Research on Embedded Systems

Transcription

1 Research on Embedded Systems Chenyang Lu Department of Computer Science and Engineering

2 Embedded Systems Any device that includes a computer (but you don t think of it as a computer) iphone. Digital camera. Game console. Wrist watch. Automobile. Airplanes. Mars rover. >95% of microprocessors are used for embedded systems. 2

3 On a Car ~100 microprocessors 4 bit microcontroller checks seat belt; Microcontrollers run dashboard devices; 16/32 bit microprocessor controls engine; NavigaPon; Entertainment: DVD, audio, satellite radio 3

4 Wireless Sensor Network Processor + Sensors + Wireless Interface Miniature hardware manufactured economically in large numbers Smart Dust (UCB) 4

5 MICA2 Mote 128KB Instruction EEPROM 2 AA Chipcon CC1000 radio, 38K or 19K baud, Manchester, 315, 433, or 900MHz SPI bus 3 LEDs 4KB Data EEPROM Atmel ATmega128L µp MHz ADC 0-7 UART 1 I2C Bus 51 pin I/O Connector UART 2 512KB External Flash Memory (16 bytes x rows) To Sensors, JTAG, and/or Programming Board 5

6 WU Sensor Network Testbed 6

7 WU Sensor Network Testbed 7

8 American Society for Civil Engineers 2009 Report Card for America's Infrastructure Bridges C Dams D Levees D Rail C Roads D More than 26%, or one in four, of the napon's bridges are either structurally deficient or funcponally obsolete. America's Infrastructure GPA: D EsPmated 5 Year Investment Need: $2.2 Trillion 8

9 Current Practice Bridges: inspected manually once every two years. Costly and Pme consuming. Highway 40 Closing for Boone Bridge Inspection Monday August 10, 2009 If you're heading to St. Charles this weekend, Highway 40 is not your best option. Westbound 40 from Long Road in St. Louis County to Route 94 in St. Charles County will be closed (weather permitting) while work crews inspect the Daniel Boone Bridge across the Missouri River. The road will close at 5:30 a.m. on August 15 and won't reopen until sometime after 9 p.m. on August 16. 9

10 Structural Health Monitoring (SHM) Detect and localize damages to structures Wireless sensor networks monitor at high temporal and spapal granularipes Key Challenges ComputaPonally intensive Resource & energy constraints Long term monitoring 10

11 Existing Approaches Centralized approach: stream all sensor data to base stapon for processing. Useful for model validapon. But too energy consuming for long term monitoring? Example: Golden Gate Bridge project [Kim IPSN'07]. 46 hop network deployed along Golden Gate Bridge Nearly 1 day to collect enough data for one computapon LifePme of 10 weeks w/4 x 6V lantern balery Separate designs of sensor networks and SHM algorithm. Primarily focus on data transport issues. Not concerned with SHM algorithm used for damage localizapon. 11

12 Distributed Architecture Dilemma Too much sensor data to stream to the base stapon SHM algorithms are too complex to run enprely on sensors Raw Data Distributed Architecture Performs part of computapon on sensor nodes Send (smaller) intermediate results to base stapon Base stapon completes computapon ParPal Results 12

13 Cyber Physical Co design The term cyber physical systems refers to the Pght conjoining of and coordinapon between computaponal and physical resources. [NSF] Raw Data Cyber Physical Co design of SHM Systems IdenPfy an SHM approach that allows efficient distributed implementapon over sensor networks. OpPmally map SHM algorithms onto distributed sensor network architecture. ParPal Results 13

14 Design of Distributed SHM System Damage LocalizaPon Assurance Criterion (DLAC) [Messina 96] IdenPfy structure s natural frequencies based on vibrapon data. Signature of structure s health Match natural frequencies to structural models with damages. OpPmally parppon computapon stages between sensors and the base stapon. Minimize energy consumppon Subject to resource constraints 14

15 (3a) Coefficient ExtracPon (3b) EquaPon Solving D Integers 5*P Floats Healthy Model (1) FFT (2) Power Spectrum (3) Curve Fivng (4) DLAC D Floats D/2 Floats P Floats D: # of samples P: # of natural freq. (D» P) Damaged Location Data Flow Analysis DLAC Algorithm 15

16 4096 bytes (1) FFT D: 2048 P: 5 (3a) Coefficient ExtracPon 100 bytes (3b) EquaPon Solving (2) Power Spectrum EffecPve compression rapo of 204:1 (3) Curve Fivng 8192 bytes 4096 bytes 20 bytes Integer: 2 bytes Float: 4 bytes Healthy Model (4) DLAC Damaged Location Data Flow Analysis DLAC Algorithm 16

17 Implementation Sensor plaxorm: Intel/Crossbow Imote2 + ITS400 sensor board MHz XScale CPU 32 MB ROM, 32 MB SDRAM CC compliant radio 3 axis accelerometer on sensor board Data collecpon and processing applicapon wrilen with TinyOS KB ROM, 71 KB RAM 17

18 Evaluation: Truss 5.6 m steel truss structure at UIUC m long bays Sivng on four rigid supports 11 Imote2s alached to frontal pane Damage correctly localized to third bay 18

19 Energy Consump8on Evalua8on 19

20 Summary Cyber physical co design of a distributed SHM system. Reduces energy consumppon by 71% Reduces latency by 66% Implemented on imote2 using <1% of its memory EffecPvely localized damage on two physical structures. Demonstrated the promise of cyber physical co design of sensor network systems. G. Hackmann, F. Sun, N. Castaneda, C. Lu, and S. Dyke, A HolisPc Approach to Decentralized Structural Damage LocalizaPon Using Wireless Sensor Networks, RTSS

21 Other Research We Do Wireless papent monitoring at BJH Resilient sensor network for homes Real Pme sensor network for process control Power efficient wireless network protocols Thermal control for real Pme systems hlp:// 21