Characterizing Bandlimited Nonlinear (and Linear) Systems with Memory while Reducing the Curse of Dimensionality using a new Volterra Expansion Technique

#non-linear #signal #processing #Nuttall #Wiener #Volterra #algorithm
Share

The overarching goal of this research is to provide a new active sonar signal processing method to exploit the nonlinearity inherent in underwater sound channels. This new approach aims to enhance detection, classification, and localization (DCL) through increased frequency coverage and selection beyond the limits of the transmit bandwidth. One possible technical advantage gained from this nonlinear characterization is in mono-static and multi-static sonar. The Navy has always desired multi-static sonar capabilities within shallow water and littoral regions for underwater connectivity. Thus, any new capabilities gained within shallow water are desired since communication and tactical awareness are troublesome in high-reverberation environments. Thus, the proposed signal-processing technique obtains information not traditionally utilized in DCL, which could improve undersea sonar systems like submarines, torpedoes, and sonobuoys.



  Date and Time

  Location

  Hosts

  Registration



  • Date: 17 Oct 2017
  • Time: 06:00 PM to 09:00 PM
  • All times are (GMT-05:00) US/Eastern
  • Add_To_Calendar_icon Add Event to Calendar
  • Roger Williams University
  • 1 Old Ferry Road
  • Bristol, Rhode Island
  • United States 02809
  • Building: Engineering Building
  • Room Number: SE 201
  • Click here for Map

  • Contact Event Host
  • Starts 29 September 2017 12:00 AM
  • Ends 18 October 2017 12:00 AM
  • All times are (GMT-05:00) US/Eastern
  • Admission fee ?


  Speakers

Dr. Derke R. Hughes of Naval Undersea Warfare Center

Topic:

Characterizing Bandlimited Nonlinear (and Linear) Systems with Memory while Reducing the Curse of Dimensionality using a

The overarching goal of this research is to provide a new active sonar signal processing method to exploit the nonlinearity inherent in underwater sound channels. This new approach aims to enhance detection, classification, and localization (DCL) through increased frequency coverage and selection beyond the limits of the transmit bandwidth. One possible technical advantage gained from this nonlinear characterization is in mono-static and multi-static sonar. The Navy has always desired multi-static sonar capabilities within shallow water and littoral regions for underwater connectivity. Thus, any new capabilities gained within shallow water are desired since communication and tactical awareness are troublesome in high-reverberation environments. Thus, the proposed signal-processing technique obtains information not traditionally utilized in DCL, which could improve undersea sonar systems like submarines, torpedoes, and sonobuoys.

Biography:

Dr. Derke R. Hughes is a 2001 graduate of North Carolina A&T State University where he holds a Doctor of Philosophy Degree from the College of Engineering. Dr. Hughes is a senior researcher and signal processor in the Devices, Sensors, and Materials Research and Development Branch (NUWC 1512) of the Science and Technology Division (NUWC 151) in the Sensors and Sonar Systems Department at the Naval Undersea Warfare Center (NUWC) in Newport, RI.

 

As a senior signal processor, Dr. Hughes executed two leading roles as Technical Program Manager (TPM) and Principal Investigator for a bio-inspired sonar project for DARPA ($450 K funding level). This bio-inspired sensor concept was based upon the anatomical sound generation mechanism of the insect, cicada. Dr. Hughes accomplished his initial goal to measure and analyze the cicada mating call in an experiment conducted at the University of Michigan Biological Station in Pellston, Michigan with the perennial Okanagana rimosa. A male cicada song is largely produced from three different anatomical structures: two tymbals and an abdomen where the tymbals are located on opposite sides of the insect in front of the abdomen (the abdomen is an air cavity which is considered a resonator). In prior research, the consensus opinion held that the sound generating mechanism from these anatomical structures was based on a linear process such as the familiar linear Helmholtz resonator. Dr. Hughes et. al. initial journal article considered several nonlinear signal processing techniques such as mutual information, bi- and tri-coherences, and the Volterra expansion to ascertain the significance of nonlinearity in the cicada mating call.

 

 

Dr. Hughes has worked collaboratively with Dr. Albert Nuttall for more than ten years on the Nuttall Wiener Volterra (NWV) algorithm for underwater applications. Dr. Hughes has presented the development of the NWV to national and international audiences detailing the signal processing breakthroughs in this method including applications for sonar. Included in these advanced techniques are kernel under-sampling, data decimation, time domain and spectral (i.e., Fourier) domain processing and analysis for modified-Volterra kernels and functional estimations, filtering methods, Doppler, and multipath signal processing. Most significantly, the signal processing portions of Dr. Hughes' projects have supported the development of techniques to alleviate the Curse of Dimensionality (COD) through maximally sparse kernels. By alleviating the COD the NWV modifications to the Volterra expansion have emerged as a practical method, thus increasing the potential of its utility in underwater acoustic applications, and specifically for improved DCL.

 

As a senior systems analyst and engineer at NUWC, Dr. Hughes was project lead and has written reports on projects such as: hydrophone drift caused by sea conditions; sensitivity analysis on hydrodynamic coefficients for torpedo and salvo weapons; a GPS buoy to hydrophone uncertainty study due to sea wind and current dynamics; and uncertainty study for hydrophone survey location analysis, to mention a few. Also, as lead system engineer, Dr. Hughes executed this role for a sensor system that monitored environmental conditions to produce the sound velocity and turbidity for targeting range operations. This system records those parameters and transmits remotely to a data collection unit located on another platform or onshore. Another sensor system obtains GPS locations and wirelessly telemeters the data to a range operation center (ROC) to identify the location of targets traveling within the test range. This sensor unit is a verification unit for a multi-phenomenological sensor that measures magnetic and acoustic information simultaneously.

Dr. Derke R. Hughes of Naval Undersea Warfare Center

Topic:

Characterizing Bandlimited Nonlinear (and Linear) Systems with Memory while Reducing the Curse of Dimensionality using a

Biography:






Agenda

Join us for pizza at 6 pm with talk to follow.  Lot 9 will be open for parking after 5 pm.