Characterization of network traffic processes under adaptive traffic control systems

摘要: Abstract We present a compact characterization of network-level traffic processes for dense urban area operating under an adaptive control system. The is based on state classification scheme that employed at detector level, and transition model works with combinations detectors are topologically dependent. Jointly, the two models provide concise but rich representation network level. key insight identification transient states, termed under-utilized (U) where effects such as insufficient downstream capacity captured. In states green time not fully used. approach provides space-time evolution across network, conditional probabilities upstream drive propagation in near term, probabilistic information congested paths described sequence detectors. paper presents empirical evidence SCATS system Dublin, insights provided by proposed approach, importance which represent much 20% unused along certain corridors peak periods. results basis future procedures.