Published on February 17, 2014
Smart Sensor Architecture Smart Sensor System Architecture in Mimosa project Iiro Jantunen / Nokia Research Center Tutorial Ambient Intelligence, Toulouse, March 10, 2006 1 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Contents •What is a smart sensor system? • Transducer, analog electronics, ADC , digital electronics •O pen architecture for smart sensors • BluLite • Public S S I protocol • nanoUDP/nanoIP networking • O pen API:s for developing applications for using smart sensors •Mimosa architecture for smart sensors • Wired & wireless BluLite or R F ID connection • Many sensors in many devices over many radio technologies 2 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
From sensors to smart sensors Passive sensor (photodiode) S ensor Active sensor (S T 3-axis accelerometer) Wireless smart sensor 3 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Smart Sensor Systems • C ombine functions of sensors and interfaces • S ensing • Amplification • S ignal conditioning • AD conversion • Bus interfacing • Include higher level functions • S elf-testing • Auto-calibration • Data processing and evaluation • C ontext awareness • C ommunications • Modularity and/or integration 4 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ • Transducer • physical reality → electrical measurable • Amplifier & filtering • sensor impedance, signal strength and quality • Analog-digital conversion • Microcontroller / DS P / AS IC • Digital signal processing • C ommunications to outside world • Networking • S erial interface (US B, S PI, MMC ) • R adio interface (optional)
Transducer • C onverts between physical properties Measurement Typical/common techniques • The resulting property can be Acceleration Piezoelectric, capacitive Displacement, position, proximity R eluctance, optoelectronic, ultrasonic, radar • Voltage F low Pressure difference • O ptical power F orce Piezoresistive • E lectrical measurand needs an amplifier/filtering circuit to provide electrical power, impedance etc. Humidity R esistive, capacitive Location G PS Time C lock signal • O ptical measurand needs a optoelectrical transducer with the same properties S ound, pressure C apacitive R adiation O ptoelectronic G as concentration Tuned laser & optoelectronic • E lectrical capacitance • C urrent 5 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Non-ideal behavior of sensors C haracteristic S ensor Design Nonlinearity C onsistent R educe Drift Minimize C ompensate O ffset S ensor Interface C alibrate Time dependence of offset Minimize C alibrate/reduce Auto-zero Time dependence of sensitivity Nonrepeatability MC U/DS P Auto-range R educe C ross-sensitivity to temp and strain C alibrate Hysteresis Predictable Low resolution Increase Amplify Low sensibility Increase S tore value and correct Amplify Unsuitable output impedance Buffer S elf-heating Increase cooling Unsuitable frequency response Modify R educe power use F ilter Temperature dependence of offset S tore value and correct Temp. dependence of sensitivity S tore value and correct F rom R . F rank: Understanding S mart S ensors, 2n ed., 6 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Amplifying and analog filtering of sensor output • Amplifier provides signal strength and sensor impedance Issues: • F iltering needed to suppress noise • S ignal transmission • If needed, quite complex functions can be done with analog electronics, but many functions are better done in digital electronics • Data display • S ignal conditioning • O perating life • C alibration • easier • Impedance of sensor and system • cheaper • S upply voltage • need less room • F requency response • F iltering 7 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Analog-digital conversion (ADC) • The measurement must usually be changed to digital form for • further processing and calculations • storing to memory • sending the data • or just an economical reason • ADC often included in MC U:s, e.g. MS P430F 1xx • Is the resolution (8 or 12-bit) enough? 8 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ Issues: • S ample rate • R esolution (8, 12, 16-bit) • Accuracy • Power consumption • Price
