Ultra Low Power Sensor Node for Security Applications, Facilitated by Algorithm-Architecture Co-design

Design of an ultra-low power sensor node for identifying human trespassing is proposed, that can be attractive for security applications. Approximate Frequency Transformation (AFT) technique has been employed to characterize and classify the acoustic signals in order to identify sounds produced by human motion and human voice, under practical surrounding conditions. The approximate frequency spectrum is constructed using simple computations like the detection of zero-crossings and local peaks. The algorithm being memory-intensive mandates careful memory access scheme, along with optimum choice of classifying feature parameters. A custom designed Generalized Regression Neural Network (GRNN) block is used to classify the AFT results. The proposed design employs tight co-optimization of the algorithm and the corresponding computing architecture to achieve highly energy efficient "Edge-Computing" on the sensor node and hence, can facilitate deployment of large scale Wireless Sensor Network, with high node-density for security applications.

[1]  T. Teixeira,et al.  A Survey of Human-Sensing : Methods for Detecting Presence , Count , Location , Track , and Identity , 2010 .

[2]  Yu Hen Hu,et al.  Vehicle classification in distributed sensor networks , 2004, J. Parallel Distributed Comput..

[3]  Electronics Letters , 1965, Nature.

[4]  Anna W. Topol,et al.  Stable SRAM cell design for the 32 nm node and beyond , 2005, Digest of Technical Papers. 2005 Symposium on VLSI Technology, 2005..

[5]  Armando Astarloa,et al.  Hardware architecture for a general regression neural network coprocessor , 2007, Neurocomputing.

[6]  Denis McKeown,et al.  Vehicle classification by acoustic signature , 1998 .

[7]  Tülay Yildirim,et al.  FPGA implementation of a General Regression Neural Network: An embedded pattern classification system , 2010, Digit. Signal Process..

[8]  Shuvra S. Bhattacharyya,et al.  Energy-Aware Data Compression for Wireless Sensor Networks , 2007, 2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP '07.

[9]  W. Gosling,et al.  Time-encoded speech , 1978 .

[10]  G. Asada,et al.  Wireless integrated network sensors: Low power systems on a chip , 1998, Proceedings of the 24th European Solid-State Circuits Conference.

[11]  Prasant Mohapatra,et al.  Power conservation and quality of surveillance in target tracking sensor networks , 2004, MobiCom '04.

[12]  Gerhard P. Hancke,et al.  Industrial Wireless Sensor Networks: Challenges, Design Principles, and Technical Approaches , 2009, IEEE Transactions on Industrial Electronics.

[13]  Ohbyung Kwon,et al.  Acoustic Sensor Based Recognition of Human Activity in Everyday Life for Smart Home Services , 2015, Int. J. Distributed Sens. Networks.

[14]  Ian F. Akyildiz,et al.  Sensor Networks , 2002, Encyclopedia of GIS.

[15]  Donald F. Specht,et al.  A general regression neural network , 1991, IEEE Trans. Neural Networks.

[16]  I. Pollack,et al.  Effects of Differentiation, Integration, and Infinite Peak Clipping upon the Intelligibility of Speech , 1948 .

[17]  A.P. Chandrakasan,et al.  A 256 kb 65 nm 8T Subthreshold SRAM Employing Sense-Amplifier Redundancy , 2008, IEEE Journal of Solid-State Circuits.

[18]  Anantha Chandrakasan,et al.  Low-power wireless sensor networks , 2001, VLSI Design 2001. Fourteenth International Conference on VLSI Design.

[19]  Georgios P. Mazarakis,et al.  Vehicle classification in Sensor Networks using time-domain signal processing and Neural Networks , 2007, Microprocess. Microsystems.

[20]  Krish Ahuja,et al.  A Review of the Role of Acoustic Sensors in the Modern Battlefield , 2005 .