Resilient and Federated Acoustic Defence Approaches for Maritime Security (Ref: SF-KK-2025)

About the Project

We are seeking an outstanding and highly motivated PhD candidate to join the Wolfson School of Mechanical, Electrical and Manufacturing Engineering. This project will investigate next-generation machine-learning methods for detecting small, covert maritime threats using distributed acoustic sensing, while ensuring that sensitive underwater recordings never leave their owners.

Modern maritime security zones face growing challenges from quiet, low-signature threats such as USVs, UUVs and divers, where radio-frequency cues are often weak or deliberately suppressed. Acoustic sensing is therefore essential, yet the most informative underwater recordings are scattered across navy bases, coast guards, port authorities and critical-infrastructure operators who cannot share raw audio for sovereignty, commercial or security reasons. This results in a capability gap: all stakeholders need strong detection models, but no one can centralise sensitive data.

Federated learning helps close this gap by allowing each organisation to train models locally and share only model parameters, enabling collaboration while keeping audio in place. However, decentralisation brings major challenges: highly variable acoustic environments shaped by bathymetry, traffic noise, oceanography and bio-acoustics make classification difficult, and the learning process itself becomes an attack surface. Adversaries may poison local datasets, craft malicious model updates that appear benign, or implant backdoors triggered by rare acoustic patterns, requiring robust, uncertainty-aware defences.

To this end, this PhD will design and evaluate an uncertainty-aware, robust to adversarial ML attacks federated learning framework for maritime acoustic detection and classification across heterogeneous, multi-owner hydrophone networks (and, where available, distributed acoustic sensing). The work will involve developing:

• Global models with site-specific personalisation to reduce false alarms and improve trust.
• Uncertainty-aware inference and calibrated decision support, enabling operators to set thresholds appropriate to local noise and risk.
• Red-team attack models that simulate realistic poisoning, backdoors, and model-update manipulations under operational constraints.
• Blue-team defence mechanisms, including robust aggregation, client-reputation tracking, update forensics, and selective “unlearning”.
• Evaluation testbeds to measure detection accuracy, calibration, robustness under attack, and communication efficiency.

The successful candidate will have the opportunity to work with leading researchers in the field and to present their work at international conferences and publish in high-impact journals. Furthermore, there will be an opportunity to attend summer schools and Continuing Professional Development courses. The position has a flexible starting time.

The student will work with Dr Kostas Kyriakopoulos (ML, decision making under uncertainty) and Professor Paul Lepper (underwater acoustics, signal processing). To apply, please submit your CV, a cover letter, and contact details of two referees. Please contact the primary supervisor, Dr Kostas Kyriakopoulos for more details and for raising your interest.

Name of primary supervisor/CDT lead:

Dr. Kostas Kyriakopoulos elkk@lboro.ac.uk

https://www.lboro.ac.uk/schools/meme/staff/konstantinos-kyriakopoulos/

Name(s) of secondary supervisor(s) if known:

Prof. Paul Lepper

https://www.lboro.ac.uk/schools/meme/staff/paul-lepper/

Entry requirements:

The successful applicant should have a 1st class or high 2:1 honours (or equivalent) degree in electronic/electrical engineering, computer science or a closely related discipline. An MSc with Distinction is desirable. Strong research abilities with appropriate coding skills are required.

The ideal candidate will have strong programming skills (e.g., Python or Matlab) and experience in one or more of the following areas:

• Machine learning or signal processing
• Acoustic sensing, sonar, or maritime systems
• Distributed/federated learning

The successful candidate should be an enthusiastic team player who can work both independently and collaboratively.

English language requirements:

Applicants must meet the minimum English language requirements.

Bench fees required: No

Closing date of advert: 15 June 2026

Start date: 01 April 2026, 01 July 2026, 01 October 2026

Full-time/part-time availability: Full-time 3 years

Fee band: 2025/26 Band RB (UK £5,006, International £28,600)

How to Apply:

All applications should be made online. Under Campus, please select Loughborough and select Programme Electronic, Electrical and Systems Engineering. Please quote the advertised reference number SF-KK-2025 in your application under the ‘Finance’ section.

Applications must include a personal statement, up-to-date curriculum vitae (CV), details of two referees (one from your highest degree qualification), certified certificates and transcripts for all completed degree programmes, and a reference to the project ‘SF-KK-2025’. Submission of a research proposal is not essential but may strengthen your application.

To avoid delays in processing your application, please ensure that you submit the minimum supporting documents.

Project search terms:

acoustics, acoustics engineering, cyber security, machine learning

Email address Wolfson:

ws.phdadmin@lboro.ac.uk