Impact of emerging memory technologies on the energy efficiency of wireless sensor nodes

Abstract

Nowadays, battery-powered portable electronics are gaining more importance in our everyday use from smart phones, to Wireless Sensor Nodes (WSNs). WSNs are used in variety of applications in healthcare, agricultural, smart structural monitoring and so on. Increase demand on such devices for higher functionality, smaller size, lower cost and near perpetual operation are posing big challenges for system designers. Hence, the system's lifetime is a major concern and is determined by the finite energy source supplied by the batteries. Battery lifetime is impacted by the higher power consumption incurred by the added functionalities and increase leakage due to technology scaling. Unfortunately, battery energy density improvement at a rate of [7% - 10%] is not sufficient to help extend the sensor node's operational lifetime. In addition, enabling smaller scale WSN puts more restrictions on the available volume for energy. Hence, the WSN's lifetime is getting shorter. The effect of memory technology used in WSNs has not been fully exploited not only the memory contribution to the overall power consumptions but also the usage of memory as part of the power management unit. This motivates us to focus our research on emerging nonvolatile memory (NVM) technologies. Their non-volatility nature will suppress the power wasted as leakage in SRAM during idle periods which is typical for low duty-cycle WSNs. Thus, there is a need to investigate and study some of the emerging memory technologies and their suitability in WSN applications for the purpose of prolonging their operational lifetime. In this work, a study of two emerging memory technologies, Memristor and STT-RAM, as potential memory candidates for SRAM replacement. In our analysis for WSN applications, Memristor and STT-RAM total consumed power is compared to traditional 6T-SRAM-based memory using physics-based MATLAB models. Simulation results shows potential saving of 87% and 77% reduction in power for memristor and STT-RAM respectively as compared to conventional 6T-SRAM at 1% duty cycle. Memory was used to further exploit the energy efficiency at the sensor node system-level. A detailed analysis using Semi-Markov model to investigate different operational modes of WSN with the present of SRAM shows an improvement of 2x in the node's lifetime. In addition, a NVM is also explored to further improve the WSN energy efficiency. The conclusion of this work shows that utilizing an on-chip NVM can further improve WSN lifetime by 1x for low activity μW range sensor nodes.

Date of Award	2014
Original language	American English
Supervisor	Baker Saleh (Supervisor)