Determination of Network Origin-Destination Matrices Using Partial Link Traffic Counts and Virtual Sensor Information in an Integrated Corridor Management Framework

015292912014FINAL REPORTPDFTech ReportProject No. 056PY03DTRT07-G-0005Origin and destinationTraffic countsIntegrated corridor managementTrip matricesTrip distributionSensorsData collectionTravel demandResearch HubHighwaysPlanning and ForecastingI72: Traffic and Transport PlanningIntegrated corridor managementLocationOrigin and destinationSensorsTraffic countsTraffic flowTravel demandUnited StatesNEXTRANS Center (U.S.)Hu, Shou-RenLiou, Han-TsungPeeta, SrinivasZheng, HongNEXTRANS Center (U.S.)United States. Dept. of Transportation. Research and Innovative Technology AdministrationUS Transportation CollectionTrip origin-destination (O-D) demand matrices are critical components in transportation network modeling, and provide essential information on trip distributions and corresponding spatiotemporal traffic patterns in traffic zones in vehicular networks. Trip O-D matrices also reflect traffic loadings and flow intensities in transportation networks, and are crucial inputs in determining short-term traffic control schemes and long-term transportation improvement programs, as well as offline transportation planning and online traffic management. Trip O-D demand matrices have traditionally been estimated by conducting household surveys or roadside interviews; however, this is infeasible because of the high cost and data recording errors involved. Inferring network O-D demand matrices using corresponding link flows is an effective alternative approach, because link flows, which are a set of traffic flows associated with the vehicular trip distributions of different O-D pairs, are easily obtained. Past studies on the estimation of network O-D demand matrices using link flow information have generally assumed that link flows are fully observable. However, in practice, highway agencies face budget constraints in implementing comprehensive sensor deployment plans, and assuming the full observability of link flows is unreasonable. Because of the rapid development of information and communication technologies (ICTs), applications of advanced sensor technologies to traffic management and operation have become widespread and essential. As a result, determining the strategic deployment of traffic sensors to obtain necessary traffic information for network O-D demand estimation has become crucial in transportation network research