1 Reduce Noise Level Accelerometer Board Filtering hardware - PDF document

Health Monitoring of Categories of WSN Applications Civil I nfrastructures Using W ireless Sensor Netw orks � Monitoring environments � Great duck island, Redwood forest � Focus on low-duty cycle and low power consumption � Monitoring objects – High Fidelity Sampling Chenyang Lu � Machine health monitoring, earthquake monitoring, CSE 520S structural health monitoring � Focus on fidelity (quality) of sample S. Kim, S. Pakzad, D. Culler, J. Dem mel, G. Fenves, S. Glaser, � Interacting with space and objects and M. Turon, Health Monitoring of Civil Infrastructures Using Wireless Sensor Networks, IPSN, April 2007. � Lighting control � Focus on control 2 Structural Health Monitoring Traditional Approach vs. WSN � Two Damage Detection Approaches: � cost of equipment is high � direct (visual inspection, x-ray, etc.) � installation is very expensive due to wiring � indirect (detecting changes in structural � maintenance is expensive properties/ behavior) � Two Major Categories � WSN provides the same functionality at a � disaster response (earthquake, explosion, much lower price � higher spatial density etc.) and � $600 per point compared to thousands of dollars � continuous health monitoring (ambient for a data point in traditional sensor networks vibrations, wind, etc.). 3 4 Major Requirements of WSN System Architecture � Sensitive Data Acquisition System � High-frequency Sam pling w ith Low Jitter � Time Synchronized Sampling � FTSP � Large-scale Multi-hop Network � MintRoute � Reliable Command Dissemination � Flooding � Reliable Data Collection � Straw 5 6 1

Reduce Noise Level Accelerometer Board � Filtering � hardware single-pole -6db low-pass filter with a cutoff frequency of around 25Hz Top two: ADXL202E � Oversampling Bottom two: Silicon Design 1221L � ADC much faster than sampling frequency 16-bit ADC � allow oversampling and averaging � desired sampling rate 200Hz � real sampling rate 1KHz 7 8 Power Consumption Accelerometer Board (more) � � Thermometer The accelerometer board consumes about twice the energy as the mote. � for calibration � Only the mote should be directly connected to the battery � accelerometers are and turn on/ off the other components. sensitive to temperature � Voltage Regulator � 3V to the mote � 5V to the accelerometer board � constant voltage essential for the calibration of the accelerometers 9 10 Calibration (important!) iMote2 Calibration on a Beam -WUSTL � Board placed in an underground vault to Free vibration 0.15 measure static noise floor. 0.1 � Tilt test process at different angles. 0.05 � SiliconDesigns 1221L saturates at + / - 150 mG � ADXL202 ranges + / -2G Amplitude 0 � Shake bed tests (0.5Hz ~ 8Hz) verifies −0.05 dynamic performance −0.1 � Oven test for temperature calibration −0.15 WIRELESS SENSORS � Chips not only sensitive to temperature, but also to WIRED SENSOR temperature change −0.2 0 2 4 6 8 10 12 Time 11 12 2

Jitter Analysis of Jitter � Spatial: between different nodes Sampling � variation in crystals � Flooding Time Synchronization Protocol Other jobs like EEPROM write [ FTSP] is adequate Time Non-preemptible portion Preemptible portion � Temporal: different samples in a node Probability � Temporal jitter > spatial jitter in high � C: context switch time frequency sampling � T i : length of atomic section i � W: CPU wake up time P 2 / T 2 Jitter P 1 / T 1 0 C W T 2 + C T 1 + C 13 14 Jitter Measurement Reduce Jitter � MicroTimer: async timer in Khz � Turn off all external component except FLASH during sampling � worst case jitter is determined by the longest atomic section � avoid atomic sections from other components 15 16 Reduce Jitter Reliable Data Collection � Atomic sections are unavoidable on a single � SHM requires 100% data delivery microcontroller. � Challenges � Reduce jitter � Unreliable links � turning off unnecessary components � reduce � Interference number of atomic sections � Straw: Scalable Thin and Rapid Amassment � faster microcontroller � shorten atomic sections � Test-and-set instruction � shorten atomic sections � Lossless data collection protocol � May eliminate jitter by using a separate microcontroller to sample sensors. 17 18 3

