GRAPH SIGNAL PROCESSING FOR TRAFFIC ANALYSIS AND FORECASTING

Gacek, Andrew

Permalink

https://hdl.handle.net/2142/125096 Copy

Description

Title

GRAPH SIGNAL PROCESSING FOR TRAFFIC ANALYSIS AND FORECASTING

Author(s)

Gacek, Andrew

Issue Date

2020-05-05

Keyword(s)

Graph Signal Processing

Abstract

The application of Graph Signal Processing (GSP) techniques to traffic networks in urban settings is a challenging problem that has become the topic of many studies in recent works. In this paper, we analyze the effectiveness of these techniques to sparse image data collected in suburban environments using the STREETS dataset. We start with spectral analysis on these traffic states with the goal of discovering patterns in traffic data not obvious to the casual observer. Second, we introduce structure into the traffic forecasting problem with an adaptive graph filter. The performance of this filter is compared to non-graphical methods. Finally, we attempt to classify anomalous traffic incidents via graph wavelet decompositions.

Type of Resource

text

Language

eng

Owning Collections