IJCSNS International Journal of Computer Science and Network Security, VOL.7 No.2, February 2007 106 Manuscript received February 5, 2007 Manuscript revised February 25, 2007 A GIS-Based Design and Implementation Approach for Modeling Driver’s Behavior in Route Selection Using Fuzzy-Neural Networks Parham Pahlavani, Mahmoud Reza Delavar Center of Exellence in Geomatics Eng. and Disaster Management, Dept. of Surveying and Geomatics Eng., Eng. Faculty, University of Tehran, Tehran, Iran Summary For modeling a driver's behavior in route selection in outdoor situations we have two problems: 1-real situation is very often not crisp and deterministic and cannot be described precisely, 2-the complete description of driver's behavior in route selection often require much more detailed data than driver could ever recognize process, and understand simultaneously. In this paper we have designed and implemented a GIS-based fuzzy-neural approach for modeling driver’s behavior which represents the correlation of the attributes with the driver’s route selection. A recommendation or route fitness is provided to the driver based on a training of the fuzzy adaptive neural network on the main criteria of route selection such as length, time and the degree of difficulty. Tests of route selection for a part of North-West of Tehran traffic network are conducted and the results show the efficiency of the algorithm and support our analyses. Key words: Route selection, deriver’s behavior, fuzzy adaptive neural network(FALCON) 1. Introduction Decisions are often evaluated on the basis of quality of the processes behind. It is in this context that geospatial information systems (GIS) and spatial decision support systems (SDSS) increasingly are being used to generate alternatives to aid decision-makers in their deliberations. Decision making itself, however, is broadly defined to include any choice or selection of alternative course of actions, and is therefore of importance in many fields in both the social and natural sciences including geospatial information sciences. Among so many implementations GIS, a GIS application for Transportation (GIS-T) has become an outstanding one. It is possible to state unequivocally that GIS-T has arrived and now represents as one of the most important application areas of GIS. Advanced Traveler Information Systems (ATIS) assist travelers with planning, perception, analysis and decision making to improve the convenience, safety and efficiency of travel. ATIS is one component of the Intelligent Transport Systems (ITS) that currently being developed to improve the safety and efficiency of automobile travel. Route planning is therefore an essential component of ATIS, aiding travelers in choosing the optimal path to their destinations in terms of travel distance, travel time and many other criteria. It is this multi-criteria aspect of route planning that we wish to tackle. For the first time, we outlined a GIS-based novel approach for using a genetic algorithm for urban multi-objective optimized route selection in static environment [1] and an innovative method that extends the previous novel approach in order to include driver's unspecified sites [2]. Although the above approaches proposed the quasi-optimal route by the driver's consideration for the importance of each route criterion, it is essential to say that none of them concerned the driver's behavior. It is believed that each driver has a set of route choice preferences. Very often, drivers would try to select the route which is optimum with reference to their preferences. In the other words, each quasi-optimal route may have an especial meaning for each driver, so it would be necessary to have a ranking route engine for proposed genetic algorithm in previous approaches in order to calculate the fitness of each route based on every driver's behavior and preferences. The objectives of this paper are to design a ranking route engine as follows: • It is a decision support system for route selection. • It can model the behavior of the drivers by storing their preference and previous choices. • It can adapt and learn from the recent decisions of the drivers. Each route candidate has a set of attributes. A GIS-based fuzzy-neural approach is used to represent the correlation of the attributes with the driver’s route selection. A recommendation or route ranking can be provided to the drivers based on a training of the fuzzy-neural network on the main criteria of route selection. This convenience is needed and may happen when planning a special trip on a particular day. It is used as a quick and convenient means