Loading...

The Detection of False Data Injection Attacks in Large-Scale Residential Energy Grids

Babaei Ghazvini, Hamid Reza | 2022

96 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55626 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Haeri, Mohammad
  7. Abstract:
  8. This dissertation investigates the detection of false data injection attacks in largescale residential energy grids. A typical multi-micro-grid framework is investigated by considering the geographical dispersion of equivalent models, wherein each microgrid is assumed as an agent and managed by a local aggregator. Furthermore, a fully distributed Nash equilibrium seeking algorithm in the partial-decision information scenario is proposed, where each microgrid estimates the actions of all the microgrids by relying only on the information exchanged with some neighbors over a communication network. To effectively reduce unnecessary signal transmission among the agents, an adaptive algorithm, where threshold parameters are adjusted based on the behavior of the estimated signals in the last two iterations, is proposed. Due to the strong reliance of multi-micro-grids on advanced communication networks for monitoring and control purposes, they are severely vulnerable to false information injected by adversarial agents. Among adversarial agents, there are Byzantine agents who are allowed to send out arbitrary and even different false data to different neighbors.They can send out false data about their and all other agents’ strategies in a few iterations of an iterative non-cooperative game. It is observed that when false datais injected into an agent, its estimated signals begin to diverge from the equilibrium point, and as a result, the threshold value becomes increasing. Therefore, the increase in threshold value is a sign of Byzantine agents’ presence, and the verification of received data is done when the presence of Byzantine agents among the set of inneighbors is detected. By doing so, the computational burden reduces compared to the existing algorithms in which the correctness of received data is made in all iterations.The Byzantine agents are identified and isolated using a reputation-based mechanism and are allowed to rejoin the network when they behave normally again. An idea based on receding horizon is utilized to deal with users’ highly random behavior.Finally, a set of simulation results is presented to evaluate the effectiveness of the suggested algorithm.
  9. Keywords:
  10. Microgrid ; Distributed Management ; Energy Management ; Byzantine Attacks ; Byzanyine Agents Detection ; Byzantine Agents

 Digital Object List

 Bookmark

No TOC