TY - JOUR
T1 - RVFL-LQP
T2 - RVFL-based link quality prediction of wireless sensor networks in smart grid
AU - Xue, Xue
AU - Sun, Wei
AU - Wang, Jianping
AU - Li, Qiyue
AU - Luo, Guojun
AU - Yu, Keping
N1 - Funding Information:
This work was supported in part by the National Natural Science Foundation of China under Grant 51877060, in part by the Fundamental Research Funds for the Central Universities of China under Grant PA2019GDQT0006 and Grant JZ2018HGTB0253, and in part by the Science and Technology Project of State Grid Corporation of China (Research and Application of Key Technologies for Operation and Maintenance of Smart Substation Based on the Fusion of Heterogeneous Network and Data).
Publisher Copyright:
© 2013 IEEE.
PY - 2020
Y1 - 2020
N2 - In the application of wireless sensor networks (WSNs) to smart grid, real-time and accurate wireless link quality prediction (LQP) is important to determine which link is reliable enough to undertake the communication task. However, the existing LQP methods are neither suitable to describe the dynamic stochastic features of link quality nor to ensure the validity of prediction results. In this paper, a random-vector-functional-link-based LQP (RVFL-LQP) algorithm is proposed. The algorithm selects the signal-to-noise ratio (SNR) as the link quality metric and decomposes the raw SNR sequence into the time-varying sequence and the stochastic sequence according to the analysis of wireless link characteristics. Then, the RVFL network is used to establish the prediction model of the time-varying sequence and the variance of the stochastic sequence. Lastly, the probability-guaranteed interval boundary of SNR is predicted, and the validity and practicability of prediction results are evaluated by comparative experiments and real-world application, respectively.
AB - In the application of wireless sensor networks (WSNs) to smart grid, real-time and accurate wireless link quality prediction (LQP) is important to determine which link is reliable enough to undertake the communication task. However, the existing LQP methods are neither suitable to describe the dynamic stochastic features of link quality nor to ensure the validity of prediction results. In this paper, a random-vector-functional-link-based LQP (RVFL-LQP) algorithm is proposed. The algorithm selects the signal-to-noise ratio (SNR) as the link quality metric and decomposes the raw SNR sequence into the time-varying sequence and the stochastic sequence according to the analysis of wireless link characteristics. Then, the RVFL network is used to establish the prediction model of the time-varying sequence and the variance of the stochastic sequence. Lastly, the probability-guaranteed interval boundary of SNR is predicted, and the validity and practicability of prediction results are evaluated by comparative experiments and real-world application, respectively.
KW - RVFL network
KW - Wireless sensor networks
KW - link quality prediction
KW - probability-guaranteed interval boundary
UR - http://www.scopus.com/inward/record.url?scp=85078295861&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85078295861&partnerID=8YFLogxK
U2 - 10.1109/ACCESS.2020.2964319
DO - 10.1109/ACCESS.2020.2964319
M3 - Article
AN - SCOPUS:85078295861
SN - 2169-3536
VL - 8
SP - 7829
EP - 7841
JO - IEEE Access
JF - IEEE Access
M1 - 8951146
ER -