{"id":"https://openalex.org/W7152022486","doi":"https://doi.org/10.1145/3798065.3798069","title":"Pipelining Network and Compute for Real-Time LLM Serving","display_name":"Pipelining Network and Compute for Real-Time LLM Serving","publication_year":2026,"publication_date":"2026-04-04","ids":{"openalex":"https://openalex.org/W7152022486","doi":"https://doi.org/10.1145/3798065.3798069"},"language":null,"primary_location":{"id":"doi:10.1145/3798065.3798069","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3798065.3798069","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 36th Workshop on Network and Operating System Support for Digital Audio and Video","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3798065.3798069","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125639151","display_name":"Aman Sahu","orcid":null},"institutions":[{"id":"https://openalex.org/I60158472","display_name":"Concordia University","ror":"https://ror.org/0420zvk78","country_code":"CA","type":"education","lineage":["https://openalex.org/I60158472"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Aman Sahu","raw_affiliation_strings":["Concordia University, Montreal, Quebec, Canada"],"raw_orcid":"https://orcid.org/0009-0008-4773-7233","affiliations":[{"raw_affiliation_string":"Concordia University, Montreal, Quebec, Canada","institution_ids":["https://openalex.org/I60158472"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5098768017","display_name":"Abdechakour Mechri","orcid":"https://orcid.org/0009-0002-6937-9373"},"institutions":[{"id":"https://openalex.org/I60158472","display_name":"Concordia University","ror":"https://ror.org/0420zvk78","country_code":"CA","type":"education","lineage":["https://openalex.org/I60158472"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Abdechakour Mechri","raw_affiliation_strings":["Concordia University, Montreal, Quebec, Canada"],"raw_orcid":"https://orcid.org/0009-0002-6937-9373","affiliations":[{"raw_affiliation_string":"Concordia University, Montreal, Quebec, Canada","institution_ids":["https://openalex.org/I60158472"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005139138","display_name":"Abdelhak Bentaleb","orcid":"https://orcid.org/0000-0002-5382-6530"},"institutions":[{"id":"https://openalex.org/I60158472","display_name":"Concordia University","ror":"https://ror.org/0420zvk78","country_code":"CA","type":"education","lineage":["https://openalex.org/I60158472"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Abdelhak Bentaleb","raw_affiliation_strings":["Concordia University, Montreal, Quebec, Canada"],"raw_orcid":"https://orcid.org/0000-0002-5382-6530","affiliations":[{"raw_affiliation_string":"Concordia University, Montreal, Quebec, Canada","institution_ids":["https://openalex.org/I60158472"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I60158472"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"22","last_page":"28"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12203","display_name":"Mobile Agent-Based Network Management","score":0.1679999977350235,"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"}},"topics":[{"id":"https://openalex.org/T12203","display_name":"Mobile Agent-Based Network Management","score":0.1679999977350235,"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"}},{"id":"https://openalex.org/T12216","display_name":"Network Time Synchronization Technologies","score":0.11060000211000443,"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"}},{"id":"https://openalex.org/T10080","display_name":"Energy Efficient Wireless Sensor Networks","score":0.0746999979019165,"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/pipeline","display_name":"Pipeline (software)","score":0.33640000224113464},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.3089999854564667},{"id":"https://openalex.org/keywords/minification","display_name":"Minification","score":0.250900000333786},{"id":"https://openalex.org/keywords/power","display_name":"Power (physics)","score":0.24130000174045563}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7261999845504761},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.33640000224113464},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.32899999618530273},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3237000107765198},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.2946000099182129},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2750999927520752},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.250900000333786},{"id":"https://openalex.org/C163258240","wikidata":"https://www.wikidata.org/wiki/Q25342","display_name":"Power (physics)","level":2,"score":0.24130000174045563},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.22630000114440918}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3798065.3798069","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3798065.3798069","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 36th Workshop on Network and Operating System Support for Digital Audio and Video","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3798065.3798069","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3798065.3798069","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 36th Workshop on Network and Operating System Support for Digital Audio and Video","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":9,"referenced_works":["https://openalex.org/W2339106079","https://openalex.org/W2507174453","https://openalex.org/W2960833983","https://openalex.org/W3188452751","https://openalex.org/W4387321091","https://openalex.org/W4402727272","https://openalex.org/W4402727512","https://openalex.org/W4404573917","https://openalex.org/W4416100799"],"related_works":[],"abstract_inverted_index":{"Real-time":[0],"interaction":[1],"with":[2,53,76],"multimodal":[3,130],"Large":[4],"Language":[5],"Models":[6],"(M-LLMs)":[7],"remains":[8],"impractical":[9],"because":[10],"conventional":[11],"serving":[12,42,131],"pipelines":[13],"impose":[14],"a":[15,82],"receive-then-process":[16],"barrier:":[17],"video":[18,51,60],"must":[19],"be":[20],"fully":[21],"uploaded":[22],"and":[23,33,64,120,137],"preprocessed":[24],"before":[25],"inference":[26,138],"can":[27],"begin,":[28],"leading":[29],"to":[30,99],"accelerator":[31],"cold-start":[32],"high":[34],"end-to-end":[35,41,94],"latency.":[36],"We":[37],"present":[38],"RabbitLLM,":[39],"an":[40,100],"architecture":[43],"that":[44,128],"enables":[45],"network-compute":[46],"pipelining":[47],"by":[48,96,118,123],"coupling":[49],"progressive":[50],"transport":[52,136],"transport-triggered":[54],"incremental":[55],"KV-cache":[56],"construction.":[57],"RabbitLLM":[58,91,112],"streams":[59],"frames":[61],"using":[62],"WebRTC":[63],"immediately":[65],"prefixes":[66],"arriving":[67],"visual":[68],"chunks":[69],"into":[70],"the":[71],"model":[72],"state,":[73],"overlapping":[74],"transmission":[75,116],"prefilling":[77],"while":[78,104],"deferring":[79],"decoding":[80],"until":[81],"warm":[83],"cache":[84],"is":[85],"available.":[86],"Across":[87],"realistic":[88],"4G/5G":[89],"traces,":[90],"reduces":[92,114],"interactive":[93],"delay":[95,117],"43.5%":[97],"compared":[98],"upload-first":[101],"TCP":[102],"baseline":[103],"preserving":[105],"output":[106],"quality":[107],"(BERTScore":[108],"F1":[109],">":[110],"0.82).":[111],"also":[113],"per-frame":[115],"95.4%":[119],"bandwidth":[121],"consumption":[122],"99.1%.":[124],"These":[125],"results":[126],"suggest":[127],"low-latency":[129],"requires":[132],"co-design":[133],"between":[134],"streaming":[135],"execution,":[139],"rather":[140],"than":[141],"optimizing":[142],"either":[143],"layer":[144],"in":[145],"isolation.":[146]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-04-09T00:00:00"}
