Content area
As 5G mobile network continues to grow, making it more performant and secure has become an important issue. Recent advances in programmable hardware and confidential computing have opened new possibilities to addressing these concerns together.
In this dissertation, we propose three methods to address the performance, security, and privacy concerns in the 5G network. First, we developed a high performance 5G data plane User Plane Function (UPF) that allows for both high amount of concurrent user and bandwidth while maintaining low latency in both data and control plane. Second, we built on top of the UPF to introduce a combined on-path and off-path malicious app detection framework that uses domain name queries to allow scalable detection of malicious apps on users’ devices. Lastly, we deploy a practical edge 5G network using confidential computing and secure aggregation to build and run a privacy-preserving real-world traffic analysis system.