Digital signal processing • More complex signal processing • summing many sensors (e.g. 3-axis accelerometers) • pattern recognition, e.g. step counters, speech recognition • C an be done in Issues: • F ixed-point vs. floating-point DS P • Data precision • S peed • Power usage • AS IC , for cheap mass-production • Price • DS P, for high-speed number crunching • Internal ADC • MC U (microcontroller), medium level calculations, control software 9 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ • Need for flexible programming: MC U vs. DS P • Programming languages: Assembly vs. C /C ++ or J ava
Microcontroller • S oftware controlling a smart sensor system, e.g. • measurements • communications • data memory • real time clock • Usually includes • ADC (DAC ) • timers • serial ports (S PI, I2C , UAR T) • flash memory • Power-efficient, cheap, flexible • Also called MC U or µC 10 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ MS P430F169 by Texas Instruments • Low supply-voltage 1.8 - 3.6 V • Ultra-low power consumption: • Active: 330 µA at 1 MHz, 2.2 V • S tandby: 1.1 µA • O ff (R AM retention): 0.2 µA • Wake-up from standby in less than 6 µs • 16-Bit R IS C , 125-ns instruction cycle • Program memory 60 kB (flash), R AM 2048 B • 8-channel 12-Bit ADC with internal reference, sample-and-hold and autoscan • Dual channel 12-Bit DAC with synchronization • 16-Bit Timer_ A & Timer_ B with 3/7 C C R • O n-chip comparator • 2 serial interfaces: UAR T or S PI or I2C ™ • S upply voltage supervisor/monitor • Brownout detector • Bootstrap loader • S erial onboard programming
Networking sensors over radio RF ID Internet WL AN L ow E nd Bluetooth UWB 11 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ C ellular GPRS/3G network
BluLite (or Bluetooth Low End Extension) • Low E nd E xtension for Bluetooth is designed • to complement Bluetooth by creating a wireless solution that • allows small devices that are limited in battery power, size, weight and cost to have a wireless connection with mobile terminals • without adding yet another radio to mobile terminals (as Z igBee) • O ptimized for irregular data exchange between Bluetooth enabled mobile terminals and button cell batter powered small devices • C oncept assumes two device classes • Dual-mode (Bluetooth 1.2 with Low E nd mode) for terminals • S tand-alone (Low E nd mode alone) for sensors and enhancements Dual-mode device (i.e. terminal) E /(BB) C ommon LE MAC C ommon Host Bluetooth RF IF BB/MAC 12 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ Stand-alone device (e.g. 3D accelometer node) S ensor AS IC LE E AS IC Stand-alone device (e.g. fitness gadget) S ensor AS IC LE E AS IC
BluLite radio Channels of the proposed system 2401 2400 IEEE 802.11b channel in North America and Europe 2402 2403 Bluetooth channels 2480 2481 2482 2483 MHz Low End Extension channels 2403 2406 2433 2436 2463 2466 2478 2481 MHz One default initialization channel, non-overlapping with Bluetooth T secondary initialization channels, for jamming resistance wo 24 Unicast channels for user data 13 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
BluLite radio parameters with 1 Mbps •Physical bit rate 1 Mbps •F requency band 2.4 G Hz (IS M band) •Duplex TDD •C o-existence of multiple devices • C onnection setup channel C S MA • Data delivery F DMA •J amming avoidance F DMA •PDU payload • Byte aligned, variable length, max 255 bytes •Bit rate excluding PHY and MAC overheads • Uni-directional with AR Q , max 890 kbps • Bi-directional with AR Q , max 2 x 471 kbps 14 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Basic functions of BluLite radio ADVERTISE OFF IDLE SCAN CONNECTED CONNECT 1. 2. 3. 4. 15 ADVE RTIS E : Makes the local device visible and connectable to all remote devices within reach. Low-power protocols optimized for this state. A possibility for an application dependent trade-off between connection set-up delay and the power consumption. S C AN: R eturns the addresses and short description of the advertising remote devises within reach. C ONNE C T: E stablishes a point-to-point connection with an advertising remote device. C ONNE C TE D: Provides point-to-point bi-directional data delivery with error detection, AR Q , segmentation, role switch and a low-activity mode. An evolution path to point-tomultipoint, requiring additions only in master capable devices. © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Mimosa Architecture Radio sensor node Back-end server RFID Sensor MSP430 4. LEE ASIC SSI server Sensors nanoIP Internet FPGA Digital sensor management (memory map) D M-SPI UART SPI/IIC A BT LEE RFID (front end) UMTS G PR S M-SPI 1. 3. 1. Host (N6630) Application space M-API SSI SSI (Client) 5. RF-module 1. (virtual Server) 4. LOCOS 2. SEMBO MSP430 Communication Layer (U-ULIF) SSI server Symbian MCU SPI ADC Device drivers CART FPGA UART SPI/IIC D A DAC Radio BBs POP SPI 3. Sensor-HW 16 © 2006 Nokia Device drivers ULIF S mart S ens or Architecture.ppt / 2006-03-10 / IJ 2. Sensor-HW
