PaperlayerView exact revision on arXiv

Source-linked AI summary

Practical Zero-Trust for Mission-Critical Robotic Fleets via Hardware Attestation and Packet Timing Watermarking

Ryne Gonzales, Ethan Liesdyanto, Rex Worley, Michael Frederick, Jaewon Kim, Eman Hammad

arXiv:2609.05741v1eess.SYcs.MAcs.SY

TL;DR

Open wireless links and standard ROS 2 middleware expose robotic fleets to attacks that can disrupt real-time control. The paper evaluates a layered defense combining TPM attestation, SIEM telemetry, and an IPD watermark on a live testbed. Volume filters isolate brute DoS floods, while watermark kurtosis detects MITM injections with complete accuracy.

  • Problem

    Open wireless links and unauthenticated ROS 2 middleware expose robotic fleets to stealthy protocol attacks, while benign RF variability complicates anomaly detection.

    Source: §I · pp. 1–2

  • Method

    The paper evaluates a layered framework combining a TPM hardware root of trust, centralized SIEM telemetry, and a non-cryptographic IPD timing watermark on a live ROS 2 testbed.

    Source: §I · p. 2, 2) Packet-Layer Inter-Packet Delay… · p. 3

  • Results

    Volume-based filters isolate brute DoS floods, while IPD-watermark kurtosis detected active MITM injections in 20/20 trials.

    Source: §V · p. 7, 5) Attack Scenario 4: Middleware… · p. 7

  • Takeaways & Limitations

    The testbed evaluation demonstrates layered resilience for hardware-constrained wireless robotic networks without application-layer cryptographic overhead.

    Source: §V · p. 7

Abstract

from arXiv · show

Autonomous unmanned vehicles are vital to tactical missions, mission-critical public-safety operations like search and rescue and disaster response. However, their reliance on open wireless links and standard Robot Operating System (ROS 2) middleware exposes a broad cyber-physical attack surface. A compromise of these systems can disrupt real-time control loops, leading to mission failure or asset loss in high-stakes environments. This paper presents and empirically evaluates a layered, context-aware cybersecurity framework enforcing Zero-Trust principles for a multi-node robotic fleet over Wi-Fi. The framework integrates an active hardware root of trust (TPM 2.0), centralized in-band and out-of-band SIEM telemetry monitoring (ELK Stack and Kismet), and a non-cryptographic Inter-Packet Delay (IPD) timing watermark. Evaluated on a live ROS 2 mobile testbed under multi-layer exploits (OSI Layers 2-5), results demonstrate that while volume-based filters isolate brute denial-of-service floods, tracking the statistical sample kurtosis (K) of the embedded IPD watermark exposes stealthy Man-in-the-Middle command injections with complete detection accuracy without payload overheads.

I. INTRODUCTION

Open wireless links expose robotic fleets to evolving protocol-level attacks, while benign RF variability complicates anomaly detection. The paper presents a live-testbed framework that combines hardware trust, SIEM telemetry, and IPD watermarking to address these risks.

I. INTRODUCTION

  • Wireless links support real-time navigation, telemetry, and coordination in critical UxV deployments, but open shared channels create a broad attack surface.The motivating settings include search and rescue, disaster response, and hazardous-environment inspections.Source: §I · p. 1
  • Benign RF variability makes static anomaly thresholds prone to false alarms or failures to isolate low-power attacks.Multipath fading, noise, and interference vary jitter and packet loss, complicating detection based on latency-related metrics.Source: §I · p. 1
  • Adversaries can use protocol-specific manipulation, including selective dropping, de-authentication flooding, and MITM injection, without raising the ambient RF noise floor.The paper identifies vulnerabilities in unauthenticated 802.11 management frames and default ROS 2 DDS discovery.Source: §I · pp. 1–2
  • The framework combines a hardware root of trust, centralized SIEM monitoring, and IPD watermarking; watermark kurtosis detects stealthy MITM command injections with complete accuracy.Live hardware evaluations examine exploits across OSI Layers 2–5 and their effects on state estimation and trajectory control.Source: §I · p. 2

II. BACKGROUND AND RELATED WORK

Prior defenses tend to focus on either lower-layer interference or application-layer security, while network-timing defenses have often been evaluated offline. This paper combines hardware attestation, SIEM telemetry, and packet-timing verification on an active ROS 2 fleet.

