BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:b4c49cd03c85bffef015b4cf7c46a3d9
DTSTAMP:20260607T140441Z
SUMMARY:Exjobbspresentation: Generative AI for Anomaly Diagnosis in 5G NR S
 cheduling Logs
DESCRIPTION:Kontakt: susanna.lonnqvist@eit.lth.se\n\n5G NR base stations pr
 oduce detailed scheduling logs that record per-slot decisions\, channel me
 asurements\, and HARQ feedback at sub-millisecond granularity. Diagnosing 
 anomalies in these logs currently requires hours of manual expert inspecti
 on per file. This thesis presents a system that automates this process end
 -to-end. Raw binary traces are parsed into a relational database\, where s
 tatistical and machine learning methods detect anomalies at both the indiv
 idual-record and temporal levels. A correlation-based module differentiate
 s root causes with similar signatures\, a rule engine maps findings to ver
 ified diagnostic patterns\, and a Large Language Model (LLM) agent synthes
 izes the evidence into causal explanations. A feedback mechanism allows en
 gineers to confirm or correct diagnoses\, building a case store for future
  reference. On 11 logs from a controlled lab environment\, the system clas
 sified all anomaly patterns correctly\, diagnosed two engineer-confirmed f
 ailures\, and surfaced subtle events that conventional alarms missed. Diag
 nosis time dropped from roughly one hour to 30 seconds\, allowing engineer
 s to quickly validate results and focus on resolution rather than investig
 ation.Handledare: Johan Thunberg (EIT)\, Ali Moradian (Ericsson)&nbsp\;Exa
 minator: Michael Lentmaier (EIT)\n\nMer information om händelsen: https:/
 /www.eit.lth.se/evenemang/exjobbspresentation-generative-ai-anomaly-diagno
 sis-5g-nr-scheduling-logs-0
DTSTART;TZID=GMT:20260603T121500
DTEND;TZID=GMT:20260603T130000
LOCATION:E:3139
END:VEVENT
END:VCALENDAR