Sensors Converge 2021

Miniaturized sensors allow for data collection in a wealth of new use cases. Challenges arise in terms of power budgeting, design and assembly of miniaturized devices. In this talk we present a case study in designing a fingertip game controller based on a Bluetooth-enabled IMU. We present a power analysis of the sensor in different operation modes, the relationship between power requirements and battery characteristics, and the assembly of such a small form-factor device.

The push for greater intelligence has paved the way for machine learning at the edge of portable, battery-powered devices. While enabling greater privacy and performance, such devices are power-inefficient because they’re designed to digitize a vast amount of (mostly irrelevant) sensor data. To extend battery life, designers can choose more power-efficient components, use analog for computation in digital chips, or select an analogML chip to classify data at the start of the signal chain.

This session will use a top-down approach to discuss how to derive a system architecture and design specifications from marketing requirements; how to ensure all technical challenges, feasibility, manufacturing cost and development time are addressed at all levels; how to design a POC and minimize potential technical roadblocks; and how to assess manufacturability and product cost including DFx. This session will also introduce some of the low power, energy harvesting, wireless IoT and industrial sensor and system design techniques.

Energy harvesting is an excellent power solution for a wide variety of low power electronics and IoT devices. Well suited applications rely on a balance between available power, device functionality, and cost. These parameters aren't always easy to quantify and seemingly small changes to functionality or use case can have significant power impact. This presentation will provide simple guidelines for what makes an application well suited for energy harvesting and ways to overcome common roadblocks. 

Whether billions or trillions of devices becomes more semantics depending on how one defines devices, but when one considers the ubiquitous deployment of sensors in everything from Smart Cities/Buildings to Internet of Things (IoT) / Industrial IoT (IIoT) to conditional/asset monitoring to wearables to industrial automatic, it is pretty easy to build the case for the number of sensor deployments to approach 1 T devices even in the short- to mid-term timeframe, say by 2025.