Distributed Voting Classifier for DoS Detection in LLNs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Detecting Denial of Service (DoS) attacks in Low Power and Lossy Networks (LLNs) is challenging due to resource constraints, dynamic conditions, and the difficulty in distinguishing between legitimate and malicious traffic, especially in distributed DoS attacks.
Innovation Solution
A distributed voting mechanism using learning machines across the network, where potential voting nodes execute classifiers to select labels based on input features, and a set of eligible nodes is selected to determine a voting result, reducing information exchange and improving classification robustness even in constrained networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple classifiers are used to improve detection accuracy, then classification accuracy improves, but device complexity and resource consumption increase
Solution Approach 1:
The system segments the classification task by deploying multiple independent classifiers across different network nodes instead of using a single complex classifier. Each classifier operates autonomously on its local data, and their results are aggregated through voting. This segmentation reduces the complexity burden on individual devices while maintaining or improving overall classification accuracy through collective decision-making.
Solution Approach 2:
The system merges the outputs of multiple distributed classifiers through a voting mechanism to produce a final classification decision. By combining the strengths of multiple classifiers that operate in parallel across the network, the system achieves higher accuracy than any single classifier could provide alone, while distributing the computational load across multiple nodes.
2Reliability
If multiple classifiers are deployed across the network, then detection robustness improves, but information exchange overhead increases
Solution Approach 1:
The system extracts only the essential classification results from each distributed classifier rather than exchanging complete datasets or detailed intermediate computations. Each classifier processes local data independently and contributes only its classification outcome to the voting process, minimizing information exchange overhead while maintaining detection robustness through the collective input of multiple classifiers.
3Measurement precision
If distributed voting mechanism is implemented, then classification accuracy improves, but network resource consumption increases
Solution Approach 1:
The system implements a practical voting mechanism where a sufficient number of classifiers participate in the voting process to achieve reliable accuracy without requiring every possible classifier to vote. This partial action approach balances the need for accurate classification with the constraints of network energy consumption, allowing the system to achieve diminishing returns at a reasonable cost.
Data Source
AI summary
In one embodiment, possible voting nodes in a network are identified. The possible voting nodes each execute a classifier that is configured to select a label from among a plurality of labels based on a set of input features. A set of one or more eligible voting nodes is selected from among the possible voting nodes based on a network policy. Voting requests are then provided to the one or more eligible voting nodes that cause the one or more eligible voting nodes to select labels from among the plurality of labels. Votes are received from the eligible voting nodes that include the selected labels and are used to determine a voting result.