nDPI Security Analysis
When nDPI Analysis is enabled at upload time, each conversation is
inspected by nDPI v5 via the ndpiReader command-line tool
(ndpiReader -i <file> -v 2). The backend parses its per-flow output lines
in a single streaming pass and enriches each conversation record.
What nDPI Provides
Data point |
Description |
|---|---|
Application name |
Identified application (e.g. |
Category |
Traffic category (e.g. |
Risk flags |
Per-conversation risk alerts extracted from the
|
JA3C fingerprint |
Client TLS fingerprint (MD5 hash of TLS ClientHello parameters).
Extracted from the |
JA3S fingerprint |
Server TLS fingerprint (MD5 hash of TLS ServerHello parameters).
Extracted from the |
SNI (Server Name Indication) |
Hostname extracted from TLS ClientHello. Extracted from the
|
TLS certificate metadata |
Subject DN, issuer DN, not-before date, and not-after date. Extracted
from the |
Misclassification Corrections
The backend applies post-processing corrections to known nDPI misclassifications before storing results:
UFTP (UDP port 1044) — misclassified as
BitTorrentby nDPI 5.x because binary file-transfer payloads trigger BitTorrent heuristics. Corrected toUFTP.H.225 (TCP port 1720) — misclassified as
Cassandrain some nDPI builds. Corrected toH225.H.323 sub-protocols — nDPI reports all H.323 flows as
H323. Corrected toH225(TCP port 1720, call signaling) orH245(other TCP ports, media control).
Viewing nDPI Results
nDPI data surfaces across multiple views:
Overview tab — summary counts of detected applications, categories, and risk alerts.
Conversations tab — per-row columns for Application, Category, Risk, JA3C, JA3S, and SNI; filterable by any of these fields.
Node Detail Panel — risk badges on individual host nodes in the network graph.
Risk Flags
Risk flags are color-coded badges displayed alongside conversations:
critical— redhigh— orangemedium— amber / yellowlow— purple
Custom signature rules (see Custom Signature Rules) add additional badges to these nDPI-native detections.
Enabling nDPI
By default nDPI analysis runs automatically on every upload. If the
deployment has VITE_ANALYSIS_OPTIONS=true set, an Analysis options
modal appears after file selection where the Protocol & application
classification stage can be unchecked to skip nDPI for that upload.
Either way, nDPI results cannot be added after upload — re-upload the file if you need nDPI data for a capture that was processed without it.
Graceful Degradation
If ndpiReader is not installed or fails, all appName, flowRisks,
and related fields remain empty and analysis continues normally with tshark
data only. A warning is logged but no error is surfaced to the user.