Wireless Device Detection Using Streaming Survival Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wireless device detection systems struggle to accurately identify and track devices with obfuscated identities due to MAC address randomization and temporary identifiers in Wi-Fi and cellular protocols, making it difficult to distinguish between known and unknown devices.
Innovation Solution
A survival modeling approach is applied to process temporary identifiers, constructing a cumulative distribution function based on delta times between access events to assign survival probabilities, maintaining a streaming list of likely devices, and calibrating with ground truth data to detect anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If MAC address randomization and temporary identifiers are used to obfuscate device identity, then device security and privacy are improved, but device detection and tracking accuracy deteriorate
Solution Approach 1:
The system implements feedback by continuously monitoring temporary identifier usage patterns and updating the detection model with observed data. The survival analysis model learns from actual identifier lifecycles and adjusts probability estimates based on observed patterns, creating a closed-loop system that improves detection accuracy while respecting privacy-preserving identifier rotation.
Solution Approach 2:
The patent introduces survival probability analysis as an intermediary layer between the obfuscated temporary identifiers and the detection decision. Instead of directly tracking identifiers, the system calculates survival probabilities that mediate between the randomizing identifiers and the final detection outcome, enabling accurate tracking without requiring direct identifier persistence.
2Device complexity
If traditional device tracking methods are used, then device identification is simplified, but computational resources and memory consumption increase
Solution Approach 1:
The system extracts only the essential temporal characteristics from the identifier data stream, focusing solely on the time-based survival analysis rather than processing complete identifier histories. This extraction approach reduces computational burden by concentrating on the critical temporal dimension while ignoring redundant identifier details.
Solution Approach 2:
The patent employs lightweight survival probability calculations that can be quickly computed and discarded as new data arrives. Each survival probability estimate is a computationally inexpensive object that serves its purpose for immediate detection decisions and is then replaced by new calculations, avoiding the need to maintain complex long-term state structures.
3Measurement precision
If comprehensive device monitoring is implemented, then detection accuracy is improved, but real-time processing capability deteriorates
Solution Approach 1:
The system performs preliminary action by pre-computing the cumulative distribution function from historical data before real-time detection occurs. This pre-processing step creates a ready-to-use statistical model that can be quickly applied to incoming data streams without requiring complex real-time computations, thus maintaining both accuracy and speed.
Solution Approach 2:
The patent implements dynamic updating of survival probabilities as new data arrives, allowing the system to adapt to changing conditions in real-time. The model continuously updates its probability estimates based on new identifier observations while maintaining computational efficiency through incremental updates rather than complete re-computations.
Data Source
AI summary
A wireless device detection system includes sensors to receive temporary identifiers transmitted between a base station and a mobile wireless device. A cumulative distribution function for survival probability is generated based on delta times between multiple access events for each temporary identifier. In response to a new access event, a corresponding temporary identifier is added to a streaming list and assigned a survival probability value, based on the cumulative distribution function, to a latest access event for each temporary identifier contained in the streaming list. Temporary identifiers that have a survival probability value less than a threshold value are removed from the streaming list. The number of temporary identifiers contained in the streaming list are compared to a number of devices known to be present. The probability that a device corresponding to each of the temporary identifiers contained in the streaming list is present is determined.