3GPP AI packet analysis

A core network signaling analysis system combining a large model and a protocol parsing engine to reconstruct signaling procedures from PCAP files and locate fault root causes automatically.

One-line positioning

A core network signaling analysis system combining a large model and a protocol parsing engine to reconstruct signaling procedures from PCAP files and locate fault root causes automatically.

Product overview

This product addresses fault diagnosis and procedure analysis on the 5G core network side. It combines the semantic understanding and reasoning of large language models (LLMs) with a protocol parsing engine in MCP services to process core network signaling automatically. The system extracts key signaling fields across interfaces from PCAP files and uses a built-in case knowledge base and RAG (retrieval-augmented generation) to load 3GPP TS 23/24/29 specifications dynamically. It analyzes core network and NAS signaling procedures, diagnoses faults automatically, and produces standardized sequence diagrams and structured diagnostic reports.

Supported protocols and interfaces

Category Supported content
Core network 3G/4G/5G core network interfaces including S1-MME, S6A, S11, N2, N4, N26 and SBI, with signaling parsing and field-level decoding; automatic signaling procedure decoding, element identification, anomaly root cause location and fault diagnosis
Optional interfaces VoLTE / VoNR extended analysis: IMS registration / authentication procedures and VoLTE/VoNR call signaling tracing
Protocol extensions RAG-loaded 3GPP specifications and local case libraries, model training and custom field parsing, with on-demand expansion to NB-IoT and IoT scenarios

Core functions

Module Function Specification
Automatic parsing Invoke MCP service tools to parse PCAP files automatically Parallel parsing; individual files ≤10MB
Content extraction Extract key signaling fields automatically Field completeness ≥90%; custom fields supported
Signaling procedure analysis Reconstruct complete procedures automatically Multiple UE scenarios; automatic anomaly marking
Standards comparison Load and compare 3GPP specifications through RAG TS 23/24/29 series coverage ≥95%
Fault diagnosis Root cause analysis and location Fault identification accuracy ≥90% (small-packet testing)
Link and service quality Delay / jitter / packet loss monitoring Fault identification accuracy ≥90% (small-packet testing)
Sequence diagram generation Produce Mermaid sequence diagrams for import into Draw.io and other tools —
Report generation Generate diagnostic reports automatically PDF / Word export, including root causes, recommendations and procedure diagrams

Performance specifications

Item Specification
Processing time per file ≤5 minutes, including parsing, analysis and generation
Concurrent parsing ≥10 parallel tasks
Signaling procedure coverage ≥90% of core network signaling
Automatic standards conformance matching ≥95%
Root cause diagnosis accuracy ≥90%
Automation ≥90%, without manual intervention
Availability 7×24 hours

Deployment and expansion

  • Private and public cloud deployment; private deployment can use a DGX Spark AI workstation
  • Analysis results in Chinese, English, Japanese, Spanish and other languages
  • Custom filtering by protocol, port, IP and procedure characteristics
  • Model fine-tuning and knowledge base updates