10th EAI International Conference on Communications and Networking in China

Research Article

Wearable Computing Going Green: Energy-Optimal Data Transmission for Multi-Sensor Wearable Devices

  • @INPROCEEDINGS{10.4108/eai.15-8-2015.2261044,
        author={Weizheng Hu and Weiwen Zhang and Han Hu and Yonggang Wen},
        title={Wearable Computing Going Green: Energy-Optimal Data Transmission for Multi-Sensor Wearable Devices},
        proceedings={10th EAI International Conference on Communications and Networking in China},
        publisher={IEEE},
        proceedings_a={CHINACOM},
        year={2015},
        month={9},
        keywords={buffer management; lyapunov optimization; wearable devices; wearable computing; energy consumption},
        doi={10.4108/eai.15-8-2015.2261044}
    }
    
  • Weizheng Hu
    Weiwen Zhang
    Han Hu
    Yonggang Wen
    Year: 2015
    Wearable Computing Going Green: Energy-Optimal Data Transmission for Multi-Sensor Wearable Devices
    CHINACOM
    IEEE
    DOI: 10.4108/eai.15-8-2015.2261044
Weizheng Hu1,*, Weiwen Zhang1, Han Hu1, Yonggang Wen1
  • 1: Nanyang Technological University, School of Computer Engineering
*Contact email: huwe0012@e.ntu.edu.sg

Abstract

Energy management stands out as a crucial design and operational factor for the emerging wearable devices. In this paper, we investigate how to reduce the data-transmission energy for multi-sensor wearable devices over stochastic wireless channels. Specifically, we formulate the data-transmission scheduling issue into a constrained optimization problem, with an objective to minimize the time-average energy cost under the constrains of limited queueing buffer size. We adopt the canonical Lyapunov optimization framework to derive an online algorithm to minimize the drift-plus-penalty function. Under this framework, we first characterize a fundamental trade-off between energy cost and data-transmission delay with both theoretical performance bounds and numerical verifications. We then reveal a threshold effect of the total buffer size on the energy consumption, below which the buffer size limit will be active and the energy cost rises as the buffer size decreases. We further elaborate the relationship between channel gain and energy consumption. Finally, compared to a random transmission algorithm, our approach can save up to 85.08% of energy.