Development of a Sensor Density Reduction Methodology for Thermal Monitoring of Electronic Systems

This paper presents a preliminary methodology for identifying and localizing anomalous heat sources on 2D surfaces in electronic systems using sparse sensor networks. As power densities in modern electronics continue to rise, effective thermal monitoring has become essential for preventing failures and ensuring long-term reliability. However, dense sensor deployment is often impractical due to cost constraints and spatial limitations. The proposed approach leverages heat-diffusion physics to predict the expected response of the sensor network to anomalies occurring at predetermined locations of interest. In this way, the computationally intensive simulation stage is decoupled from the real-time identification process, substantially reducing the monitoring system's computational burden. Experimental validation shows that, even with a 50 -fold reduction in sensor count (a 98 % decrease), the worst-case localization error increases only by a factor of 25 . By enabling more cost-effective monitoring network deployments, the methodology supports the acquisition of detailed information about component operating conditions without interfering with primary system functions.

Keywords: PCB thermal monitoring, heat source localization, sparse sensor networks, electronic systems reliability.