Multi-Variate Data Fusion Technique for Reducing Sensor Errors in Intelligent Transportation Systems
Gunasekaran Manogaran,Venki Balasubramanian,Bharat S. Rawal,Vijayalakshmi Saravanan,Carlos Enrique Montenegro-Marin,Varatharajan Ramachandran,Priyan Malarvizhi Kumar
DOI: https://doi.org/10.1109/jsen.2020.3017384
IF: 4.3
2021-07-15
IEEE Sensors Journal
Abstract:Connected vehicles in intelligent transportation system (ITS) scenario rely on environmental data for supporting user-centric applications along the driving time. Sensors equipped in the vehicles are responsible for accumulating data from the environment, followed by the fusion process. Such fusion process provides accurate and stable data required for the applications in a recurrent manner. In order to enhance the data fusion of connected vehicles, this article introduces multi-variate data fusion (MVDF) technique. This technique is competent in handling asynchronous and discrete data from the environment and streamlining them into continuous and delay-less inputs for the applications. The process of data fusion is aided through least square regression learning to determine the errors in different time instances. The indefinite and definite data fusion instances are differentiated using this regression model to identify the errors in fore-hand. Besides, the differentiation relies on the application run-time interval to progress data fusion within the same or extended time instance and data slots. In this manner the differentiation along with the error identification is regular until the application required data is met. The performance of this technique is verified using network simulator experiments for the metrics error, data utilization ratio, and computation time. The results show that this technique improves data utilization under controlled time and fewer errors.
engineering, electrical & electronic,instruments & instrumentation,physics, applied