Highway Traffic Emergency Auto-Detection Using Wireless Sensor Network

Traffic emergency is tremendous harmful, it is one of the key factors of disruptingnormal traffic order, causing unexpected traffic jams and congestion, leading heavilyeconomic harm. It was estimated that the losses have reached more thanï¿¥1,000 billionYuan every year caused by unexpected traffic incident. Traffic emergency management is anefficient way to reduce harms or incident loss, to recover rapid resumption of normal trafficorder and improve transportation efficiency. Traffic emergency automatically detected is thebasic of traffic incident management; there are two processes of automatic incident detection:automatic data collection and automatic data analysis.First, data collection technology and incident detection algorithm of traditional automaticincident detection (Automatic Incident Detection, called AID) system is researched in-depth.A new detection approach, Highway Traffic Emergency Auto-Detection, called HTEAD, hasbeen designed based on wireless sensor network technology after analyzed the disadvantagesof detection mechanisms of pre-exist AID, such as the mean time to detect(MTTD) is too long,the cost price is too high, sometimes is not available, etc. HTEAD detects the unexpectedtraffic incident using the real-time traffic flow data or the features of incident. The newdetection mechanisms of HTEAD can overcome the disadvantages of detection mechanismsof traditional AID.Secondly, the features of unexpected traffic incident has been addressed primarily, thenthe key vehicle detection and data fusion in-network technologies about traffic emergencyautomatic detection using wireless sensor network has been researched. The methods andtechnology about how to get these features, such as car crash sound, real-time vehicle speed,vehicle driving direction, based on microphone and AMR (Anisotropic Magneto-Resistive)sensors integrated on the intelligent sensor node are also been discussed. In addition, somecommon road traffic emergencies' detection were studied based on incident types table usingin-network data fusion technologies, such as vehicle crashes, vehicle going in wrong direction,anchored vehicle, etc.Finally, the unexpected traffic incident detection experiment has been tested usingHTEAD, normal traffic background sounds, car collisions sounds, geomagnetic disturbances,and other physical characteristics were been collected in real traffic scene. The vehicle converse illegally and vehicle collision incidents has been detected successfully using thosedata. Result show that this method has broad application prospects.