Funzin, the AIoT specialist moving forward with joy, brings you our April updates! 😁

Hello, this is Funzin! We are excited to share another collection of Funzin’s latest achievements and news in this month’s newsletter.

[Highlights]

• Funzin’s KWM-World: Unveiling the KWM-World Hierarchy

• KWM-Ocelot Secures GS Certification Grade 1

• The Era of Data Explosion, a New Battlefield Shaped by Satellite AI: An Interview on the Development of KWM-GIL

• FUNZIN's Milestones in the AI Industry

As a Vertical AI specialist, we will continue to lead the future of national security and defense. We ask for your continued interest and support in our journey. If you found this newsletter helpful, please subscribe and share it with your colleagues and friends! 😊

Funzin has unveiled the KWM-World Hierarchy, an innovative AI-powered command and decision support system applicable to all three military branches: the Army, Navy, and Air Force.

The core philosophy behind KWM-World is clear: "Instead of one giant AI controlling the entire battlefield, we provide the right-sized AI right where it's needed." Let's dive into the core of Funzin's customized AI lineup that will lead the future battlefield.

🔍 Core Components of KWM-World

☁️ KWM-KCCS (Theater-Level Decision Support / Cloud-Based) A high-performance, AI-based cloud command and decision system deployed for missions requiring massive and complex data processing, such as counter-fire operations and air defense.

💻 KWM-onP (Battlefield-Adaptive Decision Support / On-Premise) Optimized for field environments with frequent unit movements. Directly integrated into equipment (On-Premise), it supports immediate and flexible operational command tailored to dynamic on-site situations.

🚁 MUM-T (Manned-Unmanned Teaming) & On-Device AI Solutions

• KWM-stA: Modular software integrated into manned aircraft to command and control the MUM-T system.

• AISS & AIsen: Unmanned aerial vehicles (UAVs) and sensors equipped with self-aware, autonomous On-device AI to execute frontline missions.

📡 MCTS (The Battlefield Network Enabler) A core communication solution that ensures all these systems can exchange data seamlessly without interruption, even in the harsh and unstable communication environments typical of battlefields.

From the cloud to the edge, all the way to frontline on-device AI! Funzin's multi-layered AI solutions—optimized for the scale and conditions of any battlefield—are setting a new standard for future smart defense.

Funzin’s AI-powered electromagnetic spectrum reconnaissance system, 'KWM-Ocelot,' has officially received GS (Good Software) Certification Grade 1 after passing rigorous quality verification by the Telecommunications Technology Association (TTA). Here is a summary of Funzin's innovative technology that "sees through" invisible radio waves.

📡 What is KWM-Ocelot?

• 3D Radio Wave Visualization: It visualizes mobile communication and broadband signal data collected by reconnaissance devices, such as drones, by integrating it with 3D spatial information and time-series data.

• Precision Location Tracking: Even in complex and noisy electromagnetic environments, it precisely detects specific signals and accurately identifies the exact source of the transmission.

🎖️ Proven Performance in the Field

• Award-Winning Technology: Following the 'Grand Prize at the Army AI Idea Contest' last May, the system received the 'Minister of National Defense Citation' in July for its contribution to the drone industry, gaining early recognition for its technical excellence.

• Successful Combat Trials: Last October, it was deployed in combat experiments at the Korea Combat Training Center (KCTC) in Inje, Gangwon-do, proving its operational effectiveness on the battlefield. As a result, it is currently drawing significant interest from leading global defense companies.

🚀 The Next Step for KWM-Ocelot With this GS Grade 1 certification as a stepping stone, Funzin continues its R&D to maintain an overwhelming technological lead.

• Applying Physics AI based on Physics-Informed Neural Networks (PINN).

• Developing a 'World Foundation Model' specifically for the defense radio spectrum.

• Ultimate Goal: To evolve the technology beyond simple location tracking, achieving a level where objects can be perfectly identified using only frequency detection data!

The evolution of satellite AI stepping in for the human eye! What exactly is the newly launched 'KWM-GIL (GEOINT Layer)' solution being developed by Funzin? We sat down with Seung-hoon Shin, Head of the Research Center, to find out.

Q1. What is the primary driving force behind the development of the KWM-GIL solution?

A. While the volume of collected satellite data is increasing exponentially, the processing capacity of human analysts has its limits, leading to a growing "Intel Gap." KWM-GIL is a solution designed and developed to utilize AI as an assistive tool to reduce blind spots in intelligence analysis amidst this flood of data. While the value of individual data is crucial, we initiated this development because we determined that fusing multiple datasets generates new and even more valuable intelligence.

Q2. In which areas are you currently advancing information processing through the application of AI?

A. We are currently in the process of advancing three core areas that are fundamentally the most researched right now:

• Object Detection: We are researching the automatic detection of tanks, missile launchers, and naval vessels by introducing the Vision Transformer (ViT) model, with an ultimate goal of achieving 95% accuracy.

• Change Detection: We are testing technologies designed to identify meaningful differences from time-series imagery—specifically, comparing images taken at different timestamps.

• Super-Resolution: We are conducting research to enhance the identifiability of low-resolution images using AI-based image quality enhancement technology.

We are currently focusing our advancements on these core areas. Moving forward, we plan to concentrate even more heavily on research that generates meaningful new intelligence through data fusion.

Q3. How do you plan to handle situations where the collected satellite imagery is of low resolution?

A. We are researching ways to overcome this limitation through 'Super-Resolution' technology. Currently, we are conducting tests using GAN (Generative Adversarial Networks) and Diffusion models to restore 1m-class low-resolution images to a 0.3m-class level. Our focus is on bringing out the clearest possible outlines of targets that are difficult to identify due to weather conditions or resolution issues.

Q4. In 'Change Detection,' how do you distinguish between simple seasonal changes and actual military threats?

A. That is one of the core challenges we are continuously refining our model training to address. We are advancing the system beyond simply comparing image pixel differences to enable 'Semantic Detection.' We are researching ways to minimize false positives by having the AI autonomously distinguish between natural changes—like falling snow or shedding leaves—and actual threats, such as troop movements or encampment construction.

To perform this semantic analysis flawlessly, it is necessary to fuse and interpret a variety of information. We must comprehensively evaluate factors such as the timeline of events, conflicts in specific regions, and rapid environmental changes.