{"id":"https://openalex.org/W7140285488","doi":"https://doi.org/10.1109/ton.2026.3677371","title":"MoE <sup>2</sup> : Optimizing Collaborative Inference for Edge Large Language Models","display_name":"MoE <sup>2</sup> : Optimizing Collaborative Inference for Edge Large Language Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7140285488","doi":"https://doi.org/10.1109/ton.2026.3677371"},"language":null,"primary_location":{"id":"doi:10.1109/ton.2026.3677371","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ton.2026.3677371","pdf_url":null,"source":{"id":"https://openalex.org/S5407042750","display_name":"IEEE Transactions on Networking","issn_l":"2998-4157","issn":["2998-4157"],"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Networking","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/A5111153528","display_name":"Lyudong Jin","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Lyudong Jin","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":"https://orcid.org/0009-0003-2454-6577","affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130555652","display_name":"Yanning Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Yanning Zhang","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yanhan Li","orcid":"https://orcid.org/0009-0003-6291-8132"},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Yanhan Li","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":"https://orcid.org/0009-0003-6291-8132","affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112860301","display_name":"Shuna Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Shurong Wang","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130549423","display_name":"Howard H. Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Howard H. Yang","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":"https://orcid.org/0000-0002-0256-2416","affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Jian Wu","orcid":"https://orcid.org/0000-0002-3230-6392"},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jian Wu","raw_affiliation_strings":["Zhejiang Key Laboratory of Medical Imaging Artificial Intelligence, Zhejiang University, Hangzhou, China"],"raw_orcid":"https://orcid.org/0000-0002-3230-6392","affiliations":[{"raw_affiliation_string":"Zhejiang Key Laboratory of Medical Imaging Artificial Intelligence, Zhejiang University, Hangzhou, China","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5130588059","display_name":"Meng Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I157725225","display_name":"University of Illinois Urbana-Champaign","ror":"https://ror.org/047426m28","country_code":"US","type":"education","lineage":["https://openalex.org/I157725225"]},{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN","US"],"is_corresponding":false,"raw_author_name":"Meng Zhang","raw_affiliation_strings":["Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China"],"raw_orcid":"https://orcid.org/0000-0002-4893-6946","affiliations":[{"raw_affiliation_string":"Zhejiang University&#x2014;University of Illinois at Urbana-Champaign Institute, Zhejiang University, Haining, China","institution_ids":["https://openalex.org/I157725225","https://openalex.org/I76130692"]}]}],"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":0.33136105,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"34","issue":null,"first_page":"4637","last_page":"4651"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.25429999828338623,"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"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.25429999828338623,"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"}},{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.08250000327825546,"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"}},{"id":"https://openalex.org/T11273","display_name":"Advanced Graph Neural Networks","score":0.07370000332593918,"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/inference","display_name":"Inference","score":0.3970000147819519},{"id":"https://openalex.org/keywords/interpretation","display_name":"Interpretation (philosophy)","score":0.31690001487731934},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.29919999837875366},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.29440000653266907}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6449999809265137},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46549999713897705},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.3970000147819519},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.31690001487731934},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.29919999837875366},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.29440000653266907},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2806999981403351},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2660999894142151},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.26350000500679016},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2574000060558319}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/ton.2026.3677371","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ton.2026.3677371","pdf_url":null,"source":{"id":"https://openalex.org/S5407042750","display_name":"IEEE Transactions on