Archival Version

The overview of the proposed contraband wireless device detection system consists of hardware and software components. The hardware component is a receiver with multiple antennas, each of which receives the wireless signal at different locations. The received wireless signals carry unique characteristics of the device, known as device fingerprints and unique information (e.g., arrival strength and phase angle) of their propagation paths, known as location fingerprints. The software component takes the received multi-channel signals as input and applies RF signal analysis approaches including mathematical models and machine learning algorithms to differentiate devices and locations. To achieve the project goals and objectives, the study team developed and implemented the following three types of methods: device fingerprinting, location fingerprinting, and content summarization. For further technical details for each method please refer to the related publications.

本次提出的违禁无线设备检测系统（contraband wireless device detection system）整体架构包含硬件与软件两大组成部分。硬件组件为多天线接收机，各天线可在不同部署位置采集无线信号。所采集的无线信号携带有设备的独有特征，即设备指纹（device fingerprints），以及信号传播路径的独有信息（例如到达强度与相位角），即位置指纹（location fingerprints）。软件组件以采集到的多通道信号作为输入，采用包括数学模型与机器学习算法在内的射频（Radio Frequency, RF）信号分析方法，实现设备与位置的区分识别。为达成本项目的既定目标，研究团队开发并实现了三类核心方法：设备指纹分析、位置指纹分析与内容摘要分析。若需了解各方法的详细技术细节，请参阅相关学术出版物。