|M.Sc Student||Margalit Elad|
|Subject||Real-Time Optimization in Event Based Systems for Traffic|
|Department||Department of Industrial Engineering and Management||Supervisor||Dr. Opher Etzion|
|Full Thesis text - in Hebrew|
The purpose of this work is to examine, using Complex Event Processing (CEP), the ability to solve problems that require decisions making in real time. The problems presented in this work are characterized by the need for making decisions involving two constraints; one which involves decisions that have to be as close as possible to optimality during deterministic time, and the second one should involve a learning process. One kind of problem that is typical to this issue constitutes timing of traffic light.
Ever since the existence of motor vehicle the development of roads infrastructure was slower than the development of transportation vehicles. This fact consistently creates heavy traffic in day to day life. These circumstances create a situation of traffic movement in the urban and inter-urban areas that quiet often entail some serious problems of operation. Passing an intersection usually involves delays, halting, lost of working hours and waste of energy. Those multiple halts and accelerations while getting stuck in traffic generate an increase in air pollution and a higher consumption of gasoline not to mention the noise.
During the research a model of traffic lights' net was established based on the Even-Driven Architecture (EDA) which is architecture of pattern promoting the production, detection, consumption and reaction of events.
Following the EDA, different processes were running in the background in order to solve the optimization problem of the model in real time. Finally the data was processed back into the system with running/activation results. The model includes several scenarios and examines the ability to deal with them efficiently and successfully.
It was found that using the Complex Event Processing (CEP) which is an event that involves number of previous events within a certain context provides a significant improvement in regard to the system functionality particularly in unexpected cases. The implementation possibilities of this strategy are depending on the ability of the system to be accurate and to its ability to identify data in real time.