The challenge

Background and context

CAF’s current detection and monitoring capabilities rely on purpose-built, dedicated sensor systems that are costly to procure, slow to deploy, and difficult to scale across the breadth of Canada’s urban landscape. Meanwhile, Canadian cities already operate thousands of cameras, acoustic arrays, air-quality monitors, vibration sensors, and wireless access points that generate rich, continuous data streams—none of which are currently exploited for situational awareness by DND/CAF.

The opportunity is to develop an AI middleware layer that sits on top of these existing feeds, transforming ambient urban data into actionable intelligence without installing new hardware. This is a force-multiplier concept: turning every city into a sensor grid that can detect unauthorized drone activity, unusual vehicular or pedestrian patterns, anomalous radio frequency (RF) emissions, chemical or environmental signatures, and emerging threats in real time and at a fraction of the cost of dedicated systems.

This challenge supports the CAF Digital Campaign Plan, Strong, Secure, Engaged urban defence priorities, domestic operations readiness, critical infrastructure protection, and NORAD (North American Aerospace Defence Command) modernization’s push for persistent, all-domain awareness across the Canadian homeland. The resulting platform becomes a reusable building block for any future scenario requiring rapid urban intelligence—from G7 summit security to Arctic community monitoring to allied interoperability in coalition urban operations.

Canadian privacy law, municipal data-sharing frameworks, bilingual operational requirements, and the unique characteristics of Canadian urban and Arctic environments demand a purpose-built, Canadian-developed solution. This challenge seeks to grow Canadian industrial capacity and intellectual property in ambient urban intelligence—a capability that does not yet exist as a sovereign Canadian product—while ensuring compliance with Canadian values and legal standards from the ground up.

DND/CAF are hosting this challenge to observe advancements in AI technology to resolve this issue.

Examples of application can be described as, but are not limited to:

Urban Counter-Uncrewed Aerial Systems (UAS): Fuse traffic camera video, acoustic sensors, and passive RF analytics to detect, classify, and track unauthorized drones over Canadian cities—without deploying any new hardware.

Critical Infrastructure Protection: Repurpose vibration sensors on bridges and buildings, air-quality monitors near water treatment plants, and surveillance cameras at power substations to create an anomaly-detection mesh.

Environmental Early Warning: Use environmental sensors as secondary indicators for abnormal activity.

DND/CAF will not provide any data (classified, unclassified, operational, synthetic, or representative) for use in training, fine-tuning, or validating AI models. Participation in this challenge assumes that innovators possess sufficient sensor-domain expertise to independently generate or obtain appropriate datasets for model development and testing.

Essential outcomes

Proposed solutions must:

• Build one or more AI-enabled modules capable of ingesting and processing data from multiple (at least two), heterogeneous sources
• Perform real-time analysis to identify anomalous patterns or potential threat indicators within the ingested data streams
• Produce machine readable outputs (e.g., alerts, scores, or flags) corresponding to detected anomalies or threat indicators for further review by an operator
• Be compliant with applicable Canadian legal, privacy, and data protection requirements when processing data derived from civilian or non- defence sources. The offeror is responsible for determining which are applicable

Desired outcomes

Proposed solutions should include capabilities and considerations such as, but not limited to, the following:

• Rapid deployability: A portable, edge-deployable solution that can connect to locally available data feeds and become operational in a new environment
• Adaptive sensor ingestion: The ability to connect to diverse and evolving data feeds using standard protocols, when individual sources degrade or go offline
• Explainable and actionable outputs: Anomaly detections include sufficient context for operators to understand what was detected, which sources contributed, and the confidence level of the assessment
• Scalable architecture: A design that can grow from a localized and that is architecturally compatible with international standards