Distributed coordination control of traffic network flow using adaptive genetic algorithm based on cloud computing

Abstract Adaptive traffic signal control can automatically adjust signal cycles, offsets and green ratios with the fluctuations of traffic flow to reduce the delay of vehicle fleets. At present, cloud computing has provided a technically feasible realization measure for the real-time optimization processing of adaptive traffic signal control with traffic big data. This paper proposes a by-level optimization strategy of signal cycles, offsets and green ratios on the cloud computing platform. The optimization of cycles and offsets among road intersections coordinates traffic flow on road links, called coordination control. The optimization of green ratios at each intersection further controls traffic flow on each road link, called distributed control. The coarse-grained parallel adaptive genetic algorithm (CPAGA) is developed for the optimization of distributed coordination control. On the cloud computing platform, the common data of road network and traffic flow is located in the bottom layer and transparently shared to related computing nodes. The CPAGAs at computing nodes are deployed to perform coordinated distributed optimization of adaptive traffic signal control. A migration strategy is developed for the parallel genetic algorithm where a portion of worse chromosomes at one computing node is replaced by better chromosomes at another node with a replace probability. The distributed coordination control procedures are designed with computing nodes being dynamically deployed at different optimization phases. The simulation experiments with regard to a realistic traffic network are carried out on the cloud computing platform. The numerical results demonstrate that the CPAGA can avoid falling into a local optimum and has fast solution efficiency. The proposed distributed coordination control of traffic signals is superior to the commonly used timing control at reducing traffic delay.