An Adaptive Method for Acquisition of Dynamic Data in the Internet of Things

Internet of Things is considered of great importance in Industry 4.0 trend. Being on the “peak of inflated expectations” it is being widely implemented in businesses and production. PricewaterhouseCoopers is expecting IoT market volume to reach 1.2 trillion dollars by 2020. The hype is coming from great opportunities of the technology and its wide sphere of application. Growing amount of interest and constantly increasing number of implementations leads to the sensor’s amount growth and thus higher interconnection rate, leading to data transfer and storage volumes to build up. Enterprises tend to use this data for analysis and process optimization. With introduction of machine learning and deep learning, businesses are becoming more flexible as quick reaction to context changes is possible, thus lowering costs and maximizing return. IoT implementation is capable of both automation, which leads to processes optimization, and data gathering, which enables further analysis and return to the actor. However, the majority are struggling use the opportunity. Some do not see the value in metadata, others are not coping with the data amount requirements. What actually prevents enterprises from getting the full return from the technology are limitations on the sensor’s resources, heterogenous data, making data mapping without resources being drained impossible. The goal of this research is to offer an adaptive method for dynamic data acquisition, capable of gathering data with low-latency and high quality.