Wireless Sensing ML Training With Deceptive Features for Security
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless sensing technologies lack security measures to protect against illegitimate users who perform passive sensing by intercepting and analyzing legitimate user's sensing signals, making it difficult to detect unauthorized access.
Innovation Solution
A method for training a machine learning model by introducing deceptive features into the input instances, allowing the model to differentiate between legitimate and illegitimate sensing signals, thereby enhancing security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If wireless sensing signals are transmitted without security measures, then sensing functionality is maintained, but illegitimate users can perform passive sensing and intercept data
Solution Approach 1:
The patent applies parameter changes by modifying the sensing signals through deception techniques. The system changes signal parameters such as phase, frequency, or timing characteristics to create deceptive sensing signals that mislead illegitimate users. This allows the system to maintain sensing functionality while preventing unauthorized passive sensing, as the altered parameters make it difficult for attackers to extract accurate information from intercepted signals.
2Reliability
If deception techniques are applied to sensing signals, then security against passive sensing is improved, but signal processing complexity increases
Solution Approach 1:
The patent implements preliminary action by pre-processing the sensing signals with deception characteristics before transmission. The system prepares deceptive signal patterns in advance and embeds them into the transmitted sensing signals. This preliminary preparation ensures that when illegitimate users intercept the signals, they already contain misleading information, reducing the need for complex real-time detection and response mechanisms.
3Measurement precision
If machine learning models are trained with deceptive features, then classification accuracy is improved, but training data requirements increase
Solution Approach 1:
The patent applies copying by creating synthetic training data that replicates deceptive signal patterns. Instead of requiring extensive real-world captured data with deception, the system generates artificial training samples that mimic the characteristics of deceptive sensing signals. This allows the machine learning model to learn effective classification patterns without needing a prohibitively large volume of actual intercepted deceptive signals.
Data Source
AI summary
The present disclosure relates to improving security in sensing applications. A sensing system utilizing a ML model is provided. In a method for training the ML model, a first dataset including a plurality of first input instances and associated first output labels is obtained. Further, a second dataset including a plurality of second input instances and associated second output labels is obtained, wherein the second dataset is obtained, by selecting a subset of the plurality of first input instances and associated first output labels, generating the plurality of second input instances by introducing a deceptive feature into each of the subset of the first input instances, and setting the first output labels associated with the subset of the first input instances as the second output labels. The training method further includes training a machine learning model using the second dataset.