Open Collections

Automated tuning of wireless mesh networks for smart meter applications

University of British Columbia

Many power utilities use wireless mesh networks to interconnect the smart meters that support monitoring, protection and optimization of the distribution grid. Network performance is critically dependent on the distribution of smart meters/mesh nodes and the path-loss over the links between nodes and degrades as the number of nearest neighbours seen by each node: 1) increases in regions of high node density (leading to mutual interference) or 2) decreases in regions of low node density (leading to reduced reliability). Although this number can be reduced or increased by adjusting the transmit power of existing nodes and/or adding additional relay nodes as appropriate, manual tuning is extremely labour intensive and automated tuning algorithms that support both functions have not been previously reported. This work contributes to the development of practical automated tuning algorithms for smart meter networks in three ways. First, we show how the accuracy of simple path-loss models that characterize the relationship between mean path-loss and distance by a linear regression line degrades as path length decreases and the degree of shadow fading increases when using: 1) the ordinary least square (OLS) approach when distance measurements are error-free and 2) the errors-in-variables (EIV) approach when distance measurements are corrupted by errors. The results allow researchers to assess the reliability of short-range data sets and determine when EIV should be used in place of OLS. Second, we use these insights to develop a measurement-based short-range power-law path-loss model applicable to the smart meter environment using massive amounts of data obtained from BC Hydro's multi-service grid network (MSGN). The result is much more reliable than previous works based on more limited data and, for the first time, reveals that the long-term temporal variability of each link follows a lognormal distribution with the standard deviation across all links in a given region itself following a lognormal distribution. Finally, we propose and demonstrate the first distributed and combined relay node placement-transmit power adjustment (RNP-TPA) algorithm for wireless mesh networks that reduces mutual interference in high density parts and improves connectivity in low density parts of the network.

Text

2020-01-03

Attribution-NoDerivatives 4.0 International

http://hdl.handle.net/2429/73094

Electrical and Computer Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nd/4.0/