Learning in Compressed Domain Development of Light Learning Engine in Fog Computing Environments

PI: An-Yeu Wu (National Taiwan University)

Champion: Yen-Kuang Chen (Intel)

Status Quo:  

–     Compressing” after “sampling” wastes time and memory

–     “Compressive sensing” combines sampling and compressing 

  • Only measure what should be measured
  • Reduce cost and latency in sampling
  • Reduction of sensor and transmission power by CS

–     Fog Computing for IoT (promoted by Cisco)

  • Proximity to end-users
  • Latency reduction for quality of service
  • Edge analytics

–     Resource constraints in Fog Computing

  • Limited computing hardware and memory in (layered) fog nodes
  • High computational complexity algorithm

Key New Insights:  

–     Real-time reconstruction is required for continuous monitoring

–     High computation overhead for CS reconstruction

–     Reduce complexity by removing the overhead of reconstruction

 

 (Updated in Jul, 2017)