Signal Processing Aspects of Realtime Wireless Sensor Network

Signal Processing Aspects of Realtime Wireless Sensor Network Applications György Orosz Department of Measurement and Information Systems Budapest University of Technology and Economics, Hungary HNI - MIT Knowledge Sharing Symposium Budapest, Hungary, February 10. , 2010

Wireless signal processing o o Sensor network from signal processing aspects Real-time signal processing n n o Advantages of Wireless Sensor Networks (WSNs) n n o Easy to install Flexible arrangement Difficulties of utilization of WSN n n o Fast changing signals Hard real-time operation Data loss Undeterministic data transfer Limit of the network bandwidth Lots of autonomous systems Topics n n Signal sensing Synchronization of autonomous subsystems Network protocol Distributed signal processing

ANC: a case study o o Plant to be controlled: acoustic system Noise sensing: microphone mote 1 Berkeley micaz motes mote 2 o o o mote. N Actuators: active loudspeakers Gateway: network DSP Signal processing: n DSP board o o o DSP board codec reference signal DSP n mote. G Motes o o gateway mote o o o ADSP-21364 32 bit floating point 330 MHz 8 analog output channels Tiny. OS ATmega 128 Sensor boards 250 kbps radio Identification

Physical arrangement active loudspeaker DSP board gateway mote sensor mote

Sampling precision 1. Sampling with low priority shared timer Sampling with high priority dedicated timer

Sampling precision 2. □ Middle level timing priority □ 25 samples size packets □ Effects of disturbances n n Random disturbance: contributes to noise Periodic disturbance : spurious spectrum lines Deviation from average period ( td ) t Increasing deviation (td) from periodic disturbance Average period

Synchronization 1. Tn-2 Tn-1 TS_mote Tn tmote Tt Tt o Tt Delay: Td = Tt + dt Unsynchronized subsystems: n n dti– 1 TS_DSP o Ti-2 dti Ti-1 Ti t. DSP o Goal: constant delay n TS_mote : sampling period of the motes TS_DSP : sampling period of the DSP Tt : data transmission delay Tt dt Changing delay Stability problems in feedback systems n Tt=const. : deterministic protocol dt=const. : synchronization

Synchronization 1. (stability) noise samples sent by the sensor Ti sampling Tn estimated noise according time to the estimated delay real delay anti noise from estimated noise anti noise: real noise signal: Stable: noise suppression the delay estimation is correct

Synchronization 1. (stability) noise samples sent by the sensor Ti-1 sampling Tn-2 estimated noise according time to the estimated delay real delay anti noise from estimated noise anti noise: real noise signal: Unstable: noise amplification the delay estimation is incorrect

Synchronization 2. tsyst 1 Td 1 Physical synchronization: n n Td 2 tsyst 2 Tn Td 1=Td 2=const o o o d 1 f(t) n Ti dt Sampling frequencies are the same Tuning of the timers Interpolation: Signal value is estimated in signal processing points Algorithm transformation: algorithm parameters are transformed into Ta (when data arrived). Synchronization in the ANC system: n d 2 d 3 o Motes: physical Motes DSP: linear interpolation Tn t TSmote Physical synch. tmotes Tt Ta: arrival time of data Interpolation Interp. Ti t. DSP

Data transmission methods Transmission of row data o o 1. 8 k. Hz sampling frequency on the motes Synchronization of WSN DSP LMS and resonator based ANC algorithms Bandwidth restriction: about 3 sensors Transformed domain data transmission o o o 1. 8 k. Hz sampling frequency on the motes Transmission of Fouriercoefficients Increased number of sensors: 8 sensors (expansion possible)

Distributed ANC system A(z) error signals acoustic plant mote 1 FA DSP mote 2 FA gateway ANC algorithm R(z) mote. N FA : synchronization messages : data (Fourier-coefficients) transmission messages o o Fourier analysis on motes Control algorithm on DSP Synchronization of base functions Computational limits reference signal control signals

Summary and future plans o o Utilization of WSN in closed loop signal processing systems Importance of signal observation n n o o Sampling Synchronization Distributed signal processing Future research goals n n Searching for possible ways of data reduction Analysis of the effect of data loss