Realisierungskonzepte für drahtlose Sensornetzwerke – Ein Überblick Stefan von der Mark, Georg Böck

Overview What are Wireless Sensor Networks? Similarities and differences to RFID Some published approaches PicoRadio/PicoBeacon (Berkeley) WiseNET (CSEM) MUSE and ORBIT (WINLAB) The AVM eGrain project Concept WakeUp Demonstrator

What are Sensor Networks? Smart Dust: University of Geneva in Switzerland

Applications Logistics, Locationing Environmental monitoring Goods in a warehouse or shopping center Books in a library Environmental monitoring Indoor: Temperature, humidity, intruders Outdoor: Pollution, agricultural research Structural monitoring Bridges, skyscrapers, large halls Ageing, stress from snow, earthquakes Military

Properties of Sensor Networks Tiny little low cost sensor nodes Wireless peer to peer communication Self-sustained operation for prolonged time Preferably completely integrated (CMOS) Ad Hoc Networking: Only some sensors communicate with base stations Data is routed through the sensors

State of the Art Existing sensor arrays are usually wired Classical: Analog wire from each sensor More modern: Digital bus systems Existing wireless sensors usually communicate with dedicated access points Sensor communication mostly proprietary, but IEEE standard 802.15.4/ZigBee exists New IEEE 1451.4 „plug&play“ standard for Sensor ID Type of measurement (Units!) Calibration data

Sensor Networks vs. RFID Transponder and interrogator All nodes equal Tags reply only on request of Interrogator All nodes can initiate transmission High transmission power available from interrogator Very low transmission power Transponder usually powered by incoming RF Own power source necessary

Similarities Low data rates Only occasional communication Receivers can be similar But transmission is completely different

Realisation approaches Nothing coming close to the vision has been realized so far Different approaches are being pursued: Big and power hungry, but functional nodes (for protocol develompment, application research) Demonstration of particular technologies (low power circuits, sensing, energy scavenging) Attempts towards complete low power hardware (with reduced functionality) And anything in between Kleine Auswahl, bei Weitem nicht vollständig

PicoNode I (UC Berkeley) „PicoRadio Project“ at Berkeley Wireless Research Center, University of California at Berkeley (UCB) Strong ARM CPU Xilinx FPGA Proxim RangeLAN or Bluetooth HW with own protocols 24 hr operation out of 2 x 1200mAh Li-Ion Variety of sensor boards (modular concept) January 2002

PicoBeacon (UCB) Energy scavenged from light and vibration 180 W out of 1 cm3 from vibrations 1.9 GHz transmission (no receiver) 10m range 2.4 x 3.9 cm2 Agilent Film Bulk Acoustic Resonator (FBAR) for 1.9 GHz Oscillator

WiseNet (CSEM) Swiss Center for Electronics and Microtechnology (CSEM) 2 mW RX, 32 mW TX 433 / 868 MHz ISM 25 kbps 25 W for 56 bytes every 100 seconds WiseMAC specialized MAC protocol External Antenna

Mote (Crossbow Inc.) Commercial sensor nodes based on UCB design and TinyOS operating system

Mote (cont.) Variety of different nodes MICA2 or 802.15.4/ZigBee protocols 315/433/868/916 MHz options (MICA) or 2.4 GHz (ZigBee) 1 yr operation out of AAA batteries

MUSE (WINLAB) Wireless Information Network Laboratory, Rutgers University, New Jersey Commercial embedded computers and WLAN transceiver Target is complete integration

ORBIT (WINLAB) 400 nodes Pure software testbed No development of sensor hardware

The AVM eGrain Project AVM – „Autarke Verteilte Mikrosysteme“ 3 year BMBF project with these partners: MWT - ANT - TKN AVM BMBF grant No. 16SV1658

Concept Development of completely autarkic ultra low power pico cell network Nodes are self organizing, no master/slave principle Highly integrated, node size ~1 cm3 RF frequency 24 GHz Development of low power system architecture Development of ultra low power RF components

Wakeup Strategies Nodes need to be in a sleep mode most of the time, but how and when to activate them? Periodic Wakeup Wakeup Receiver  No extra components  Network synchronization necessary  Waste of power through unnecessary wakeups  Delay in communications  Immediate response  No reference clock  Standby power consumption

Block Diagram of the Wakeup Circuit

Detector Principle Zero bias Schottky Diodes FET size increases from first to third stage

„MOS Diode“ Diode-like behavior of an NMOS Transistor:

Alternative Principle MOS rectifier CMOS compatible, no BiCMOS necessary But: less sensitivity, more standby power

Wake-Up Address Decoder Main requirement: low power consumption Block diagram: Detector PWM-signal serial Input Shift registers Adress correlator A1-A8 Discriminators & Logic serial data (0/1) clock reset Adress preset Wakeup parallel data valid

