Intelligent SOC Ecosystem

An Intelligent SOC Ecosystem for Monitoring, Detection, and Response to Cyber Attacks

Explore Project
Real-time Detection
Analytics
Security
AI/ML Engine
SOAR Automation
Threat Intel

Introduction

Watch our AI-generated introduction to the Intelligent SOC Ecosystem

Video Thumbnail
Watch Now

This video was generated by NotebookLM AI. Content may contain inaccuracies and is for reference purposes only.

Project Overview

An Intelligent SOC Ecosystem for Monitoring, Detection, and Response to Cyber Attacks

Project Description

A unified Intelligent SOC Ecosystem where SIEM, SOAR, OpenXDR, Threat Intelligence, and AI/ML/LLM platforms are integrated into a cohesive operational pipeline. The system enables end-to-end security event processing: from log collection and normalization to analysis and automated incident response. SmartXDR serves as the intelligent fusion layer, providing semantic analysis, event enrichment, and risk-based triage prioritization.

SIEM SOAR OpenXDR AI/ML RAG Python Elasticsearch Wazuh MISP IntelOwl n8n SmartXDR DFIR-IRIS CTI Open Source Suricata Zeek IDS/IPS NSM MITRE ATT&CK Threat Hunting Kibana Logstash Filebeat ElastAlert2 Sigma Rules Automation Webhook Incident Response Playbook Docker Docker Compose Reverse Proxy pfSense VMware LLM Embedding Vector DB Smart Automation

Key Features

  • Real-time log monitoring & alerting
  • AI/ML-based log classification
  • CTI enrichment (MISP, IntelOwl)
  • Automated incident response (SOAR)
  • MITRE ATT&CK mapping
  • Human-in-the-loop AI support
  • Customizable dashboards
  • Integration with SIEM
  • Open Source Project
24/7
Monitoring
10+
Platforms
4
Attack Scenarios
6
Processing Phases

Security Framework

MITRE ATT&CK
Elastic Common Schema
SSL/TLS Encryption
CTI Integration

Project Scope

UIT - VNU-HCM Thesis
Academic Year 2025-2026
NT505 - Graduation Thesis
100% Open Source

Technologies

Cutting-edge technologies powering our project

Elastic Stack

Complete search and analytics platform

Elasticsearch Elasticsearch Logstash Logstash Kibana Kibana Beats Beats Elastic Agent Elastic Agent Elastic Fleet Elastic Fleet

Network Security

pfSense pfSense Suricata Suricata IDS/IPS Linux Router Linux Router Zeek Zeek - Network Security Monitor Nginx Nginx

Log Sources

Windows Events Windows Events Linux Syslogs Linux Syslogs WAF Logs WAF Logs Firewall Logs Firewall Logs IDS/IPS Alerts IDS/IPS Alerts Zeek Alerts NSM Alerts

SOAR & Automation

n8n Workflows n8n Workflows DFIR-IRIS DFIR-IRIS ElastAlert2 ElastAlert2 Playbooks Playbooks

Threat Intelligence

MISP MISP IntelOwl IntelOwl IOC Enrichment IOC Enrichment CTI Feeds CTI Feeds

AI/ML Engine

Log Classification Log Classification LLM RAG LLM RAG Anomaly Detection Anomaly Detection SmartXDR SmartXDR

System Architecture

From detection to response - seamlessly integrated

SOC Deployment Architecture

Infrastructure & Sensors

Network Sensors (Suricata IDPS, Zeek NSM), Endpoint Sensors (Wazuh Agents), and Network Infrastructure (pfSense Firewall, Nginx Reverse Proxy, WAF) provide comprehensive visibility.

Central Data

Logs Management (Logstash, Elastic Agents, Fleet) normalizes data to ECS format. Elasticsearch serves as the centralized Data Lake for fast retrieval and long-term Big Data storage.

Application & Analytics

Kibana (SIEM Dashboard), Wazuh Manager (Security Management), Elastic Detection Rules (KQL/EQL), and ElastAlert2 provide real-time analysis and multi-layer alert generation.

Orchestration & Response

n8n (Integration Hub), DFIR-IRIS (Incident Management), MISP & IntelOwl (CTI Platform), and SmartXDR (AI-powered analysis, automated reporting, human-in-the-loop decision support).

SmartXDR

Intelligent SOC Ecosystem Core - The proposed solution designed as the central brain orchestrating the entire SOC operational pipeline

SmartXDR Architecture
SmartXDR Core

SmartXDR Core

The central Control Plane responsible for orchestrating all modules, routing data, managing background tasks, and providing AI/LLM/RAG services. Includes API Gateway (Flask + Gunicorn), Re-ranking RAG with ChromaDB, Redis cache, conversation memory, and RBAC authentication with rate-limiting.

Ingest Pipeline

Ingest Pipeline

Handles data enrichment before storage by receiving ECS-compliant logs from Logstash, Elastic Agents and Fleet. Custom-designed pipeline filters noise, extracts contextual information, and generates the ml_input field as standardized input for SmartXDR ML Classification.

