CBRN-CADS as a Lattice Entity: Publishing Hazmat Sources in Real Time

Abstract

When the Sarin nerve agent dispersed across five Tokyo subway stations in March 1995, first responders possessed no real-time chemical identification, no common operating picture, and no automated decontamination trigger. Thirteen people died; more than fifty suffered severe, permanent neurological damage. Three decades later, the Salisbury Novichok poisoning exposed the same structural failure at a state-actor scale: detectors operated in isolation, decontamination teams received tasking via voice radio, and the contamination boundary was not definitively mapped for weeks. The lesson across both events is identical — chemical threat data that cannot propagate instantly into a command-wide common operating picture (COP) is operationally worthless. This article argues that publishing CBRN-CADS detections as native Anduril Lattice Entities, using the platform_type: Animal+ classification and Hazmat schema extensions, solves the integration gap that killed response coherence in Tokyo and Salisbury. We detail the entity schema design, the role of AbriIndex as a Lattice-native confidence metric, the TEMPLATE_TRACK lifecycle for plume-dynamic sources, and how BLIS-D decontamination status reporting closes the detect-decontaminate-confirm loop — all within a NATO STANAG-compliant architecture.

---

1. Historical Anchor — Tokyo Subway Sarin Attack, 1995

Inner Landscape

Shoko Asahara and the Aum Shinrikyo operational planners assumed that a nerve agent dispersed across multiple simultaneous transit nodes would overwhelm any coordinated response. That assumption was correct — but not for the reasons they modeled. The attack was tactically unsophisticated: impure Sarin in plastic bags, punctured manually. The responders' failure was not a function of the attack's complexity. It was a function of sensor isolation. Station attendants who noticed victims collapse had no chemical identification capability. Fire department HAZMAT units arriving at Tsukiji Station had no way to share confirmed Sarin identification with units simultaneously responding at Kasumigaseki and Kodemmacho. Each site operated in its own information silo, each commander constructing a local picture that was structurally unable to feed a multi-node COP.

Environmental Read

The Tokyo Metro environment in 1995 represented a closed, high-density, poorly ventilated subterranean network — precisely the conditions that maximize CWA persistence and civilian exposure. Environmental factors that compounded response failure included: the absence of any fixed chemical detection infrastructure at station level; no standardized HAZMAT reporting protocol linking transit authority, metropolitan fire department, and national police; and a political decision architecture that placed responsibility for chemical incident response in a legal gray zone between civilian emergency management and Self-Defense Force CBRN units. The Sarin plume data that did exist — gathered by individual field responders using early photoionization detectors — never aggregated into any usable common picture.

Differential Factor

What distinguished the Tokyo attack from prior CBRN incidents was not agent lethality but information topology. Previous mass-casualty chemical incidents, including industrial accidents at Bhopal (1984), had at least a single fixed source. Tokyo presented five simultaneous dispersal points across a networked infrastructure, each evolving on a different timeline as the impure Sarin continued to off-gas from residual liquid. A multi-node, temporally dynamic hazmat source demands exactly the kind of entity-tracking architecture that did not exist in 1995 and that Anduril Lattice is now capable of providing — if sensor systems are designed to publish into it natively.

Modern Bridge

The Lattice Entity model treats every tracked object — drone, vehicle, vessel, or now chemical hazmat source — as a first-class data citizen with a UUID, a geospatial position, a confidence score, and extensible metadata. Had CBRN-CADS multi-sensor nodes been deployed at Tokyo Metro station level in 1995, each confirmed Sarin detection would have instantiated a TEMPLATE_TRACK entity in a shared COP. Incident commanders would have seen five converging hazmat tracks, their plume polygons overlapping, within the first sixty seconds — enabling unified decontamination tasking rather than five isolated responses. This is not hypothetical architecture. It is the integration that UAM KoreaTech is engineering today.

---

2. Problem Definition — The Sensor-to-COP Integration Gap

The global CBRN defense market is projected to reach $17.8 billion by 2028, growing at a CAGR of 6.4% (MarketsandMarkets, 2024). Yet procurement data consistently shows that the majority of that spending flows into standalone detection hardware rather than into the data integration layer that would make those detectors operationally coherent. A RAND Corporation analysis of multi-domain CBRN exercises found that confirmed agent detections took an average of 8.3 minutes to propagate from field sensor to command-level decision-maker — a gap during which a Sarin or Novichok plume can expand its lethal boundary by several hundred meters downwind.

The Salisbury Novichok response (2018) documented in the UK Home Office's Operation FOLKLORE review illustrates this at the state-actor scale: despite the UK possessing world-class detection laboratories, the absence of a real-time sensor-to-COP integration layer meant that public health authorities, police, and military CBRN units operated on asynchronous information for the first 72 hours. Two civilians were later exposed to the primary Novichok source — a discarded perfume bottle — precisely because contamination boundary data was not live and shared.

