Impact of False Data Detection on Cloud Hosted Linear State Estimator Performance. Chakati, V., Pore, M., Banerjee, A., Pal, A., & Gupta, S. K. In 2018 IEEE Power Energy Society General Meeting (PESGM), pages 1–5, Portland, OR, August, 2018.
Impact of False Data Detection on Cloud Hosted Linear State Estimator Performance [link]Paper  abstract   bibtex   
Linear state estimation (LSE) is used to compute the complex voltages of a power system using measurements obtained only from phasor measurement units (PMUs). With the continued addition of PMUs into the grid, classical LSE solvers would have to handle large sets of high-speed data. Furthermore, security threats in the form of false data injection (FDI) attacks must also be considered in the design, which will considerably add to the computational overhead of LSE solvers. Although installing additional computation and communication hardware is a possible solution, such a solution would incur substantial infrastructure and operation costs. In this paper, we explore the design of a cost-effective and scalable cloud hosted LSE (CLSE) solver that also has false data detection (FDD). The proposed CLSE-FDD application exploits GPU parallel processing capabilities for mitigating the performance overhead of FDD to match the operation speed of classical LSE solvers. Results indicate that the GPU based CLSE-FDD application can easily scale in excess of 1,500 PMU installations.
@inproceedings{chakati_impact_2018,
	address = {Portland, OR},
	title = {Impact of {False} {Data} {Detection} on {Cloud} {Hosted} {Linear} {State} {Estimator} {Performance}},
	url = {https://ieeexplore.ieee.org/abstract/document/8586671},
	abstract = {Linear state estimation (LSE) is used to compute the complex voltages of a power system using measurements obtained only from phasor measurement units (PMUs). With the continued addition of PMUs into the grid, classical LSE solvers would have to handle large sets of high-speed data. Furthermore, security threats in the form of false data injection (FDI) attacks must also be considered in the design, which will considerably add to the computational overhead of LSE solvers. Although installing additional computation and communication hardware is a possible solution, such a solution would incur substantial infrastructure and operation costs. In this paper, we explore the design of a cost-effective and scalable cloud hosted LSE (CLSE) solver that also has false data detection (FDD). The proposed CLSE-FDD application exploits GPU parallel processing capabilities for mitigating the performance overhead of FDD to match the operation speed of classical LSE solvers. Results indicate that the GPU based CLSE-FDD application can easily scale in excess of 1,500 PMU installations.},
	booktitle = {2018 {IEEE} {Power} {Energy} {Society} {General} {Meeting} ({PESGM})},
	author = {Chakati, Vinaya and Pore, Madhurima and Banerjee, Ayan and Pal, Anamitra and Gupta, Sandeep K.S.},
	month = aug,
	year = {2018},
	keywords = {Cloud Computing, Covariance matrices, Current measurement, False Data Detection, GPU, Graphics processing units, Linear State Estimation, Phasor Measurement Unit (PMU), Phasor measurement units, State estimation, Task analysis, Voltage measurement},
	pages = {1--5},
}

Downloads: 0