Search

In this work we propose an optimal back plane biasing (OBB) scheme to be used in a UTBB FD SOI technology that minimizes the energy per operation consumption of sub threshold digital CMOS circuits. By using this OBB scheme, simulations show that more than 30% energy savings can be obtained with low threshold voltage (LVT) devices in comparison with classic symmetric back plane biasing (SBB) schemes. Additionally, this OBB scheme allows to adjust the performance of the circuit with very small energy penalties. A very simple and intuitive model, for sub threshold digital CMOS circuits, was developed to justify the benefits obtained by OBB. The results predicted by the model are confirmed with extensive simulation results. We show that the OBB approach can be applied easily to a given circuit just based on the information provided by a logic simulation of the circuit (or even an analysis of its structure) and simple electrical simulations of the pMOS and nMOS transistors. Finally, we show that the variability in the energy consumption is improved by using OBB and suggests that new sizing methodologies must be studied to fully benefit from the wide back plane voltage range available in UTBB FD SOI technology for the design of robust energy efficient digital circuits. • This work presents a novel back plane biasing scheme (OBB). • The OBB scheme reduces energy consumption of sub/near threshold digital designs. • Modeling the impact of NMOS and PMOS imbalance in energy per operation consumption. • We show there is an optimum imbalance that minimizes energy consumption. • OBB consists on using the back plane voltage to work in this optimum. • Circuits simulations show more than 30% energy reductions in the studied circuits. [ABSTRACT FROM AUTHOR]

To achieve energy efficiency in the Internet-of-Things (IoT), more intelligence is required in the wireless IoT nodes. Otherwise, the energy required by the wireless communication of raw sensor data will prohibit battery lifetime, the backbone of IoT. One option to achive this intelligence is to implement a variety of machine learning algorithms on the IoT sensor instead of the cloud. Shown here is sub-milliwatt machine learning accelerator operating at the Ultra-Low Voltage Minimum-Energy Point. The accelerator is a Transport Triggered Architecture (TTA) Application-Specific Instruction-Set Processor (ASIP) targeted for running various Machine Learning algorithms. The ASIP is implemented in 28 nm FDSOI (Fully Depleted Silicon On Insulator) CMOS process, with an operating voltage of 0.35 V, and is capable of 5.3pJ/cycle and 1.8nJ/iteration when performing conventional machine learning algorithms. The ASIP also includes hardware and compiler support for approximate computing. With the machine learning algorithms, computing approximately brings a maximum of 4.7% energy savings. [ABSTRACT FROM AUTHOR]