Prototype of Address Decoder CMOS-technology Low complexity: ca. 470 transistors No oscillator Low data rate: e.g. 50 kb/s Address preset RF Input PWM signal Wakeup-Output Bias Vcc: + 3V

RF Frontend Overview Frontend characteristics Frequency: 24.125 GHz Range: ca. 1 m Transmit power: ca. 1 mW Flip-Chip-Assembly and integrated Antenna IC-Technology: GaAs-HBT-MMICs (FBH) TU Berlin (MWT, ANT), FBH

Heterodyne Concept Standard approach Upconversion and downconversion mixers Good channel selectivity Oscillator needed for Tx and Rx

Zero IF Concept No oscillator needed in the receiver Power consumption determined by LNA Low complexity, low power consumption

Baseband demonstrator Concept Real data transmission at 24 GHz Patch antenna realised on multilayer PCB Minimum component count

Transmitter On-Off-Keying (OOK) modulation No power consumption in standby mode No power consumption for „0“ bits

Receiver Zero IF No mixer => no LO necessary (power saving!) LF amplifier has very low current consumption (ca. 100 µA) Total battery current < 15 mA Dielectric Resonator as BPF

Detector Detector Diode type HSCH-3486 (Agilent) Single stage detector Other topologies are less efficient (bridge, cascade) PTX = 0 dBm, Pathloss (1m@24 GHz) = 76 dB => PRX = –76 dBm, Gain LNA = 13 dB => Pin, Detector = –63 dBm => Uout = 3 µV Matching

LNA Measured LNA performance 14 mA DC @ 2 V 13.3 dB gain @ 24.8 GHz Bandwidth 4.2 GHz NF 5.8 dB (simulated) Chipsize 1.1x1.3 mm

Assembly Aperture coupled patch antenna Industry standard multilayer PCB RF Chip Flip-Chip mounted LF electronics in SMD Housing soldered => only standard assembly technologies

Patch Antenna Principle Whole module size is antenna base Great beam collimation Directivity 19.6 dB Gain 8.5 dB (theo. Max. 9 dB) Coax feed

Aperture coupled feed Greater bandwidth than for coaxial feed Lower directivity of 15.6 dB Gain 7.8 dB Fabrication much easier than coax feed

Photo 1 cm3 2 button cell batteries 24 GHz 2400 bps

Future Work CMOS LNA to further reduce power consumption of the receiver (7 mA @ 1.2 V) Integration of detector with LNA BiCMOS with schottky diodes Pure CMOS with MOS rectifier Complete integration as SoC

Summary Today the vision is still far from reality But many efforts and progress are made in Hardware design (digital and RF) Integration and miniaturization Energy scavenging and storage Software design Some day the vision will become reality!

References T. T. Hsieh Using sensor networks for highway and traffic applications IEEE Potentials, vol. 23, no. 2, pp. 13 – 16, Apr-May 2004 Christian C.Enz, Amre El-Hoiydi, Jean-Dominique Decotignie, Vincent Peiris WiseNET: An Ultralow-Power Wireless Sensor Network Solution IEEE Computer, August 2004, p. 62-70 Shad Roundy, Brian P. Otis, Yuen-Hui Chee, Jan M. Rabaey, Paul Wright A 1.9GHz RF Transmit Beacon using Environmentally Scavenged Energy IEEE Int.Symposium on Low Power Elec. and Devices 2003 Stefan von der Mark, Meik Huber, Mathias Wittwer, Wolfgang Heinrich, and Georg Boeck System Architecture for Low Power 24 GHz Front-End Frequenz -Zeitschrift für Telekommunikation, Special Issue Autarkic Distributed Microsystems in Sensor Networks, 3-4/2004, p. 70-73 M. Huber, S.v.d. Mark, N. Angwafo and G. Boeck Ultra low power Wakeup Circuits for Pico Cell Networks, A conceptional View Technical Report of the 1st European Workshop on Wireless Sensor Networks (EWSN), Jan 2004 Stefan von der Mark, Roy Kamp, Meik Huber and Georg Boeck Three Stage Wakeup Scheme for Sensor Networks IEEE/SBMO International Microwave and Optoelectronics Conference IMOC 2005; Brasilia, Brazil, July 25-28 http://tcs.unige.ch/doku.php/web/wirelesssensornetworks University of Geneva in Switzerland http://bwrc.eecs.berkeley.edu BWRC at UCB: PicoRadio, PicoNode, PicoBeacon http://www.csem.ch CSEM: WiseNet http://www.xbow.com Crossbow: Mote http://www.winlab.rutgers.edu WINLAB: Muse, Orbit http://www-mwt.ee.tu-berlin.de Technische Universität Berlin Microwave Engineering: AVM