[110],"on":[111,128,161],"minimizing":[112],"inter-ES":[113],"transmission":[114],"latency":[115,157],"for":[116],"vision-decoding":[117],"while":[119],"maintaining":[120],"quality.":[122],"Third,":[123],"optimize":[125],"SD":[126],"efficiency":[127],"consumer-grade":[129],"ESs,":[130],"employs":[132],"an":[133],"adaptive":[134],"scalable":[136],"tree":[138],"structure":[139,147],"solved":[140],"using":[141,185],"parallel":[143],"delta-stepping":[144],"algorithm.":[145],"maximizes":[148],"number":[150],"accepted":[152],"tokens":[153],"under":[154],"strict":[155],"Extensive":[159],"experiments":[160],"six":[162],"datasets":[164],"five":[166],"benchmarks":[167],"various":[169],"pairs":[171],"demonstrate":[172],"achieves":[175],"substantial":[176],"acceleration":[177],"throughput":[179],"gains":[180],"collaboration":[183],"scenarios":[184],"lightweight":[187],"MLLM,":[189],"achieving":[190],"3.04-5.12x":[191],"speedup":[192],"compared":[193],"baseline":[195],"methods.":[196]},"counts_by_year":[],"updated_date":"2026-07-07T06:14:10.401900","created_date":"2026-02-25T00:00:00"}