Networking","issn_l":"2998-4157","issn":["2998-4157"],"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Networking","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2823080268","display_name":null,"funder_award_id":"62572433","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G2922365951","display_name":null,"funder_award_id":"U25D9020","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7839574859","display_name":null,"funder_award_id":"62201504","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7987138042","display_name":null,"funder_award_id":"ZFKY20250912001","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":52,"referenced_works":["https://openalex.org/W2025653905","https://openalex.org/W2137983211","https://openalex.org/W2150884987","https://openalex.org/W2594560857","https://openalex.org/W2624989916","https://openalex.org/W2786070938","https://openalex.org/W2797285154","https://openalex.org/W2918714081","https://openalex.org/W2960944858","https://openalex.org/W2964015972","https://openalex.org/W2980856918","https://openalex.org/W3093266389","https://openalex.org/W3193886995","https://openalex.org/W3209221699","https://openalex.org/W3209329320","https://openalex.org/W3214462705","https://openalex.org/W4224315052","https://openalex.org/W4283204791","https://openalex.org/W4285082282","https://openalex.org/W4285115684","https://openalex.org/W4312500615","https://openalex.org/W4313162985","https://openalex.org/W4316371702","https://openalex.org/W4317435061","https://openalex.org/W4319079731","https://openalex.org/W4379115941","https://openalex.org/W4385210979","https://openalex.org/W4385245566","https://openalex.org/W4386385424","https://openalex.org/W4386449970","https://openalex.org/W4388629193","https://openalex.org/W4388644518","https://openalex.org/W4388644561","https://openalex.org/W4388874804","https://openalex.org/W4389077350","https://openalex.org/W4389519226","https://openalex.org/W4389544468","https://openalex.org/W4389664922","https://openalex.org/W4390872508","https://openalex.org/W4391582407","https://openalex.org/W4391590983","https://openalex.org/W4392152396","https://openalex.org/W4396988239","https://openalex.org/W4400447774","https://openalex.org/W4401571420","https://openalex.org/W4404783220","https://openalex.org/W4406207083","https://openalex.org/W4407838869","https://openalex.org/W7133211372","https://openalex.org/W7133215398","https://openalex.org/W7133219248","https://openalex.org/W7133240551"],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1,14],"models":[2,176],"(LLMs)":[3],"have":[4],"demonstrated":[5],"remarkable":[6],"capabilities":[7,20],"across":[8,92,115],"a":[9,44,54,100,140],"wide":[10],"range":[11],"of":[12,21,89,112,123],"natural":[13],"processing":[15],"tasks.":[16],"Exploiting":[17],"the":[18,83,87,121,124,135,170],"heterogeneous":[19],"edge":[22,50,90,151],"LLMs":[23,91],"is":[24],"crucial":[25],"for":[26,49,145,173],"diverse":[27],"emerging":[28],"applications,":[29],"as":[30],"it":[31],"enables":[32,120],"greater":[33],"cost-effectiveness":[34],"and":[35,57,67,86,126,138,158,163,177,190,193,211],"reduced":[36],"latency.":[37],"In":[38],"this":[39,96],"work,":[40],"we":[41,98,107,133],"introduceMixture-of-Edge-Experts":[42],"(MoE2),":[43],"novel":[45],"collaborative":[46],"inference":[47,63],"framework":[48],"LLMs.":[51],"We":[52,149,201],"formulate":[53],"joint":[55],"gating":[56,113],"expert":[58,75,102,116,147],"selection":[59,76,103,127],"problem":[60],"to":[61,82],"optimize":[62],"performance":[64,171],"under":[65,196],"energy":[66,191],"latency":[68],"constraints.":[69,200],"Unlike":[70],"conventional":[71],"MoE":[72],"problems,":[73],"LLM":[74,175],"becomes":[77],"significantly":[78,129],"more":[79],"challenging":[80],"due":[81],"combinatorial":[84],"nature":[85],"heterogeneity":[88],"various":[93,174,197],"attributes.":[94],"To":[95],"end,":[97],"propose":[99],"two-level":[101],"mechanism":[104],"through":[105],"which":[106],"uncover":[108],"an":[109],"optimality-preserving":[110],"property":[111,119],"parameters":[114],"selections.":[117],"This":[118],"decomposition":[122],"training":[125],"processes,":[128],"reducing":[130],"complexity.":[131],"Furthermore,":[132],"leverage":[134],"objective\u2019s":[136],"monotonicity":[137],"design":[139],"discrete":[141],"monotonic":[142],"optimization":[143],"algorithm":[144],"optimal":[146,185],"selection.":[148],"implement":[150],"servers":[152],"with":[153],"NVIDIA":[154,159],"Jetson":[155],"AGX":[156],"Orins":[157],"RTX":[160],"4090":[161],"GPUs,":[162],"perform":[164],"extensive":[165],"experiments.":[166],"Our":[167],"results":[168],"validate":[169],"improvements":[172],"show":[178],"that":[179],"our":[180],"MoE2":[181],"method":[182],"can":[183],"achieve":[184],"trade-offs":[186],"among":[187],"different":[188],"delay":[189],"budgets,":[192],"outperforms":[194],"baselines":[195],"system":[198],"resource":[199],"further":[202],"demonstrate":[203],"its":[204,212],"strong":[205],"robustness":[206],"in":[207,214],"dynamic,":[208],"non-stationary":[209],"environments":[210],"effectiveness":[213],"achieving":[215],"load":[216],"balancing.":[217]},"counts_by_year":[],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2026-03-26T00:00:00"}