Cloud Computing For Real-Time Decision Making

Traditionally, the Decision Support Systems which are used for decision making rely on analyzing large amounts of data by using traditional components such as database, models and user interfaces to display the data. These systems are in no way close to a real-time system as they often take time in gathering, processing and analyzing the input to display the results. On the other hand, the real-time systems traditionally have often focused on the speed while keeping the task on hand very simple. Although there have been a few examples of a real-time decision support system such as air traffic control system, which are highly specialized but require a significant infrastructure investment. Hence, there is a clear gap in the industry for a solution that provides an accurate decision support system for real-time decision-making while being relatively affordable, scalable and mobile. Cloud computing has been a game changer in the quest for achieving a real-time decision making by offering a seamless integration of different devices and technologies that operate and communicate under one roof. This study proposes that the powerful capabilities of cloud computing with all its offerings can be harnessed to provide a robust solution for real-time decision making. To test the proposed hypothesis systematically by using the power of the cloud to achieve the real-time decision making, a unique architecture using the building blocks of the cloud is laid out. Field experiments for achieving the real-time detection of buried pipelines and spills are conducted by varying the difficulty level and post-process analysis and auxiliary sources are used to improve the model’s accuracy. In order to achieve the real-time pipeline detection, Microsoft Azure was selected as the preferred cloud solution using the “pay-as-you-go” model and the key components of the cloud such as IoT (Internet of Things) sensors, serverless computing, edge computing, machine learning, data warehousing, and analytics were identified, leveraged and an architecture is proposed that allows the development of robust workflows, and the results of real-time detection were transmitted to users through email notifications. Furthermore, a dashboard based on Power BI which is a cloud-based business intelligence, data visualization and analytics tools were created to monitor the real-time information. In addition to the implementation of the real-time workflows, two methods were proposed and successfully implemented to improve the detection accuracy, the first of which implemented a postprocessing workflow to further improve the machine learning models which were used for the real-time application by maximizing the field data value for the decision-making process while detecting pipelines or any incident in the field. The second method used a novel approach to improve the accuracy of the real-time detection by applying machine learning on the identified auxiliary sources by quickly training with a small dataset. These findings demonstrate that the cloud and its offerings would address a key gap in the industry by providing a viable solution for a decision support system which can be implemented in real-time. Such systems can be crucial in oil and gas, power, manufacturing, health and safety industries that need real-time data to achieve continuous improvements in reliability, safety and efficiency