SENSORS FOR IOT ROBERT VAN SCOY EMBEDDED TECH CON, JUNE 10, 2015 EVOLUTION OF SENSORS » Early 20th Century – the age of the clipboard and pencil Some sensors are very old -- Seebeck observed the thermocouple effect in 1821 Early sensor readouts use analog meters and gauges “Data acquisition” involves reading the meters and writing the numbers down Data analysis is done with pencil, paper and slide rule 8-Jun-15 Photo credit: Jimi Jones Possibilities start here // 2 EVOLUTION OF SENSORS » Mid 20th Century – automation arrives Electronic amplifiers let sensors output useful levels of voltage or current Analog process controllers allow sophisticated automation of control tasks “Data acquisition” typically involves reading paper graphs from chart recorders Data analysis is still mostly manual 8-Jun-15 Photo credit: Trojan Nuclear Power Plant Possibilities start here // 3 EVOLUTION OF SENSORS » Late 20th Century – computerization and MEMS Analog sensor outputs are digitized for processing by computer systems IC technology enables direct digital outputs from many sensor types Digital data is immediately available for local or remote computer analysis MEMS technology revolutionizes the cost and size of many “physical” sensors 8-Jun-15 Photo credit: Invensys Process Systems Possibilities start here // 4 THE IC/MEMS COST REVOLUTION » IC technology has made some sensors nearly free An entire data acquisition chain -- Sensor + amplifier + filter + A/D + signal processing – can be integrated into a single IC with a digital output One example -- digital temperature sensors now cost <$0.10 in quantity, or even come free as part of other chips Sensors for light, sound, magnetic fields, etc. have also become very low cost. Precision location sensing is now a “$5 problem” for IC-based GPS receivers. 8-Jun-15 Credit : Texas Instruments Possibilities start here // 5 THE IC/MEMS COST REVOLUTION » Some “hard” sensing problems have become easy MEMS (MicroElectricalMechanicalSystems) are miniature machines with movable parts that are built with integrated circuit fabrication technology MEMS technology has created accelerometers and gyroscopes that cost a few $, and also lowered the cost of traditional functions such as pressure sensors 8-Jun-15 Credits: Institute of Navigation, ST Possibilities start here // 6 THE IC/MEMS COST REVOLUTION » Since sensors are so cheap, sensors are spreading everywhere There are 12+ sensors in a high-end smartphone, dozens in a car There is a constant stream of new sensors and new sensor-enabled devices Much of the promise of IOT is the promise of data from “cheap sensors everywhere” 8-Jun-15 Credit: IHS Possibilities start here // 7 WHAT CAN SENSORS SENSE? » Environmental Temperature, humidity, altitude, location, light, images, magnetic fields, gases » Physical Acceleration, force, vibration, pressure, strain, sound » Electrical Voltage, current, power, phase, RF » Physiological Heartrate, BP, EKG, Blood O2, etc 8-Jun-15 Credits: Analog Devices, Intel Possibilities start here // 8 WHAT DO SENSORS SENSE? » Baseline data is important Often its hard to know what “normal” is unless we do long-term data monitoring Example: vibrations as traffic passes over a bridge Trend towards 24/7 monitoring, so low power sensors are often important » Sensors sense more than the obvious Obviously webcams need security, but - Room temperature changes can give away comings and goings Electrical load changes can reveal what equipment is being used and when Even the most “boring” sensor data may raise security and privacy concerns IOT acceptance (and company reputations) depend on public confidence that data is secure and not being misused 8-Jun-15 Possibilities start here // 9 SENSORS IN COMPUTING DEVICES » Mobile devices have lots of sensors Cameras! RF: GPS, 3G, WiFi, Bluetooth, NFC Accelerometer, gyro, magnetometer And sensors to make the UI work . . . » Sensor interfaces are usually simple I2C, SPI interfaces are common for low-bandwidth sensors Often sensors have low-power modes where they can still generate interrupts Some designs have dedicated “sensor hub” processors for low-power monitoring Image sensors typically use mobile-centric interfaces such as MIPI CSI-2 8-Jun-15 Credit: Smartphoneworld.me Possibilities start here // 10 SENSORS IN COMPUTING DEVICES » Embedded computers typically have “Hardware Monitor” functions that: Measure CPU die temperature Measure temperatures at other points Control system fan or alarms (with varying amounts of software intervention) Monitor important system power supply voltages Measure the voltage of the CMOS