Automatic Generation Control (AGC) is a key control system to keep the power system’s balance between load and supply by maintaining its frequency in a specific range. It collects the tie-line power flow and frequency measurements of each control area to calculate the Area Control Error (ACE) and then adjusts power generation based on the calculated ACE. However, malicious frequency or tie-line power flow measurements can be injected and then AGC is misled to make false power generation adjustment which will harm power system operations. Such attacks can be carefully designed so that they pass the power system’s existing bad data detection schemes. In this work, we propose Long Short Term Memory (LSTM) Neural Network-based methods and a Fourier Transform-based method to detect and localize such data forgery attacks in AGC. These methods only utilize historical data, which are already available in existing AGC systems, making them easy to be deployed in the real world. They learn normal data patterns from historical data and detect abnormal patterns caused by attacks. To make it easier for users to use the solution, we also propose methods for automatically finding the proper detection threshold based on user needs. These methods are tested both on real and simulated datasets and show high detection and localization accuracy.

The volume of data is increasing rapidly currently, which poses a great challenge for resource-constrained clients to process and analyze. A promising approach for solving computation-intensive tasks is to outsource them to the cloud to take advantage of the clouds powerful computing ability. However, it also brings privacy and security issues since the data uploaded to the cloud may contain sensitive and private information which should be protected. In this paper, we address this problem and focus on the privacy-preserving and secure outsourcing of large-scale nonlinear programming problems (NLPs) subject to both linear constraints and nonlinear constraints. Large-scale NLPs play an important role in the eld of data analytics but have not received enough attention. In our outsourcing protocol, we rst apply a secure and efcient transformation scheme at the client side to encrypt its private information. Then, we use the reduced gradient method and generalized gradient method at the server side to solve the transformed large-scale NLPs under linear constraints and nonlinear constraints, respectively. We provide security analysis of the proposed protocol, and evaluate its performance. The experimental results show that we can efciently solve the large-scale NLPs and save huge amount of time for the clients.

Power grids serve as the backbone of modern society, supplying electricity for a wide range of critical uses, including the residential, commercial, and industrial sectors. As the power grid infrastructure becomes more reliant on networked digital technologies, it also becomes more vulnerable to cyberattacks. To ensure the resilience and security of the power grid, it is necessary to identify security vulnerabilities in power electronic devices. Fuzzing is an effective technique for detecting security flaws in software/firmware by subjecting the system under test to a series of unexpected inputs, which elicits unexpected behavior such as system crashes. While some work has been done on fuzzing power grid systems, there still lacks a comprehensive, comparative study of existing solutions. Furthermore, new methods are still needed to detect unexpected behaviors of power electronic devices during fuzzing. In this paper, we first conduct a survey and a comparative study of existing fuzzing frameworks for power grid systems. We then hypothesize and verify the use of electromagnetic waves as a method to detect silent system crashes in power electronic devices, aiming to advance fuzzing in this field.

Open-Source Intelligence (OSINT) is often regarded as a critical component for cybersecurity intelligence gathering to secure cyber infrastructures. As Artificial Intelligence (AI) and social media progress and become more commonplace, we have the unique opportunity to collect and analyze information from social media in real time. In this paper, we propose an AI-based framework for automatically filtering the Twitter stream for cybersecurity information, analyzing the natural language to aggregate the tweets around specific cybersecurity events, and certifying the information. The system’s applications range from discovering events security operators may not have been aware of to helping operators investigate ongoing events affecting their systems. Furthermore, we propose a method to identify the network protocols associated with software vulnerabilities based on Twitter data, leveraging the aforementioned framework. This method can further aid operators in understanding the exploitability of vulnerabilities (e.g., whether exploitation is already blocked by firewall rules) and making mitigation measures (e.g., adding a firewall rule to block exploitation). Evaluations show the effectiveness of our solutions.