Simple Sensor Interface (SSI) protocol • A simple protocol for reading smart sensors over, e.g., BT LE E • Also provided for R F ID sensor tags • Memory map of sensor data on • C ompatible with IS O 18000-4 • S upport for multiple sensors on multiple devices • S upport for data polling or streaming • Developed in co-operation with S uunto, Vaisala, Mermit, C E A-LE TI, O ulu University and Ionific • Development of the specification keeps backward compatibility (v0.4-) • F or specifications, source code and discussion forum, visit http://ssi-protocol.net 17 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ • C lient-server architecture • Terminal has client software • S ensor unit has server software • C lient can • poll sensors • ask the server (O bserver) to stream data to client (Listener) • read and modify the configuration of the server • C ommands have two forms • C apital letter (“N”), no checksum • Low case (“n”), the payload has a C R C checksum
SSI v1.0 command base C ommand Dir. Description Q /q 0x51 / 0x71 → Q uery A /a 0x41 / 0x61 ← Q uery reply C /c 0x43 / 0x63 → Discover sensors N /n 0x4E / 0x6E ← Discovery reply Z /z 0x5A / 0x7A → R eset S S I device G /g 0x47 / 0x67 → G et configuration data for a sensor X /x 0x58 / 0x78 ← C onfiguration data response S /s 0x53 / 0x73 → S et configuration data for a sensor R /r 0x52 / 0x72 → R equest sensor data V /v 0x56 / 0x76 ← S ensor data response D /d 0x44 / 0x64 ← S ensor data response with one byte status field O /o 0x4F / 0x6F → C reate sensor observer Y /y 0x59 / 0x79 ← S ensor observer created K /k 0x4B / 0x6B → Delete sensor observer L /l 0x4C / 0x6C ← R equest sensor listener J /j 0x4A / 0x6A → S ensor listener created E rror E /e 0x45 / 0x65 ⇮ E rror messages F ree data F /f 0x46 / 0x66 ⇮ F ree data for custom purposes S ensor discovery C onfiguration Read data S treaming data 18 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Example SSI command: Discovery reply • S ensor device answers to Discovery command sent by a mobile terminal C • Discovery R eply message contains information about one or more sensors. N (sensor info ) Client Server N (sensor info ) Terminal • The device answers with its address, command name (N or n) and sensor data Sensor unit . . . • E ach sensor is identified with • The reply can carry either one sensor per N command or many depending on buffer size • S ensor Id – 2bytes • Description – 16 byte AS C II • Unit – 8 byte AS C II • UAR T or nanoIP frame provides the information about the length of a single message • Type – 1 byte • S caler - signed 1 byte • Min – minimum sensor reading value • Max – maximum sensor reading value 1 Addr 19 1 N/n © 2006 Nokia 2 Sensor Id 1 16 Sensor desc. S mart S ens or Architecture.ppt / 2006-03-10 / IJ 8 Unit 1 1 Type Scaler 4 4 MIn Max . . .
nanoIP networking • An open-source networking architecture • Minimal overheads • Wireless networking SSI • Local addressing • NanoIP makes use of the MAC address of underlying network technology rather than IP addresses • Used with nanoUDP (User Datagram Protocol) or nanoTC P (Transmission C ontrol Protocol) • nanoUDP does not provide reliability or ordering • nanoTC P provides retransmissions and flow control • Mimosa uses nanoUDP • http://www.cwc.oulu.fi/nanoip/ 20 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ nanoUDP nanoIP ULIF (BluLite MAC )
Open source SSI/nanoIP implementation • S S I v0.4 over nanoUDP/nanoIP provided by O ulu University • 1 byte protocol type (nanoIP) • 4 bytes nanoUDP header • S S I v1.0 (nanoUDP/nanoIP) being finalized by Nokia R esearch C enter, will be open source • Works over BluLite (Bluetooth LE E ) • 2 byte payload + C R C length • 1 byte S ource port (40 for S S I) • 1 byte Destination port (40 for S S I) • n bytes • S S I payload • 2 bytes • optional C R C checksum 1 1 1 1 Prtcol 21 1 LenH LenL Source Dest. © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ n Payload 2 CRC
