SecureFlow: Safeguarding IoMT Sensor Security AgainstNon-Invasive Optical Attacks

Technical Summary: SecureFlow – Safeguarding IoMT Sensor Security Against Non-Invasive Optical Attacks

Problem Statement
The rapid integration of the Internet of Medical Things (IoMT) into digital healthcare has introduced critical vulnerabilities within Medical Cyber-Physical Systems (MCPS), particularly concerning in-body and on-body sensors. While these systems enable continuous patient monitoring and reduce human error, they are increasingly susceptible to sophisticated offensive technologies. This paper specifically addresses the security risks posed to Gravity-fed IV-drip monitoring systems (Saline Infusion Monitoring Systems). These systems rely heavily on visible light-based drip sensors to count drops and regulate fluid administration. The authors identify a critical gap: while previous research has explored attacks on vision, automotive, and environmental sensors, there is limited exploration of non-invasive physical attacks targeting biomedical embedded subdomains. Specifically, the paper investigates how adversarial lasers can disrupt the Light Dependent Resistor (LDR) components of IV drip sensors, leading to inaccurate drop counting, miscalculation of infusion rates, and potentially life-threatening over- or under-administration of fluids.

Methodology
The research employs a controlled laboratory approach to simulate and analyze non-invasive optical attacks on medical devices.

• Experimental Setup: The authors constructed a custom drip measurement device using a 3D-printed casing, an LED light source, and an LDR sensor. The signal processing chain utilizes an LM358 comparator IC to detect minute voltage deflections caused by falling drops, which are then processed by an ESP32-Wroom-32 microcontroller for real-time counting and monitoring.
• Threat Model: The study assumes an adversary with no direct physical or wireless access (Bluetooth, WiFi, RF) to the device and limited knowledge of the sensor's internal architecture. The attack vector involves two types of adversarial lasers positioned 10 to 50 meters away:
  1. Constant Lasers: Non-flickering sources intended to provide continuous interference, causing persistently incorrect readings.
  2. Pulsating Lasers: Flickering sources manipulated to mimic genuine light sources, deceiving the sensor into reporting controlled, false data.
• Attack Execution: The researchers directed these lasers at the sensor unit to disrupt the LDR's ability to detect falling saline drops. They compared the system's performance under normal conditions versus under attack, measuring parameters such as drip count, drop volume, remaining water volume, and estimated time to empty.
• Mitigation Strategy: To counter these threats, the authors implemented an anomaly detection mechanism within the microcontroller firmware. This system monitors the consistency of drop detection. It flags irregularities, such as prolonged illumination of the LDR followed by a return to normal, as anomalies. If such deviations occur twice or persist for over 5 seconds, the system triggers immediate alerts via buzzer alarms or IoT notifications.

Key Contributions
The paper makes four primary contributions to the field of IoMT security:

1. Target Selection: Identification and selection of pertinent laser attacks specifically targeting saline drip sensors, a widely used yet vulnerable component of medical infrastructure.
2. Vulnerability Analysis: Recording and analyzing the specific alterations, estimations, and deflections caused by laser inputs on sensor readings.
3. Impact Quantification: Estimating the impact of these deflections on patient safety, rated on a scale of 0 to 10. The authors assign a severity rating of 7 to these attacks, indicating a significant life-threatening potential due to risks like fluid overload, pulmonary edema, and electrolyte imbalances.
4. Countermeasure Proposal: Development and integration of a detection and alert system that identifies laser-induced anomalies in real-time, offering a pathway to mitigate identified threats.

Results and Observations
The experimental results highlight a significant discrepancy between the State of the Art (SoA) systems and the proposed system under attack conditions:

• Data Discrepancy: Under normal operation, the system recorded 312 drops with a volume of 15.6 ml. Under laser attack, the system recorded 363 drops with a volume of 18.15 ml.
• Volume Miscalculation: The attack resulted in a calculated difference of 2.55 ml in drop volume and a corresponding error in the remaining water calculation.
• Detection Capability: While the SoA system showed "No Detection" of the attack, the proposed system successfully detected 4 distinct attack attempts.
• Consequences: The laser interference caused the sensors to fail in accurately measuring the volume of saline entering the body. The authors note that in a real-world scenario involving thousands of samples, these discrepancies could escalate, leading to severe clinical complications such as nasal congestion, peripheral edema, internal edema, and abnormal sodium and chloride levels.

Significance and Claims
The paper claims to be a pioneering effort in assessing and quantifying the impact of Electromagnetic Wave (EMW) attacks on intravenous drip infusion sensors. Its significance lies in:

• Novelty: It introduces a new perspective on sensor security by focusing on embedded system security within the biomedical domain, a field previously under-explored compared to automotive or environmental sensors.
• Real-World Relevance: By demonstrating that lasers can disrupt critical life-support monitoring from a distance (up to 50 meters) without physical access, the research underscores the urgent need for enhanced security measures in Saline Infusion Monitoring Systems.
• Proactive Defense: The proposed anomaly detection solution offers a practical, immediate method for hospitals to identify security breaches and take precautionary actions, potentially preventing lethal outcomes.
• Future Direction: The work aims to pave the way for future research into developing more robust medical devices resilient to cyber-physical attacks, advocating for cautionary measures to safeguard current medical infrastructure against emerging optical threats.

The authors explicitly state that their study did not involve human participants, animal subjects, or clinical trials; all experiments were conducted on laboratory-developed hardware prototypes under controlled conditions.