Skip to main content

Detection Rules

NetSenX uses the proprietary NDR engine with behavioral analysis and multi-valued logic to detect threats unknown to traditional signatures.

Built-in Detections

CapabilitySeverityWhat it detects
C2 Beacon DetectionCRITICALPeriodic outbound connections to suspicious IPs
Data Exfiltration DetectionCRITICALHigh upload ratio to external IPs, especially after hours
OT/ICS Unauthorized WriteCRITICALWrite commands to OT/ICS protocols (Modbus, DNP3, BACnet)
Port Scan DetectionHIGHSingle IP scanning multiple ports on a target
Lateral Movement DetectionHIGHInternal-to-internal connections on sensitive ports (RDP, SSH, SMB)
Brute Force DetectionHIGHMultiple failed authentication attempts
DNS Tunneling DetectionHIGHDNS queries with abnormal entropy (potential data tunneling)
Shadow AI DetectionMEDIUMUnauthorized connections to AI/ML service endpoints
Anomalous Payload DetectionMEDIUMUnusual packet sizes or byte patterns
New Device DetectionLOWPreviously unseen device on the network

Detection coverage expands with higher-tier plans:

PlanCoverage
FreeCore detections
StarterCore detections
ProfessionalExtended detections
BusinessExtended detections + OT protocol support
EnterpriseFull detection suite + OT protocol support

How the NDR Engine Works

The NetSenX behavioral detection engine:

  1. Builds baseline: Learns normal network behavior over the first 24-48 hours
  2. Extracts features: For each flow, extracts a rich feature set (timing, volume, ports, protocols)
  3. Scores with multi-valued logic: Applies multi-valued logic scoring across multiple detection dimensions
  4. Explains with SHAP: Generates feature attribution for every detection
  5. Accumulates evidence: Uses temporal analysis to correlate related events

Unlike signature-based IDS (Snort, Suricata), NDR detects zero-day attacks and novel threat patterns because it analyzes behavior, not signatures.

Customizing Rules

Go to Dashboard -> Rules to:

  • Enable/disable rules (Admin only)
  • Adjust threshold (0.0 to 1.0) — lower = more sensitive, higher = fewer alerts
  • View rule statistics — how many alerts each rule has generated

Threshold Tuning

Each rule produces a confidence score from 0.0 to 1.0:

Score RangeMeaningAlert Level
0.0 - 0.3No threat detectedNo alert
0.3 - 0.5Low confidenceLogged, no alert
0.5 - 0.7Medium confidenceMEDIUM or LOW alert
0.7 - 0.9High confidenceHIGH alert
0.9 - 1.0Very high confidenceCRITICAL alert

The default threshold is 0.5. Lowering it increases sensitivity (more alerts, potentially more false positives). Raising it reduces noise but may miss threats.

Recommendation: Start with defaults for 1-2 weeks, then tune based on your false positive rate.

OT/ICS Protocol Support

Business and Enterprise plans include support for industrial protocols:

ProtocolPortsDetection
Modbus502Unauthorized function codes (write coils, write registers)
DNP320000Unauthorized control commands
BACnet47808Unauthorized property writes

These rules are critical for manufacturing, energy, and infrastructure companies subject to NIS2.