Design techniques to enhance low-power wireless communication soc with reconfigurability and wake up radio

Abstract

Nowadays, Internet of things applications are increasing, and each end-node has more demanding requirements such as energy efficiency and speed. The thesis proposes a heterogeneous elaboration unit for smart power applications, that consists of an ultra-low-power microcontroller coupled with a small (around 1k equivalent gates) soft-core of embedded FPGA. This digital system is implemented in 90-nm BCD technology of STMicroelectronics, and through the analysis presented in this thesis proves to have good performance in terms of power consumption and latency. The idea is to increase the system performance exploiting the embedded FPGA to managing smart power tasks. For the intended applications, a remarkable computational load is not required, it is just required the implementation of simple finite state machines, since they are event-driven applications. In this way, while the microcontroller deals with other system computations such as high-level communications, the eFPGA can efficiently manage smart power applications. An added value of the proposed elaboration unit is that a soft-core approach is applied to the whole digital system including the eFPGA, and hence, it is portable to different technologies. On the other hand, the configurability improvement has a straightforward drawback of about a 20–27% area overhead. The eFPGA usage to manage smart power applications, allows the system to reduce the required energy per task from about 400 to around 800 times compared to a processor implementation. The eFPGA utilization improves also the latency performance of the system reaching from 8 to 145 times less latency in terms of clock cycles. The thesis also introduces the architecture of a nano-watt wake-up radio integrated circuit implemented in 90-nm BCD technology of STMicroelectronics. The wake-up radio is an auxiliary always-on radio for medium-range applications that allows the IoT end-nodes to drastically reduce the power consumption during the node idle-listening communication phase.