Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Data-Driven Network Intelligence for Anomaly Detection and Information Privacy
Abstract
Facing groups of end devices from smart applications in vehicular networks, 5G networks, and Industrial Internet-of-Things (IIoT), an effective edge intelligence is crucial in providing efficient solutions for data intensive, low latency, and availability assured missions. Cyber security threat will remain as the main challenge for an edge intelligence. Advanced persistent threat (APT), network intrusion, data privacy, and location privacy are constantly invading the communication networks, further disturbing the guaranteed services, and destroying critical end devices and cyber infrastructure. In this dissertation, we first present a proposed framework of data-driven network intelligence for anomaly detection. The proposed system consists of three major components: edge enabled infrastructure, edge enabled artificial intelligence (AI) engine and threat intelligence. Edge enabled infrastructure provides efficient and effective computing resources for parallel computing and data storage. Edge enabled AI engine produces optimal edge learning models for threat detection and enables efficient model update both locally and globally. Threat intelligence offers real-time network monitoring and cyber threats detection. With the proposed network intelligence framework, a robust network anomaly detection scheme is further studied and applied on the proposed edge intelligence. Four phases are designed to incorporate with statistics and data-driven approaches to train an edge learning model which is able to detect and identify a network anomaly in a robust way. Positive impacts of the proposed scheme are addressed in this dissertation, including the robustness of trained model and the efficiency on the detection of specific anomalies. Furthermore, a privacy-preserving anomaly detection is proposed and studied in the edge intelligence. Specifically, we present a novel predicate encryption based anomaly detection protocol. We also present the system and security model in the proposed edge intelligence and provide detailed descriptions on the design of predicates and privacy-preserving anomaly detection procedures. Positive impacts include the efficiency of awareness over anomalous patterns and the continuity of privacy-preserving monitoring over sensitive data.
Subject Area
Computer Engineering|Information science|Electrical engineering
Recommended Citation
Xu, Shengjie, "Data-Driven Network Intelligence for Anomaly Detection and Information Privacy" (2019). ETD collection for University of Nebraska-Lincoln. AAI22584542.
https://digitalcommons.unl.edu/dissertations/AAI22584542