A node-based network sensor location model for path flow reconstruction

  • 王 聖儒

Student thesis: Master's Thesis

Abstract

Vehicle detectors or traffic sensors play an indispensable role in modern traffic management systems They can not only collect traffic parameters in a vehicular network but also can be used for traffic monitoring and/ or control purposes However traffic management agencies are not able to comprehensively deploy a larger number of sensors in practice due to an annual budget constraint It drives the need to address the problem of optimal sensor locations under a budget constraint and/ or within a specific region This problem and its variants is called the network sensor location problem (NSLP) In the previous studies of NSLP they mainly aimed at finding a smallest subset of the links or nodes for sensor deployment by minimizing the flow estimation errors or focused on the estimation of network origin-destination (O-D) demands given a set of prior deployed sensors With the advent of internet of things (IoT) and the rapid development of information and communication technologies (ICTs) application of advanced sensors for traffic data collection becomes feasible For the NSLP this study assumes a node-based smart virtual sensor is able to detect vehicle trajectories in a given network via two-way communications We seek to identify the smallest subset of nodes for the NSLP by using the vertex cover method Finally we develop mathematical models by under the limited shortest paths between each O-D pair can be recognized to estimate path flows and O-D demand Then the numerical analysis is divided into two parts: first part is to find optimal location(s) for sensor deployment by vertex cover method in order to detect all link flows in the network; second part is to construct mathematical models to estimate path flows and O-D demand under sensors are optimally located The results of numerical analysis show that there are multiple optimal locations for sensor deployment by the vertex cover method and it helps to estimate path flows and O-D demands when the partial path trajectories are captured by virtual smart sensors Addressed from a frontier viewpoint this study proposes an innovative idea for the NSLP Looking into the future collecting traffic information via advanced sensor technologies becomes popular due to matured technologies and decreasing costs In addition the developed models and solution algorithm proposed in this study are beneficial to traffic agencies for on-line traffic control and management purposes
Date of Award2018 Aug 18
Original languageEnglish
SupervisorShou-Ren Hu (Supervisor)

Cite this

A node-based network sensor location model for path flow reconstruction
聖儒, 王. (Author). 2018 Aug 18

Student thesis: Master's Thesis