Andrew W. Moore - Student Project / Masters Dissertation Proposals

These might look like MPhil work but we can rescope them to be more Part-2 sized as well.

Linux kernel expert or eBPF hacker?

Low-overhead eBPF kernel tracing.

eBPF permits a relatively naive programmer to add system (kernel) code triggered by in-system events (clocks, system calls, packet arrival or departure, and other similar events.)

This work will look to implement and evaluate an eBPF based implementation of the KUtrace authored by Dick Sites. The work will look to implement as near-complete cover as possible and then perform a comparison of overheads and impact.

The work will allow head-to-head comparison of an optimised (this is going to be the brunt of the fun work) eBPF implementation that would seek to impact a system as minimally as the KUtrace system.

Part of this work will seek to compare and as appropriate build on other work such as that of KUtrace and of other, eBPF, instrumentation approaches.

The KUtrace system is optimised for minimal impact; this will be the design criteria from the outset and become a practical comparison between the KUtrace approach and your new ideas.

KUtrace References: https://github.com/dicksites/KUtrace

and

Sites, Richard “Understanding Software Dynamics”, Addison-Wesley, 2022.

eBPF References: https://ebpf.io/

kernel instrumenttaion with eBPF blog: https://sematext.com/blog/linux-kernel-observability-ebpf/

SCION centered project

Dynamic Multipath Video Streaming Application

Focus: Software Engineering, Application Layer, Multipath Optimization.

The Core Idea: SCION naturally exposes multiple paths between a source and destination. In this project, the student will build a video streaming server/client (or adapt a protocol like DASH or QUIC) that splits video chunks across entirely different geographic paths simultaneously to optimize quality and survive network failures.

What the project will build/do:

Set up a small local SCION topology or connect to the SCIONLab testbed.

Implement a streaming client in Go or Python using the SCION socket API.

Program the client to fetch available paths and evaluate them in real-time (e.g., Path A goes through ISPs in Europe, Path B goes via North America).

Implement an algorithm that dynamically splits traffic (e.g., sending high-priority video frames over the lowest-latency path, and audio/redundant data over a higher-bandwidth path).

The “Wow” Evaluation: Simulate an ISP failure mid-stream on one path. Show how the video keeps playing without a single stutter because the client instantly falls back on the secondary path without waiting for traditional BGP convergence.

A Smart “Path Picker” Daemon for End Hosts (AI/Heuristic-Driven)

Focus: Data-driven Networking, Algorithms, Devops/Systems.

The Core Idea: SCION gives end-hosts total control over which paths their packets take. However, regular users have no idea how to choose an optimal path. This project involves building a lightweight, local daemon that sits on a user's machine, actively probes SCION paths, and dynamically changes the active path based on the application's current needs (e.g., lowest latency for gaming, highest bandwidth for downloads).

What the project will build/do:

Create a background service (daemon) that continuously queries the SCION path server for destination Autonomous Systems (ASes).

Implement a continuous, low-overhead path-probing mechanism (measuring latency, jitter, and estimated packet loss).

Expose a local API or configuration file where applications can declare their profile: Gaming (minimize jitterlatency), Backup (maximize bandwidth), or Geofenced (avoid passing through specific countriesISDs for geopolitical/privacy reasons).

Use SCION's path switching capabilities to rewrite the forwarding path used by the host's networking stack on the fly.

Designing and Simulating a SCION-native IoT Gateway

Focus: Computer Engineering, Embedded Systems/IoT, Network Security.

The Core Idea: IoT devices are notorious targets for DDoS attacks. SCION drastically mitigates DDoS because receivers must explicitly authorize paths, and source spoofing is cryptographically prevented. Because IoT devices are too weak to run full SCION stacks, the student will build an entry-level “SCION IoT Gateway” using a Raspberry Pi or an emulated environment.

What the project will build/do:

Set up a Raspberry Pi to act as a gateway router between a local legacy IP network (where smart devices live) and an external SCION network.

Implement an translation layer (similar to the SCION-IP-Gateway or SIG, but optimized for lightweight IoT traffic like MQTT or CoAP).

Program the gateway to automatically map specific IoT devices or topics to dedicated SCION isolation domains (ISDs).

The “Wow” Evaluation: Mount a simulated DDoS attack from the outside internet. Demonstrate how legacy IP IoT infrastructure collapses, while the SCION gateway isolates the attack at the border router level, keeping the internal IoT devices functioning perfectly.

FPGA / RTL / Verilog / VHDL expert?

Come take to me; I have some ideas here too but they depend on your skill-level.