Device-Free Counting via OFDM Signals of Opportunity

Abstract

Counting targets (people or things) within a monitored area is an important task in emerging wireless applications for smart environments, safety, and security. Counting via passive radars rely on signals of opportunity (i.e., signal already on air for other purposes) to detect and count device-free targets, which is preferable, in terms of privacy and implementation costs, to active radars that rely on dedicated or personal devices. However, conventional radar techniques for multi-target detection require to associate measurements sets to detected targets. Such data association may lead to high dimensionality and complexity even with few targets, despite it is a redundant operation for counting. The need of low dimensionality and complexity calls for the definition of signal features and the development of techniques for their extraction, which enable the association of measured signals directly with the number of targets (namely, crowd-centric algorithms). This paper develops a framework for the design and analysis of crowd-centric algorithms for device-free counting via OFDM signals of opportunity. Preliminary results in simple use cases show the effectiveness of the proposed techniques with respect to individual-centric algorithms.

Crowd Counting Model Training with the Method of Moments in Electromagnetics

“…A step in this direction has already be taken in [5], where a multistatic radar environment with uniformly distributed targets is simulated and used to generate channel impulse responses on which counting algorithms can be trained. However, it considers a limited bandwidth of 20 MHz, and uses a geometry-based single-bounce channel model with simple path loss, hence considering point targets and neglecting electromagnetic interactions between targets.…”