Mimosa API’s for 3rd parties Back-end server IP Local MIMO S A S W Applications UI_ API C ontext_ API J ava & C ++ implementation MIMO S A Ambient User Interface Layer MIMO S A C ontext Awareness Layer S ensor_ API MIMO S A S ensor Layer LC _ API MIMO S A Local C onnectivity Layer 22 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Sensor Management Board • C ontrols the sensors (host) • C ontrols the sensors and BTLE E (S ensor R adio Node) • R eal time clock (with its own battery) • C onnectors for • LO C O S and/or C AR T • S ensor board (UAR T, S PI/I2C , general digital I/O , 8-channel analog) • J TAG programming interface • Debugging ports (UAR T) • IR Q ports • MC U runs S S I/nanoIP and device drivers for sensors 23 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Sensor Management Board 24 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ 3.3 V POWER UART 32.768 kHz clock UART I2C SPI ADC DAC 32.768 kHz crystal IRQs USART0 I2C MCU text MSP430F169 JTAG M-SPI LOCOS and / or CART UART 8 MHz crystal CART USART1 External power • LO C O S and/or C AR T • S ensor board (UAR T, S PI/I2C , general digital I/O , 8-channel analog) • J TAG programming interface • Debugging ports (UAR T) • IR Q ports • MC U runs S S I/nanoIP and device drivers for sensors SEMBO Debugging: SPI0 + UART1 • C ontrols the sensors (host) • C ontrols the sensors and BTLE E (S ensor R adio Node) • R eal time clock (with its own battery) • C onnectors for 32.768 kHz crystal 3.3 V POWER 3V BAT ANALOG Real time clock DS1305E SEPPO / SENSOR BOARD
Mimosa terminal • Nokia 6630 phone as terminal • Attached electronics (C AR T, LO C O S , S E MBO , R F ) provide the Mimosa hardware functionality • J ava software on N6630 will provide the user interface, context awareness, sensor management etc. • LO C O S board as motherboard of attached electronics • S E MBO provides local sensor control • S E PPO (or other) sensor board connected to S E MBO with a standard interface • R F part controlled with F PG A on LO C O S on baseband-module • S eparate R F module 25 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ RF-module LOCOS SEMBO MSP430 ADC SSI server DAC CART Device drivers MCU SPI UART SPI/IIC FPGA D A Radio BBs SPI Sensor-HW
Mimosa terminal - 2 SEPPO Connection to RF board SEMBO LOCOS CART Connection to phone 26 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Sensor Radio Node • S mart sensor which can be read over BT LE E connection • E MMI board provides Bluetooth LE E for communications with mobile terminals • S E MBO for controlling BT LE E , running the S S I server and sensor drivers LEE ASIC SEMBO MSP430 EMMI SSI server Device drivers FPGA M-SPI UART SPI BT LEE • LO C O S acts as motherboard • S E PPO (or other) as sensor hardware board via standard connector • UAR T • S PI / I2C • 8 analog channels • 15 unspecified digital I/O pins 27 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ M-SPI Sensor-HW LOCOS D A
Sensor Radio Node - 2 SEPPO EMMI 28 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ LOCOS SEMBO
Sensors demonstrated on Mimosa platform S ensor Provider Notes Humidity, temperature, pressure Digital Nokia, (S ensirion, Intersema) Weather station, S S I demonstration F at % Amplified analog Nokia F itness & health EC G Amplified analog C ardiplus F itness & health, streaming data over S S I/nanoIP/BluLite Lactate, glucose Amplified analog F raunhofer-IS IT, C ardiplus, Åmic F itness & health G yroscope 29 Interface I2C F raunhofer-IS IT, C E ALE TI, S TMicroelectronics © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Weather station • Demonstration of S S I use • Weather information: Temperature, Humidity / dew point, and Pressure • Intersema MS 5534A pressure sensor • Pressure range 300-1100 mbar • Internal 15 Bit ADC • 6 coeff. software compensation on-chip • 3-wire serial interface • 1 system clock line (32.768 kHz) • 35 ms measurement time • S ensirion S HT11 humidity sensor 30 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Weather station • Demonstration of S S I use • Weather information: Temperature, Humidity / dew point, and Pressure • Intersema MS 5534A pressure sensor • S ensirion S HT11 humidity sensor • Internal 14-bit ADC • F ully calibrated • Internal power regulation • R ange • Humidity 0 – 100 % • Temperature -40 – 128°C • 2-wire serial interface (not I2C ) • 11/55/210 ms for a 8/12/14bit measurement. 31 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
