[Paper Review] Internet of Threats Introspection in Dynamic Intelligent Virtual Sensing

Victor R. Kebande, Joseph Bugeja+1 more|arXiv (Cornell University)|Jun 21, 2020

Energy Efficient Wireless Sensor Networks20 references4 citations

TL;DR

This paper investigates security threats in Dynamic Intelligent Virtual Sensors (DIVS) within IoT ecosystems, focusing on how malicious user feedback can inject false labels, compromising prediction accuracy. It proposes an introspective threat analysis framework to identify and mitigate risks to DIVS integrity, emphasizing the need for robust label validation in virtualized sensing environments.

ABSTRACT

Continued ubiquity of communication infrastructure across Internet of Things (IoT) ecosystems has seen persistent advances of dynamic, intelligent, virtualised sensing and actuation. This has led to effective interaction across the connected ecosystem of -things. Furthermore, this has enabled the creation of smart environments that has created the need for the development of different IoT protocols that support the relaying of information across billions of electronic devices over the Internet. That notwithstanding, the phenomenon of virtual sensors that are supported by IoT technologies like Wireless Sensor Networks (WSNs), RFID, WIFI, Bluetooth, ZigBee, IEEE 802.15.4, etc., emulates physical sensors, and enables more efficient resource management through the dynamic allocation of virtual sensor resources. A distinctive example of this has been the proposition of the Dynamic Intelligent Virtual Sensors (DIVS). This DIVS concept is a novel proposition that allows sensing to be done by the use of logical instances through the use of labeled data. This allows for making accurate predictions during data fusion. However, a potential security attack on DIVS may end up providing false labels during the User Feedback Process (UFP), which may interfere with the accuracy of DIVS. This paper investigates the threat landscape in DIVS when employed in IoT ecosystems, in order to identify the extent to which the severity of these threats may hinder accurate prediction of DIVS in IoT, based on labeled data.

Motivation & Objective

• To analyze the threat landscape affecting Dynamic Intelligent Virtual Sensors (DIVS) in IoT ecosystems.
• To identify how false labeling during the User Feedback Process (UFP) can undermine DIVS prediction accuracy.
• To evaluate the severity of security threats that could compromise the reliability of virtualized sensing in smart environments.
• To propose a framework for introspective threat analysis in dynamic, intelligent virtual sensing systems.
• To support secure and accurate data fusion by identifying vulnerabilities in label validation mechanisms.

Proposed method

• The authors analyze the DIVS architecture, focusing on its reliance on labeled data for accurate predictions.
• They identify the User Feedback Process (UFP) as a critical attack surface where false labels can be injected.
• A threat modeling approach is applied to assess potential attacks on the labeling process in DIVS.
• The study evaluates the impact of compromised labels on data fusion accuracy and system reliability.
• The framework emphasizes introspective analysis to detect and mitigate threats before they affect prediction outcomes.
• Security implications are assessed across IoT protocols like WSNs, RFID, Wi-Fi, Bluetooth, and ZigBee, which support virtual sensor operations.

Experimental results

Research questions

• RQ1 What are the primary security threats targeting the User Feedback Process (UFP) in Dynamic Intelligent Virtual Sensors (DIVS) within IoT ecosystems?
• RQ2 How can false labeling during UFP compromise the accuracy of DIVS predictions in virtualized sensing environments?
• RQ3 To what extent do existing IoT protocols and virtual sensor architectures expose DIVS to integrity attacks?
• RQ4 What introspective mechanisms can be implemented to detect and prevent malicious label injection in DIVS?
• RQ5 How do security vulnerabilities in virtual sensing affect the reliability of data fusion in smart environments?

Key findings

• False labeling during the User Feedback Process (UFP) poses a significant threat to the accuracy of DIVS predictions.
• The integrity of labeled data is critical for reliable data fusion in dynamic virtual sensing systems.
• Virtual sensors built on protocols like WSNs, RFID, and ZigBee are vulnerable to attacks that manipulate label inputs.
• The study identifies that current DIVS implementations lack sufficient introspective mechanisms to detect malicious label injection.
• Threats to label integrity can severely degrade the performance and trustworthiness of intelligent virtual sensing in IoT.
• The research underscores the need for proactive, introspective threat analysis to secure DIVS in dynamic, interconnected environments.