Automation of Dynamic Power Management in FPGA-Based Energy-Constrained Systems

Abstract

The era of the Internet of Things comes with a huge number of interconnected communicating devices, which are often rather limited on the energy supply (e.g. battery powered or energy harvesting). Therefore, the pressure on energy efficiency of their operation (influencing lifetime or amount of functions) is especially crucial. In spite of a growing number of IoT devices, there still are many applications that are very specific and their market is quite limited. This is where the FPGAs offer a good alternative to dedicated application-specific chips, which would be too costly for such a purpose. Therefore, we target the powermanagement automation that simplifies energy-efficient design for FPGA platforms. A designer is then able to specify just an abstract power management in the commonly used SystemC model and it is automatically transformed in the more-complex form acceptable by a specific FPGA device. The proposed simplification and automation shortens the time-to-market of energy-efficient IoT products and prevents possible humanerrors that could be otherwise introduced to the design. The alleviated verification and debugging spare even more time in the development process. The experiments have proved the benefits of the proposed automation method. INDEX TERMS Design automation, energy efficiency, FPGA design, Internet of Things, power management.

Digital Electronic System-on-Chip Design: Methodologies, Tools, Evolution, and Trends

“…In the evolution of design methods and tools, the flexibility and choice of modeling styles offered by HDLs became insufficient at some stage. C-Synthesizer and System-C are examples of more efficient system design environments/compilers, which allow high-level language modeling/simulation/design of electronics/mechatronics systems and electronic controller system-on-chip implementation, with power efficiencies, especially relevant to Internet of Things (IoT) applications with microcontrollers and FPGAs often as popular targets [12,13]. Hardware/software co-design is also facilitated [14] in a holistic approach.…”