II. BACKGROUND AND RELATED WORK

  • Prior defenses target RF interference at physical and link layers or ROS vulnerabilities at the application layer, leaving protocol-specific network and middleware exploits less covered.The described approaches include spectrum monitoring, adaptive SNR thresholds, security wrappers, access controls, and payload authentication.Source: §II · p. 2
  • The paper combines network telemetry, packet-timing analysis, and hardware attestation, validating the integrated defense on an active ROS 2 fleet under multi-layer exploits.This addresses prior implementations evaluated in offline simulations or static fabrics.Source: §II · p. 2
  • Standard DDS discovery can be unauthenticated, and plaintext ROS 2 velocity and telemetry streams are exposed to interception, replay, and command injection.The passage also notes that DDS-Security plugins can add compute and packet-size overhead on resource-constrained SBCs.Source: §II · p. 2
  • The framework assigns continuous session verification to IPD watermarking, fleet observability to SIEM, and asset identity validation to a TPM root of trust.The stated roles include detecting anomalous flows, monitoring multi-source telemetry, and preventing rogue-device enrollment and identity masquerading.Source: §II · p. 2

III. THREAT MODEL AND APPROACH

The threat model considers a nearby on-path adversary seeking to disrupt real-time feedback or hijack control signals. A red-blue-team approach examines four attack vectors across OSI Layers 2–5.

A. Threat Model/Red Team

  • A localized on-path adversary may add packet latency or overwrite topic payloads, risking path divergence, boundary violations, or vehicle immobilization.The stated objective is to disrupt real-time feedback timing bounds or hijack control signals.Source: A. Threat Model/Red Team · pp. 2–3

III. THREAT MODEL AND APPROACH

  • The architecture figure maps the operational control loop, OSI Layers 2–5 attack vectors, and security capabilities.The section frames its evaluation as a red-blue-team perspective for validating risks and assessing defenses.Source: §III · p. 2, Figure 2 · p. 3

A. Threat Model/Red Team

  • The threat model covers de-authentication floods, ICMP floods, TCP SYN floods, and ARP-poisoning MITM command replacement across OSI Layers 2–5.The attacker is assumed to use standard commercial wireless hardware and open-source penetration-testing tools.Source: A. Threat Model/Red Team · p. 3

B. The Multilayer Integrated Defense Approach/Blue Team

The integrated defense spans hardware initialization, network-flow telemetry, and packet transmission.

B. The Multilayer Integrated Defense Approach/Blue Team

  • The three security capabilities operate across hardware initialization, network-flow telemetry, and packet transmission stages.The supplied passage identifies the stages but does not name the capabilities in this section.Source: B. The Multilayer Integrated Defense… · p. 3

1) Centralized SIEM Telemetry:

A centralized ELK Stack SOC collects fleet telemetry, while Kismet monitors the ambient 802.11 spectrum for disassociation frames.

1) Centralized SIEM Telemetry:

  • The centralized SOC runs an ELK Stack on a Raspberry Pi 5 and ingests data from lightweight shippers embedded on each robotic asset.The stack comprises Elasticsearch, Logstash, and Kibana.Source: 1) Centralized SIEM Telemetry · p. 3
  • Packetbeat parses frames and Layer 3 protocols, while Metricbeat monitors host resources and Heartbeat checks interface accessibility.Monitored host measures include socket allocations, memory state, and CPU utilization.Source: 1) Centralized SIEM Telemetry · p. 3
  • A separate Kismet node samples the 802.11 spectrum and alerts on unauthenticated broadcast disassociation frames.Source: 1) Centralized SIEM Telemetry · p. 3

2) Packet-Layer Inter-Packet Delay (IPD) Watermarking:

The packet-layer watermark embeds Gaussian timing offsets in command-packet intervals, and a rolling kurtosis detector flags deviations from the expected timing pattern.

2) Packet-Layer Inter-Packet Delay (IPD) Watermarking:

3) Hardware-Based Root of Trust:

Each robotic node uses a TPM 2.0 to verify hardware identity and integrity before joining the fleet topology.

3) Hardware-Based Root of Trust:

C. Zero-Trust Strategic Mapping

The section maps hardware, network, and packet components to Zero-Trust principles; the testbed connectivity setup is identified in Fig. 3.

C. Zero-Trust Strategic Mapping

IV. EXPERIMENTAL CONFIGURATION AND EMPIRICAL RESULTS

The section introduces the empirical evaluation of a multi-layer robotic-fleet security framework. Its figures identify the TPM-and-robot initialization and the security operations center (SOC) monitoring screen.

