Compressed Sensing With Dynamic Retransmission Algorithm in Lossy Wireless IoT

• Published in: IEEE access Vol. 8; pp. 133827 - 133842
• Main Authors: Jiang, Bo, Huang, Guosheng, Li, Fufang, Zhang, Shaobo
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Compressed sensing; data collection; Data compression; Data sampling; Data transmission; dynamic retransmission; Energy consumption; Energy utilization; Error reduction; Heuristic algorithms; Internet of Things; IoT; lifetime; Packet loss; Packets (communication); Reconstruction; Sensors; Service life assessment; Wireless communication; Wireless communications; Wireless networks; Wireless sensor networks; WSN
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2020.3011150

Wireless sensor networks (WSNs) based on compressed sensing (CS) can complete data sampling and data compression simultaneously, thereby greatly reducing the data transmission volume and the energy consumption of the network. However, many studies have not considered the loss of data packets due to the unreliable wireless communication, which leads to the data reconstruction not being as accurate as the applications require. In this paper, a Compressed Sensing with Dynamic Retransmission (CSDR) algorithm is proposed to guarantee high data reconstruction accuracy, high network lifetime and high energy utilization. The CSDR algorithm dynamically determines the max packet loss retransmission times of different nodes according to their residual energies, for Internet of Thing (IoT) devices with relative high energy consumption, fewer max retransmission times is adopted to maintain a longer network lifetime. For energy-rich IoT devices, more max retransmission times is used to improve the data transmission accuracy and the performance of data reconstruction. Strict theoretical analysis and experimental results show that the CSDR algorithm significantly improves the main performance indicators compared to the previous strategy: The Normalized Mean Absolute Error (NMAE) is reduced by 64.5%, and the effective utilization of energy is improved by 34.1% on average, under the condition that the network lifetime is no less than the previous scheme.