Towards Reliable Underwater Acoustic Video Transmission for Human-Robot Dynamic Interaction

Transcription

1 Towards Reliable Underwater Acoustic Video Transmission for Human-Robot Dynamic Interaction Dr. Dario Pompili Associate Professor Rutgers University, NJ, USA

2 Semi-autonomous underwater monitoring applications q Human-robot dynamic interaction requires realtime multimedia acquisition and classification q Remotely Operated Vehicles (ROVs) are key instruments to support such interactive applications They can capture multimedia data from places where humans cannot easily/safely go Bridge pillars inspection with a team of semi-autonomous robots Scenario I: Data fusion at the buoy 2 Scenario II: In-network data fusion (at the robots)

3 Objectives of our project (1) To design novel communication solutions for robust, reliable, and high-data rate underwater multimedia streaming (hundreds of hundreds of meters) (a) Use all the resources in frequency/signal/space domains (b) Via Acoustic Vector Sensors (AVSs) measure all the three components of the acoustic particle motion (particle velocity on x, y, z as well as pressure) (c) Vehicle/channel tracking&position estimation in MIMO structures/coding (2) To integrate various communication algorithms/methods on a software-defined testbed using AVSs. The testbed should: (a) Support processing-intensive PHY-layer software-defined functionalities (b) Be real-time reconfigurable based on user s Quality of Experience (QoE) 3

4 PHY-layer solution to boost the data rate for underwater acoustic transmission (1a)-(1b) Spherical representation of the angles in the proposed solution q Vehicle-buoy comms based on AVSs and Uniform Circular Array (UCA) q The vehicle is steered via signal-space-frequency beamforming M. Rahmati and D. Pompili, SSFB: Signal-Space-Frequency Beamforming for Underwater Acoustic Video Transmission, in Proc. of the IEEE International Conference on Mobile Ad hoc and Sensor Systems (MASS), Orlando, FL, USA, Oct Best Paper Award 4

5 Vehicle-buoy signaling blocks (1a)-(1b) & (2b) q Video encoding q Space-signal designation q NC-OFDM (Non-contiguous OFDM) q Beamforming q AOA estimation q Antenna decision q NC-OFDM receiver q Data-subcarrier extraction q Detection and Demodulation q Fine tuning with feedback Space-signal designation and video scalability are adjusted based on 5 user Quality of Experience (QoE)

6 Probabilistic SDMA-MAC solution in multi-robot missions using AVSs (1b)-(1c) q Probabilistic MAC solution based on Space Division Multiple Access~(SDMA) and onboard AVSs q Unscented Kalman Filtering~(UKF) to improve the estimation and update vehicle tracking using AVS q Angle-based solution to form separate spatial beams towards the target vehicles interval estimation is exploited to calculate the initial beam 6

7 Proposed protocol for tracking a vehicle (1b)-(1c) Steps 1 & 2: Vehicle s location uncertainty estimation Step 3: Coarse buoy s Angle of Departure (AoD) estimation Step 4: Fine steering estimation of the vehicle using AVS Step 5: Tracking the vehicle Step 6: Data transmission 7

8 Robust closed-loop hybrid Automatic Repeat Request (ARQ) coding technique (1b)-(1c) q Design and validate a novel hybrid ARQ coding for a MIMO- AVS structure to assure communication reliability based on estimated AoA q Coding algorithm is designed for a MIMO-AVS structure to increase the probability of successful transmission by adapting code redundancy based on angle estimation accuracy 8

9 Work in progress: adaptive software-defined MIMO acoustic (reconfigurable) testbed (2a)-(2b) q Study the distortion-rate and diversity-multiplexing tradeoffs using a Software-Defined MIMO Acoustic testbed for video streaming q Reconfigurability using Ettus X-300 Universal Software Radio Peripheral (USRP) with high processing capabilities (better than previous generation boards, B200/B210, which are not fast enough) 9

10 Conclusion q This research program has a multidisciplinary component that involves a nexus of ideas from sensor technology, communications, networking, algorithms, statistical inference, and dynamical systems (Distributed robotics + AComms) q Investigate the design/fabrication of a new class of miniaturized and integrated MEMS-based AVS arrays (high bandwidth) to enable reliable and broadband multimedia communications to adapt and support various layering (data rates) in video streaming q We envision our methods/algorithms to have a wide applicability in areas of science and technology that concern real-time multimedia transmission of coordinated, distributed robots dynamic interaction of (co)robots with environment and humans Thank You! Questions? 10