A Smart Trust Management Method to Detect On-Off Attacks in the Internet of Things

Abstract

Internet of Things (IoT) resources cooperate with themselves for requesting and providing services. In heterogeneous and complex environments, those resources must trust each other. On-Off attacks threaten the IoT trust security through nodes performing good and bad behaviors randomly, to avoid being rated as a menace. Some countermeasures demand prior levels of trust knowledge and time to classify a node behavior. In some cases, a malfunctioning node can be mismatched as an attacker. In this paper, we introduce a smart trust management method, based on machine learning and an elastic slide window technique that automatically assesses the IoT resource trust, evaluating service provider attributes. In simulated and real-world data, this method was able to identify On-Off attackers and fault nodes with a precision up to 96% and low time consumption.

T‐D2D: A trust model for service offloading in device‐to‐device communication

“…The centralized trust models suffer from their reliance on higher level entity, which may be not available temporarily or permanently everywhere. To overcome this issue, other trust models perform both trust evaluation and provider selection in a distributed manner . In T‐D2D, all functions are executed distributively inside each IoT node, while recommendations are exchanged between different IoT nodes.…”

“…Some trust models are based on beta distribution or a variant of this technique to calculate the direct trust over the whole time, making them capable to only detect very intense on‐off attacks to some extent . Some other works use the same technique but over an interval of time, allowing them to improve their detection of very intense on‐off attacks . Some works consider twice punishment for malicious behavior, allowing them to detect intense on‐off attacks to some extent (ie, with at least half bad to good behavior ratio).…”

“…To overcome this issue, other trust models perform both trust evaluation and provider selection in a distributed manner. 5,12,14,[38][39][40]42,45 In T-D2D, all functions are executed distributively inside each IoT node, while recommendations are exchanged between different IoT nodes.…”

“…5,15,16,37 Some other works use the same technique but over an interval of time, allowing them to improve their detection of very intense on-off attacks. 14,42 Some works consider twice punishment for malicious behavior, 12,13,[38][39][40]45 allowing them to detect intense on-off attacks to some extent (ie, with at least half bad to good behavior ratio). However, if this approach is used over the whole time and not an interval of time, an attacker may compensate its misbehavior with normal conducts over time.…”

“…41 Caminha et al proposed a trust model that uses an elastic slide window to detect on-off attackers with regard to their reported data. 42 If the received data from an IoT device is out of the expected range, the window size is extended to obtain more observations. If the doubtful data are received yet in a longer window, the source of data is considered a broken node.…”