Straw Reliable Data Collection at GGB Bandwidth versus Hop Count � Receiver initiated 1400 Aug 1st � Selective NACKs 1200 Aug 7th Bandwidth (B/s) Sep 20th 1000 � Rate of transmission determined by 800 sender’s distance to base station 600 � Max 5 hops to allow pipelining 400 � Based on measured interference range 200 0 0 10 20 30 40 50 Hop Count Data is collected reliably over a 4 6 -hop netw ork, w ith a 19 bandw idth of 4 4 1 B/ s at the 4 6 th hop 20 Adjusting Packet Size The Golden Gate Bridge � Increase packet size � � reduces header overhead � higher loss rate and retransmission cost � Doubling the packet size to 72B double the bandwidth in the test � Optimistic because the test was done in the lab with 99.8% delivery success rate 21 22 Deployment Environment Sausalito SF (north) (south) 500 ft 4200 ft 1125 ft 246 ft 51 nodes 8 nodes � Distance between nodes on the span is 100ft or 50ft Fog � Initially designed as 150ft Strong and salty wind � Rapidly changing Difference in MicaZ radio output power was up to 7.5dBm ... high and scary 23 24 4

Battery (4 X 6V Lantern Battery) Node Node Zip tie around Bi-directional Antenna Path Antenna Node (Mote + Accelerometer Board) Extreme Rusting of C-clamp 25 26 Base Station Installation Crawling and Installing Hard Hat Done! Strong Wind Harness However… Ouch Laptop Students At Work Sharp Edge 27 28 Vibration Data Vibration Data Time and Frequency plots, Vertical sensors, s284n62 Time and Frequency plots, Vertical sensors, s284n45 Accel (mg) 20 50 Accel (mg) 0 0 -20 -50 0 100 200 300 400 500 600 0 100 200 300 400 500 600 Time (sec) Time (sec) 5 Accel (mg) Accel (mg) 10 0 0 -5 -10 45 50 55 60 65 45 50 55 60 65 Time (sec) Time (sec) PSD (mg/Hz) 2 PSD (mg/Hz) 4 1 2 0 0 0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 35 40 45 50 frequency (HZ) frequency (HZ) 2 PSD (mg/Hz) 4 PSD (mg/Hz) 1 2 0 0 0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5 frequency (HZ) frequency (HZ) Vertical Sensor at quarter-span Vertical Sensor at quarter-span 365m North of the South Tower 335m South of the North Tower Peak at 0.11Hz matches the fundamental frequency of the bridge in the past studies The vertical m odal properties m atch am ong ( 1 ) sim ulation 29 m odel, ( 2 ) previous study, and ( 3 ) this study 30 5

Conclusion Bonus – Spectacular Views � 59 nodes over the span and the tower � 100ft apart due to radio range � collecting ambient vibrations in two directions synchronously at 1KHz rate � Sampled data collected reliably over a 44 hop network � 461B/ s at the 44th hop for reliable transmission � less than 10 μ s jitter ( � max 6.67KHz sampling rate) � accuracy of 30 μ G � Collected data agrees with theoretical models and previous studies of the bridge. 31 32 Future Work More? Critique? � Increase packet size without affecting reliability and � Sensor calibration is crucial. transmission cost. � Verify sampling rates and jitter. � Design better power circuit for accelerometer board. � Energy harvesting, e.g., solar panels. � less energy would be consumed if sensors could be turned off by mote when not sampling � Additional microcontroller on the board � make real-time sampling easier � require more complex accelerometer board design. 33 34 Wireless Structural Control - WUSTL Acknowledgement � Some slides borrowed from Sukun Kim’s doctoral defense at UCB, Fei Sun’s slides at WUSTL sensor networks seminar, Shirley Dyke, and Nestor Eduardo Wireless sensor Castaneda Aguilar. MR damper 35 36 6