IV. EXPERIMENTAL CONFIGURATION AND EMPIRICAL RESULTS

  • The empirical section covers physical hardware, software middleware, and evaluations across four multi-layer attack scenarios.The section introduction describes the scope of the evaluations.Source: §IV · p. 4
  • Figure 4 depicts TPM and robot initialization.The associated caption identifies the figure's subject as TPM and robot initialization.Source: Figure 4 · p. 4
  • Figure 5 depicts the SOC monitoring screen.The associated caption identifies the figure as the SOC dashboard monitoring screen.Source: Figure 5 · p. 4

1) Hardware Testbed, Software Ecosystem, and Baseline Instrumentation:

The evaluation spans link-layer disconnection, network and transport floods, and MITM command injection. Volume-based monitoring detects the floods, while IPD-watermark kurtosis and ARP consistency monitoring detect the stealthier injection scenario.

1) Hardware Testbed, Software Ecosystem, and Baseline Instrumentation:

  • The testbed used a TurtleBot3 Burger with ROS 2 Humble and Fast DDS, sending velocity commands at 50 Hz over Wi-Fi; TPM identity attestation added 314 ms before network association.The vehicle used a Raspberry Pi 4B with 4 GB RAM; the access point ran on a Raspberry Pi 4.Source: 1) Hardware Testbed, Software… · pp. 4–5
  • The 802.11 de-authentication flood disconnected all devices from the gateway, severed the control loop, and triggered a real-time Kismet alert.Because the connection was dropped, standard network metrics could not be captured on the interface.Source: 2) Offensive Subsystem Configuration… · p. 5
  • ICMP flooding raised jitter from 0.5–0.6 ms to 3.0–10.0 ms and packet loss from 0.02%–0.08% to 0.4%–1.6%, disrupting feedback synchronization.The SIEM threshold monitor flagged the incoming ICMP surge; Figures 7 and 8 compare jitter and packet loss with and without the attack.Source: 3) Attack Scenario 2: Network-Layer… · p. 5, Figure 7 · p. 5, Figure 8 · p. 5
  • TCP SYN flooding raised jitter from 0.5–0.8 ms to 5.6–6.6 ms and packet loss from 0.02%–0.04% to 0.08%–0.4%, causing intermittent control-loop disassociation.A static volume threshold flagged anomalous SYN-state transitions; Figures 9 and 10 compare jitter and packet loss under attack and baseline conditions.Source: Figure 9 · p. 5, 4) Attack Scenario 3: Transport-Layer… · p. 6, Figure 10 · p. 6
  • During MITM relay, jitter remained nearly identical to baseline and packet loss varied only slightly above or below baseline, leaving standard network metrics blind to the exploit.Figures 11 and 12 compare jitter and packet loss with and without MITM attacks.Source: 5) Attack Scenario 4: Middleware… · p. 6, Figure 11 · p. 6, Figure 12 · p. 6
  • 20/20 active MITM injections were detected by IPD-watermark kurtosis monitoring, while 19/20 legitimate watermarked conditions were identified; ARP monitoring detected poisoning in 10/10 trials.The nominal watermark had sample kurtosis 2.8 ≤ K ≤ 3.2; the attacker relay produced spikes below 2.0 or above 4.5 over rolling five-second windows.Source: 5) Attack Scenario 4: Middleware… · pp. 6–7, Figure 13 · p. 6

V. CONCLUSION AND FUTURE WORK

The paper concludes that its layered framework was empirically evaluated on hardware-constrained wireless robotic networks. It reports complete detection of stealthy Layer 5 MITM injections using IPD-watermark kurtosis, alongside flood isolation by volume-based filters, and proposes automated deployment and adaptation to heterogeneous fleets as future work.

V. CONCLUSION AND FUTURE WORK

  • The framework combines TPM 2.0 hardware trust, ELK Stack and Kismet SIEM telemetry, and a non-cryptographic IPD timing watermark without application-layer cryptographic overhead.It was evaluated under OSI Layers 2–5 exploits on hardware-constrained wireless robotic networks.Source: §V · p. 7
  • Volume-based filters isolated brute denial-of-service floods, while IPD-watermark sample kurtosis exposed stealthy Layer 5 MITM command injections with complete detection accuracy.The conclusion reports these findings from the testbed evaluations.Source: §V · p. 7
  • Future work will automate node configuration for zero-touch tactical scaling and adapt real-time IPD and TPM loops to heterogeneous fleets, especially UAVs.The stated target includes UAVs' strict flight dynamics.Source: §V · p. 7
Loading 2609.05741v1…