Cyber-Physical Cold-Chain Risk Dataset

This dataset contains high-resolution operational records collected from a cyber-physical cold-chain logistics monitoring environment spanning October 2023 to October 2025. The data represent real-world refrigerated transportation workflows where physical degradation processes, logistics operations, and cyber–network conditions jointly influence shipment integrity and compliance. Measurements are recorded at an 8-minute temporal resolution, enabling fine-grained analysis of short-term dynamics and long-term degradation trends.

The dataset integrates physical sensing data, logistics context indicators, and network and integrity telemetry within a unified time-aligned structure. Each record corresponds to a rolling observation window associated with a specific logistics client and shipment segment. The dataset reflects realistic characteristics observed in operational cold-chain systems, including non-IID client behavior, class imbalance, rare extreme events, and episodic cyber anomalies, which are common challenges in distributed industrial environments.

Data Scope and Structure

Temporal coverage: 2023-10-01 to 2025-10-31

Sampling interval: 8 minutes

Observation window: 64 minutes (rolling)

Clients: Multiple independent logistics clients with heterogeneous data volumes

Shipments: Time-varying shipment identifiers per client and day

Each row represents the state of a shipment monitoring window and includes identification metadata, physical measurements, derived degradation indicators, cyber-network statistics, and outcome labels.

Feature Categories

1. Identification and Time Metadata

Client identifier

Shipment identifier

Window start and end timestamps

Window duration (minutes)

2. Physical State Measurements

Product temperature (°C)

Ambient temperature (°C)

Relative humidity (%)

Cooling system activity ratio

Power supply status

Door open ratio

Vibration intensity (RMS)

3. Thermal Degradation and Dynamics

Temperature change rate

Maximum temperature excursion

Excursion duration

Temperature exposure area (AUC)

Thermal recovery time

Rolling temperature variance

Humidity change rate

4. Logistics and Operational Context

Product category

Target temperature bounds

Remaining travel time

Distance to destination

Stop frequency

Container utilization ratio

Handling event count

5. Network Quality and Communication Performance

Packet loss rate

Uplink latency

Latency jitter

Message interval mean

Message interval variance

Out-of-order packet ratio

6. Data Integrity and Consistency Indicators

Checksum error ratio

Time synchronization offset

Sensor cross-consistency score

Physical feasibility flag

GPS feasibility flag

7. Rolling Window Descriptors

Mean, minimum, maximum, and standard deviation of temperature

Temperature trend slope

Labels and Outcome Variables

The dataset provides multiple labels to support both predictive modeling and diagnostic analysis:

Spoilage risk score: A continuous value in the range [0, 1] representing shipment degradation risk

Attack state: Categorical indicator of operational and cyber anomaly conditions

Compliance status: Discrete compliance outcome (Compliant, Near-Violation, Violation)