Our paper titled "Energy Aware Memory Mapping for Hybrid FRAM-SRAM MCUs for IoT Edge devices" won the Best Paper Award at the 29th International Conference on VLSI Design and 15th International Conference on Embedded Systems. Related news could be found here.
The paper talks about two techniques em-Map and Energy-Align that make intermittently powered IoT system energy-efficient. Intermittently powered IoT systems suffer from sudden and abrupt power failures, which makes it challenging to perform reliable computations on them. Em-Map is a dynamic memory mapping technique that maps functions constituting a program across SRAM and FRAM (Ferroelectric RAM), thereby optimizing energy-consumption and improving performance without sacrificing reliability. Energy-Align is a technique that reduces the overhead required to perform the necessary book-keeping that enable the IoT systems to compute across the power failures. More information could be found in the paper.