Performance Improvement and Energy Saving Based on Increasing Locality of Persistent Data in Embedded Systems

The performance of embedded systems often suffers from strict cost/size and power/energy limitations. This is especially important for very small systems that are sold in a high number of items like smart cards and RFID controllers. On one hand, small hardware platforms are developed to minimize cost and power consumption. On the other hand, applications have to fulfill strict performance demands. We introduce a high-level performance optimization for EEPROM-based embedded systems. Writing to persistent memory is known to be very costly in the meaning of time and energy consumption. Our proposal shows how to reduce the number of write operations. We introduce an approach to increase spatial data locality without losing reliability of data that has to be stored persistently.