Promet – Traffc&Transportation, Vol. 30, 2018, No. 2, 205-215 205 ABSTRACT Safety systems detect unsafe conditions and provide warnings for travellers to take action and avoid crashes. Es- timation of the geographical location of a moving vehicle as to where it will be positioned next with high precision and short computation time is crucial for identifying dangers. To this end, navigational and dynamic data of a vehicle are pro- cessed in connection with the data received from neighbour- ing vehicles and infrastructure in the same vicinity. In this study, a vehicular location prediction model was developed using an artifcial neural network for cooperative active safe- ty systems. The model is intended to have a constant, short- er computation time as well as higher accuracy features. The performance of the proposed model was measured with a real-time testbed developed in this study. The results are compared with the performance of similar studies and the proposed model is shown to deliver a better performance than other models. KEY WORDS cooperative active safety systems; inter-vehicular commu- nication; vehicular location prediction; artifcial neural net- works; 1. INTRODUCTION Road transportation has an indispensable role in human life despite all the hitches in economic, en- vironmental and social aspects. Thus, multi-dimen- sional research and development studies have been carried out for safer and more effective road transpor- tation. Developments in information and communica- tion technologies have been rapidly deployed in road transportations in the last decade. The main principle of safety applications is that moving vehicles recognize possible dangers and elim- inate them. Inter Vehicle Communication (IVC) is one of the approaches that makes the road transporta- tion safer, more effcient and more comfortable via wireless technology. In this approach, vehicles ex- change their information such as dynamic and nav- igational values between each other and road infra- structures. Therefore, IVC makes possible cooperative active safety systems (CASS), which provide more ef- fective safety through cooperation among vehicles and infrastructures. Not only are data received from the vehicles and the infrastructures in the surrounding area, but also the dynamic and navigational values fetched from the vehicles are processed to recognize the possible dan- gers. Real-time estimation of the future geographical po- sition of a vehicle in motion is called vehicular location prediction (VLP). It is also referred to by such terms as state estimation, mobility prediction, path prediction, and vehicle tracking. High precision and short compu- tation times are essential components of VLP models. A perusal of literature shows that state-space equa- tions are commonly used for prediction questions. The matrix algebra used in these equations affects the computation time adversely. In particular, the com- putation time increases gradually as prediction time interval increases. The VLP methods with long com- putation times are not convenient. Moreover, the VLP methods with inconstant computation times are much more likely to cause desynchronization between vehi- cles or devices. This study focuses on how to conduct VLP with constant, short computation times as well as high accuracy on the highways and the rural roads for CASS. For this purpose, artifcial intelligence-based solutions are preferred over state-space-based mod- els. Artifcial Neural Network (ANN), which becomes prominent in predicting techniques through artifcial intelligence methods, was selected. The development and performance tests were conducted in real world Dörterler M, Bay ÖF. Neural Network Based Vehicular Location Prediction Model for Cooperative Active Safety Systems Intelligent Transport Systems (ITS) Preliminary Communication Submitted: 24 Mar. 2017 Accepted: 27 Nov. 2017 MURAT DÖRTERLER, Ph.D. 1 (Corresponding Author) E-mail: dorterler@gazi.edu.tr ÖMER FARUK BAY, Ph.D. 2 E-mail: omerbay@gazi.edu.tr 1 Department of Computer Engineering, Faculty of Technology, Gazi University 06500, Teknikokullar, Ankara, Turkey 2 Department of Electrical - Electronic Engineering, Faculty of Technology, Gazi University 06500, Teknikokullar, Ankara, Turkey NEURAL NETWORK BASED VEHICULAR LOCATION PREDICTION MODEL FOR COOPERATIVE ACTIVE SAFETY SYSTEMS