Classification

Classification

Uses Bylastic - a pre-trained DistilBERT model by Byviz Analytics - optimized for Elastic ecosystem. Automatically classifies logs into INFO, WARNING, and ERROR severity levels by analyzing ml_input semantics, providing prediction_probability scores for confidence assessment.

Analysis

Analysis

Deep analysis layer performing semantic analysis on Logs and IoCs. Features alert aggregation with Risk Score (0-100) calculated via multi-factor algorithm (Severity Distribution, Attack Escalation, ML Confidence). Integrates LLM+RAG for CTI-style reports and DFIR-IRIS for IoC enrichment via VirusTotal and MISP.

Reporting

Reporting

Extracts and aggregates classified log events to generate administrative reports from Elasticsearch and ML Classification results. Supports automated periodic reports and detailed triage reports by log source, distributed via Email/Telegram and Webhook with cloud storage upload for auditing compliance.

Assistant

Assistant

AI Security Assistant combining LLM with Advanced RAG using Two-Stage Retrieval: Bi-encoder Semantic Search (ChromaDB) for candidate filtering, and Cross-encoder Re-ranking for precision. Features MMR diversity optimization, Dynamic Thresholding, Token-aware Windowing, and MITRE ATT&CK mapping via Telegram Bot and API/CLI interfaces.

Processing Pipeline

End-to-end security event processing pipeline of the
Intelligent SOC Ecosystem

SOC Ecosystem Pipeline
1

Log Collecting

Elastic Agent and Fleet collect raw logs from Firewall (pfSense), Linux Router, IDPS (Suricata), NSM (Zeek), WAF (ModSecurity). Logs are parsed, normalized to ECS standard and stored for pipeline processing.

2

Log Pre-processing

SmartXDR Ingest Pipeline filters redundant fields, extracts contextual information, and generates the ml_input field - a structured, condensed log representation serving as input for ML classification.

3

Classification and Enrichment

SmartXDR Classification uses Bylastic (DistilBERT-based model by Byviz Analytics) to analyze ml_input semantically, assign severity labels, and generate prediction_probability scores stored in ml.prediction.* fields.

4

Alert Generation and Correlation

Elastic Detection Rules query ML fields for anomaly detection. ElastAlert2 monitors alerts index and forwards matched alerts with IoCs, timestamps, and Kibana links to DFIR-IRIS for case management.

5

Contextual Enrichment and Incident Response

Analysts review alerts in DFIR-IRIS, convert to Incident Cases. IntelOwl performs IoC pre-analysis via MISP and VirusTotal. SmartXDR Analysis uses RAG for intelligent interpretation and risk assessment.

6

Automated Response and Reporting

n8n workflows orchestrate Wazuh Active Response for endpoint isolation, pfSense/Suricata rule updates for network blocking. SmartXDR Reporting generates DOCX reports, uploads to cloud storage, and sends Telegram notifications.

System Demo

Watch our Intelligent SOC Ecosystem in action through real-world attack scenarios

Scenario 1: C2 Outbound Connection

Simulates a user opening a disguised PDF file on a Windows 11 workstation (192.168.85.150). The executable, camouflaged with a PDF icon, launches a background process and establishes a beacon connection to an external Command & Control (C2) server.

Watch Demo

Scenario 2: SQL Injection Attack

Demonstrates SQL Injection exploitation against DVWA web application behind ModSecurity WAF (192.168.85.111). The attacker targets the database server (192.168.85.112) to expose sensitive data, manipulate records, or escalate privileges.

Watch Demo

Scenario 3: NTLM over SMB Brute-force

An attacker from the GUEST network (192.168.95.100) uses automated tools to perform password brute-forcing against SMB service on Windows Server (192.168.85.115), attempting NTLM authentication bypass for lateral movement.

Watch Demo

Scenario 4: Malware Execution

Simulates social engineering attack where a user downloads a fake MediaPlayer installer (.zip) containing malware. Wazuh Agent detects the threat via VirusTotal integration, automatically removes the malicious file, and reports to the central system.

Watch Demo

Scenario 5: SmartXDR ChatOps

Demonstrates ChatOps workflow via Telegram integration. Administrators interact bidirectionally with SmartXDR through the chatbot, requesting AI-powered incident analysis or executing response actions directly from the messaging platform.

Watch Demo

Resources

Access demo videos and project documentation

GitHub Organization

Explore source code and repositories

View GitHub

YouTube Playlist

Watch all demo videos in one playlist

Watch Playlist

Google Drive

Access additional resources and files

Open Drive

Project Team

The minds behind the Intelligent SOC Ecosystem

Lai Quan Thien

Lai Quan Thien

SOC Analyst

Specialized in SIEM, SOC operations, and advanced threat detection. Skilled in building centralized monitoring systems with ELK, Suricata, and security automation solutions.

Ho Diep Huy

Ho Diep Huy

SOC Analyst

Experienced in SOC workflows, SIEM systems, and threat detection techniques. Proficient in deploying ELK Stack, Suricata IDS/IPS, and creating automated security solutions.