Fat percentage • Actually measures body water content (TBW) • Amplified analog output • Low-noise O PA2350 two-channel • Also used for 2-point measurement of surface impedance, G alvanic S kin R esponse (G S R ) • S kin stratum corneum humidity • S tress measurement 32 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ C50KHZ V_FAT ADC • 4-point measurement of impedance with 50 kHz signal +3.3V SEMBO INTERFACE • F rom TBW, fat-% and dehydration (∆TBW) calculated FAT_CS I_FAT REF_FAT GND ANALOG INTERFACE ELECTRONICS SIGNAL IN M1 M2 SIGNAL OUT E1 E2 E3 E4
Lactate & glucose sensors • F raunhofer-Institut für S iziliziumtechnologie (Itzehoe, G ermany) makes the sensor • Åmic (Uppsala, S weden) makes a needle array to penetrate outer layer of skin • Measures the lactate or glucose content of interstitial fluid • Lactate and glucose sensors differ only on the enzyme used • Amplified analog connection to S E MBO 33 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
RFID sensor tag •V irtual server on the terminal device answers to the S ensor API when connecting to R F ID sensors •V irtual server translates the commands to the R F ID reader to read a specified memory area of the tag •0x53 (for AS C II “S ”) in address 0x0C (Tag memory layout) defines the memory layout to be S S I compliant •Memory mapping designed for one or more sensors per tag •Memory layout designed to be convenient for S S I use Bytes Field 8 0x00 – 0x07 Tag ID 4 0x08 – 0x09 Tag manufacturer 0x0A – 0x0B Tag hardware type 0x0C – 0x11 Tag memory layout. This defines tag to be S S I compliant sensor. 0x12 – User data (layout defined by S S I). 6 34 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
RFID memory mapping: Sensor data space Byte address Field name Type Description Example 0x12 – 0x15 Sensor value 4 byte HEX Variable sensor value 0x00000014 0x16 Type 1 b HEX Describes the type of the Sensor value (0x00 = float, 0x01 = signed integer…) 0x01 0x17 Multiplier 1 b HEX 0x18 Status 1 b HEX Sensor status. Bits 0 – 7 indicate if the sensor values are valid data (bit = 1) or not yet valid (bit = 0). 0x19 Empty 1b For future use. Could be, e.g., number of sensors. 16 0x1A – 0x29 Sensor description 16 b ASCII Constant sensor description “Temperature” 8 0x2A – 0x31 Unit 8 b ASCII Constant unit description. “C” 0x32 – 0x35 Minimum value 4 b HEX Minimum value the sensor can provide. 0x00000000 0x36 – 0x39 Maximum value 4 b HEX Maximum value the sensor can provide. 0x00000064 0x3A Activate sensor 1b Bits 0 – 7 will be written by the reader to request the tag to write the sensor data 0x01 0x3B – 0x3D Sensor control 3b Not defined yet. Could be, e.g., time needed before sensor value is valid. n bytes Optional configuration data or 2nd sensor 8 8 4 n 0x3E – 35 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ 0x00 0x01
Questions & Answers 36 © 2006 Nokia S mart S ens or Architecture.ppt / 2006-03-10 / IJ
IEEE Xplore. Delivering full text ... A Smart Sensor Architecture for Marine Sensor Networks Full ... 16-18 July 2006 Page(s): 93 Print ISBN:
Sensor networks for environmental data acquisition will play an important role in the development of future large data acquisition systems, particularly ...
A Smart Sensor Architecture for Marine Sensor Networks J. Sorribas Unitat de Tecnologia Marina. Passeig Marítim 37-49. 08003 Barcelona. Spain
Smart Dust: Sensor Network Applications, Architecture and Design. Mohammad Ilyas, Imad Mahgoub
Smart Dust: Sensor Network Applications, Architecture and Design book description and order
5 5. General Architecture of smart sensor: One can easily propose a general architecture of smart sensor from its definition, functions. From the
Computer Architecture Andreas Loos. Search. Search: ... 2006, pp. 374-379. ... A combined space-time multiplex architecture for a stacked smart sensor chip ...
Smart Sensor Networks The ... and control messages so it may put constraints on the network architecture. Querying ... 2006 Bluetronix ...
IEEE Xplore. Delivering full text ... This paper presents a vision system based on a smart sensor. ... 10.1109/ISIE.2006.296069 Publisher: IEEE ...
Flexible smart sensor framework for autonomous structural health monitoring429 final user-specified sampling frequency. This combination of filters ...