Estimation of its sensor operational states by analyzing measurements with errors using a Hidden Markov Model

• Published in: KSCE journal of civil engineering Vol. 17; no. 7; pp. 1740 - 1748
• Main Authors: Randeniya, Duminda, Kim, Hoe Kyoung
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01-11-2013; Springer Nature B.V; 대한토목학회
• Subjects: Autonomous; Civil Engineering; Computer simulation; Detectors; Engineering; Geotechnical Engineering & Applied Earth Sciences; Industrial Pollution Prevention; Intelligent transportation systems; Markov chains; Mathematical models; Modules; Noise measurement; Reliability analysis; Sensors; Simulation models; Statistical analysis; Stochastic models; Stochasticity; Transport; Transportation Engineering; 토목공학; microscopic simulation model; VISSIM; inductive loop detector; hidden Markov model; Intelligent Transportation System
• ISSN: 1226-7988; 1976-3808
• DOI: 10.1007/s12205-013-0284-2

One of the prominent problems in introducing autonomous and advanced transportation technologies to existing traffic network systems is the lack of a framework that constantly evaluates measurement reliability and consistency of Intelligent Transportation System (ITS) sensors. In order to tackle this problem, one needs to identify the sensor state rather than the measurement itself because statistical properties of the measurement noise profile in the ITS sensor vary from situation to situation, whereas the number of states a sensor can attain remain the same. Therefore, this paper develops a stochastic model that uses a Hidden Markov Model (HMM) to identify ITS sensor states from observed measurements with two supplemental modules (i.e., a Rule-based Diagnosis Module and a Statistic and Neighbor Feedback Module). Also, this paper tests the proposed model using an aggregated dataset (15 minute interval) obtained from a portion of Interstate 40 (I-40) in Knoxville, Tennessee, which was simulated to increase the resolution by an off-the-shelf microscopic simulation model (i.e., VISSIM). The case study results indicate that the HMM coupled with two supplemental modules can accurately identify sensor operational states more than 98% of the time and its performance is consistent along all the Inductive Loop Detector (ILD) stations.