Minimum Description Feature Selection for Complexity Reduction in Machine Learning-Based Wireless Positioning

Background

 

Wireless positioning (WP) is typically conducted using a group of wireless sensors that exchange signals with a target of interest in order to collect measurements that are informative for location estimation. These sensors form a network, and the measurements from each sensor are collected by a data fusion center (DFC) for centralized processing to estimate the target location. Although these WP algorithms have attained excellent and consistent performance against complex channel environments, the computational complexity coming from processing high dimensional features can be prohibitive for mobile applications. Recently, deep learning approaches have provided solutions to these rising difficult problems in wireless positioning.

 

Invention Description

 

Researcher at Arizona State University and Purdue University have developed a Positioning Neural Network (P-NN) that introduces a multi-channel deep learning architecture that processes a minimal set of the largest power measurements and their temporal locations instead of full high-dimensional power delay profiles. By converting these features into sparse images and numerical matrices, and combining convolutional and self-attention layers, P-NN efficiently captures spatial and temporal information for zone-based location classification. An adaptive feature size selection technique optimizes the balance between accuracy and complexity without prior training, enabling robust performance in challenging signal environments.

 

Potential Applications:

Benefits and Advantages: