Use of multi-objective genetic rule selection for examining the effectiveness of inter-vehicle communication in traffic simulations

Recently, inter-vehicle communication (IVC) has been actively studied to attempt to avoid traffic congestion. In this article, we propose the idea of using fuzzy rules to examine the effectiveness of IVC. In the proposed approach, we first collect travel records (e.g., travel time, travel path, traffic volume) of vehicles with IVC from our cellular automata-based traffic simulator. Various kinds of available information for vehicles with IVC are used in the antecedent part of our fuzzy rules. The level of effectiveness of IVC is discretized into four categories (i.e., four classes) in this article. The consequent class of each fuzzy rule is one of those four classes. Next we generate a large number of fuzzy rules from the collected data. Then we select only a small number of fuzzy rules by multi-objective genetic rule selection. We use three objectives: to maximize the accuracy, to minimize the number of selected rules, and to minimize the total rule length (i.e., the total number of antecedent conditions). Our approach can find a number of nondominated fuzzy-rule-based systems with respect to their accuracy and complexity. Finally, we analyze the effectiveness of IVC using fuzzy rules in the fuzzy-rule-based systems obtained through their linguistic interpretation.