Around the globe, manufacturers, energy and utility companies are actively seeking to capitalize on the promise of big data, by transforming massive datasets into newfound knowledge that will enable timely data-driven decision-making and lead to new business opportunities. At present, the main blocker in the widespread use of data driven methodologies is the complexity of the software tools and the lack of access to data scientists capable of deploying these tools. In addition to the numerous challenges that we face, namely, the sheer magnitude of available data, the speed of newly available data (e.g., produced by sensors), the variety of data (i.e., represented in diverse formats), and data veracity (i.e., trustworthiness), one of the challenges that requires greater attention and investment is providing domain experts with appropriate data science tools and training to address their data driven business needs.
In MILAB, we focus on industrial, telecommunications, and other time series data typically collected from sensors during the manufacturing or maintenance process. Time series data has its own challenges and majority of the existing software tools do not handle time series data, or provide only partial support. The methodologies to be used will be able to handle composites of time series and a mix of numeric and categorical metadata that describe processes, structures and hierarchies typical of the IoT devices.
As a specific area, the latest mobile technologies enable high granularity real-time reporting of all conditions of individual sessions, which gives rise to use data analytics methods to process and monetize this data for network optimization. We participate in planning for ML/AI systems for B5G/6G radio.