Development and experimental verification of data-driven approaches to real-time energy monitoring and diagnosis of telecommunication sites

The paper presents a diagnostic tool aimed at improving energy conservation and thermal management within telecommunication sites. Malfunctions of data acquisition and electronic and cooling systems were analyzed. Fault tree analysis methodology was applied to develop the targeted fault signature matrix-based detection and isolation schemes. Specifically in this work, an automatic data-driven procedure, relying on physical and statistical knowledge, was conceived to get full and measurement-uncertainties-free energy characterization of telecommunication sites in non-faulty conditions, also distinguishing pure central offices from multi-use central offices (eg, combined TLC rooms and offices), with the introduction of a synthetic and easy-to-calculate threshold. According to this threshold, a TLC site is classified as a multi-use central office if the auxiliaries absorb more than 10% of the total energy consumption of the site. The developed tool allows for the valorization and processing of the available measurements, without any additional modification of the existing network. Experimental verification allowed confirming the applicability of proposed diagnostic procedure as a data-driven energy intelligence tool to simplify and lean maintenance operations. Main achievements were accomplished only relying on electricity measurements, thus positively influencing sensing costs and reliability. The proposed procedure contributed to achieving up to 11% yearly energy savings in high energy demanding central office without additional sensor costs, thus being strategic for implementing new regulatory standards, as well as reducing the impact of thermal management on telecommunication sites energy consumption.