Social insects such as ants, bees and wasps perform complex tasks with pheromone communication despite lack of top-down style control. We have examined applications of this pheromone mechanism towards ITS. In this paper, a car is regarded as an insect that releases virtual pheromone that represents traffic congestion level. We propose a method to predict traffic congestion of the immediate future through a pheromone mechanism without resort to the use of a traffic control center. Furthermore we introduced multi semantics of pheromone to refine the prediction performance. We evaluate our method using actual traffic data and the results indicate our method is superior to other conventional techniques and our previous method.