Optimal decentralized regulation for a class of flow networks

Optimal decentralized regulation is considered for a class of flow networks where, at each intersection, a selective switch links a single input-output pair at anytime. It is shown that a special case of the above is an urban traffic network with signalized intersection. We analyze first the traffic situation of an isolated intersection based on the point-queuing model of traffic and model the intersection dynamics via two states: either unsaturated or saturated. According to the different traffic characteristics of the two states, we design two intersection controllers and then combine them into one hybrid controller. This hybrid controller is extended to the multi-intersection case and becomes a decentralized hybrid intersection controller. A simulation study is given in this paper and it shows that this decentralized hybrid intersection control method can improve the performance of traffic networks and by extension the performance of more general flow networks.