battery under load (for life prediction) Monitor external battery charge state and/or control battery charging » Kontron and other vendors provide APIs to access internal sensor data Data is very useful for system thermal validation Many customers don’t take full advantage of operational data for PM, fault monitoring Module/carrier board designs allow adding sensors that make sense for application It’s likely more sensors will be added as usefulness is understood – e.g. logging of shock handling history when the computer controls delicate equipment 8-Jun-15 Credit: Kontron Possibilities start here // 11 EXTERNAL SENSOR INTERFACES » Historically sensor interfaces were analog Many different analog “standards”: raw transducer outputs, 0-10V, 4-20mA, etc Sensors had point to point wiring to the host system Long cables needed in industrial settings, subject to noise pickup & other issues 8-Jun-15 Credit: Emerson Global Users Exchange Possibilities start here // 12 EXTERNAL SENSOR INTERFACES » Digital interfaces have proliferated in recent decades Serial “busses” like RS-422, RS-485, CAN, MODBUS, Fieldbus are commonly used for connecting sensors to host computers Typically these busses are multidrop and addressable – allow connection of many sensors to one bus and so eliminate a lot of wiring Ethernet variants such as PROFINET and EtherCAT are gaining popularity Digital sensors are often “smart” (e.g. run own self-calibration routines) Many vendors provide “transmitters” that interface analog sensors to busses 8-Jun-15 Credit: Kontron, wut.de Possibilities start here // 13 EXTERNAL SENSOR INTERFACES » RF links gaining traction now Sensors can connect to hosts or each other via WiFi, Bluetooth, ZigBee, etc Mesh topologies have reliability advantages in large sensor networks Sensor and radio power consumption is sometimes critical – some applications need to run for years from a small battery 8-Jun-15 Credit: Kontron, EMWhiz Technologies Possibilities start here // 14 SOME EXAMPLES » Many industrial PC customers already implement some form of “IOT” System controller often also provides a gateway for sensor data Remote monitoring, PM, problem diagnosis, customer support all have high value Hardware and software support from system vendors makes “industrial IOT” easier 8-Jun-15 Credit: Kontron Possibilities start here // 15 SOME EXAMPLES » New applications are based on ubiquitous / dispersed / mobile sensors Often there’s value in 24/7 baseline monitoring – but this generates lots of data Wired network uplinks often not available, need to use 3G or other radios. Power consumption is critical when line power is not available Gateways often needed to aggregate data onto a single uplink. Local data preprocessing can save communications costs, bandwidth, power 8-Jun-15 Credit: Decentlab Possibilities start here // 16 SOME EXAMPLES » Rugged mobile computers interface with truck sensors via CAN Wealth of data is available on system performance for PM “IOT is integrated with other functionality for e.g. scheduling, maps, driver logs Typically 3G or satellite uplink transmits data for real-time remote monitoring Some benefits are indirect (theft prevention, driver compliance with rules and laws) Economic value well proven – major Kontron customer has 250K+ units on road 8-Jun-15 Credit: Decentlab Possibilities start here // 17 PRECISION AGRICULTURE Joint partnership with Mimos, and UPM, funded by Malaysian Government Purpose: Increase crop yield Results: 3 yields in 1 year (normally only 1) Details 3 greenhouses, semi and fully controlled in west and east Malaysia Used sensors to monitor growing variables Data collected through Kontron gateway and tracked with Mimos software 8-Jun-15 Possibilities start here // 18 WIRELESS SENSING NETWORK (WSN) FOR PRECISION AGRICULTURE Sensor Node Temperature Sensor Node pH CLOUD COMPUTING AGRICULTURAL CORPORATION ANALYST Sensor Node Humidity Router Sensor Node Moisture Sensor Node Soil Conductivity 8-Jun-15 Gateway In Greenhouse / IP Rated Housing FRONTEND INTERFACE Possibilities start here // 19 WSN OVERALL SYSTEM ARCHITECTURE Sensor Node pH & EC Water & fertilizer Water tank Mixing & dosing tanks Valves Embedded Controller CAN Network Sensor Node Moisture Sensor Node Humidity Sensor Node pH & EC Sensor Node Temperature Zigbee Network SYSTEM OFF ON WSN Gateway Harvest!!! THANK YOU FOR YOUR ATTENTION www.kontron.com Augsburg | Deggendorf | Toulon | Copenhagen | Chichester | Brussels | Pilsen | Warsaw | San Diego | Fremont | Montreal | Beijing | Hong Kong | Taipei | Penang | Bangalore | Mumbai | Sydney | Tokyo
© Copyright 2024