NATO STANAG 2352 mandates standardized CBRN warning and reporting formats, but compliance is almost universally achieved through voice radio or batch-mode digital message traffic rather than live entity publication into a shared COP. The gap between what STANAG mandates conceptually and what multi-domain C2 platforms like Lattice can operationalize technically is the precise market space that CBRN-CADS Lattice integration addresses.

---

3. UAM KoreaTech Solution — CBRN-CADS as Native Lattice Entity Publisher

CBRN-CADS (CBRN Chemical Agent Detection System) is a quad-modal detection platform integrating Ion Mobility Spectrometry (IMS), Raman spectroscopy, gamma/neutron detection, and quantitative PCR for biological agents. At the software layer, the system's edge compute module has been architected to publish detections directly into Anduril Lattice using the platform's documented Entity API.

The schema design centers on three choices. First, platform_type: Animal+ — Lattice's classification for non-traditional, non-vehicle sensor sources — correctly categorizes a fixed or mobile CBRN detector node as a source of entity-grade intelligence without forcing it into a vehicle or aircraft schema. Second, the Hazmat extension fields carry: agent_class (TIC, TIM, CWA, BIOHAZ, RAD), concentration_mgm3, detection_modalities[], persistence_rating, and abri_index — UAM KoreaTech's proprietary AbriIndex contamination severity score ranging from 1 (trace, non-actionable) to 5 (confirmed CWA, mass-casualty threshold). Third, each detection publishes as a TEMPLATE_TRACK entity — Lattice's schema for objects that must be tracked as they evolve spatially and temporally — with geospatial polygon updates pushed every 30 seconds as atmospheric dispersion modeling refines the plume boundary.

When AbriIndex reaches level 3 or above, a linked Lattice Mission object automatically tasks available BLIS-D decontamination units. BLIS-D's bleed-air dry decontamination cycle — 90 seconds, waterless, effective across Schedule 1 CWAs and Class A biological agents — completes and posts decon_confirmed: true back to the originating hazmat entity UUID, updating its status from ACTIVE to MITIGATED in the shared COP. The entire detect-identify-decontaminate-confirm loop is auditable within a single Lattice Entity record, satisfying STANAG 2352 reporting without out-of-band voice traffic.

---

4. Strategic Context — Why Korea, Why Lattice, Why Now

Korea occupies a unique strategic position for this architecture. The Republic of Korea faces the world's largest declared chemical weapons program to its north — the DPRK is assessed by the IISS to maintain a stockpile of 2,500–5,000 metric tons of CWAs including Sarin, VX, and mustard agent. This creates a domestic procurement urgency that no NATO European member currently faces at equivalent scale, generating a home market that de-risks the product at a rate no purely export-dependent CBRN firm can match.

Simultaneously, the US-Korea alliance is deepening its multi-domain operational architecture. Combined Forces Command exercises now incorporate Anduril Lattice as a C2 layer for select mission sets, creating an immediate integration demand for Korean-developed sensors that can publish natively into the platform. A CBRN detection system that requires a middleware translation layer — rather than publishing directly as a first-class Lattice Entity — will face increasing procurement disadvantage as interoperability requirements tighten.

The regulatory environment is equally favorable. Korea's Defense Acquisition Program Administration (DAPA) 2025 reform directives explicitly prioritize open-architecture, software-defined CBRN systems over closed, proprietary platforms — a direct structural preference for the CBRN-CADS integration model. European NATO allies, accelerating CBRN investment post-Ukraine, are similarly moving toward platform-agnostic sensor architectures compatible with multi-domain C2 layers, opening export pathways that closed systems cannot access.

---

5. Forward Outlook

UAM KoreaTech's Lattice integration roadmap targets three milestones within the next 18 months. By Q3 2026, the CBRN-CADS Lattice Entity publisher will complete interoperability certification with Anduril's entity API, enabling live hazmat entity publication in a Combined Forces Command-adjacent exercise environment. By Q1 2027, AbriIndex level escalation triggers will be validated against STANAG 2352 reporting thresholds in a NATO CBRN working group tabletop, establishing the compliance bridge for European procurement. By Q3 2027, BLIS-D decontamination status reporting — the decon_confirmed field closing the detect-decontaminate-confirm loop — will be demonstrated in a joint ROK-US CBRN field exercise with live agent simulant release.

Parallel to platform integration, UAM KoreaTech is developing AbriIndex as a licensable scoring standard, positioning it as the CBRN equivalent of the Common Vulnerability Scoring System (CVSS) in cybersecurity — a vendor-neutral severity metric that any Lattice-integrated CBRN sensor could adopt, expanding the addressable market beyond UAM KoreaTech's own hardware.

---

Conclusion

Tokyo's thirteen dead and Salisbury's near-catastrophe share a single systemic cause: chemical threat data that could not propagate in real time into a common operating picture. CBRN-CADS publishing as a native Anduril Lattice Entity — with AbriIndex-scored Hazmat extensions, TEMPLATE_TRACK plume dynamics, and BLIS-D decontamination confirmation closing the loop — is the architectural answer that 1995 responders needed and that 2026 commanders can now demand. The sensor is only as valuable as its integration layer; UAM KoreaTech is building both.