DBST-FL: Dynamic Behavioural and Semantic Trust for Robust Federated Learning in Industrial IoT
Ammar Alazab, Abin Kumbalapalliyil Tom, Tony Jan, Md Whaiduzzaman, Thien Nguyen, Ansam Khraisat, Ali Almazrouei · 2026
Read DBST-FL: Dynamic Behavioural and Semantic Trust for Robust Federated Learning in Industrial IoT by Ammar Alazab, Abin Kumbalapalliyil Tom, Tony Jan, Md Whaiduzzaman, Thien Nguyen, Ansam Khraisat, Ali Almazrouei (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13468717.txt.
Description
Ammar Alazab and colleagues propose in 2026 a trust system to defend federated learning in industrial Internet of Things networks. In such networks, factories train a shared model without shipping raw data to a center. The weak point is poisoned clients quietly corrupting the shared model, including stealthy backdoor attacks that hide from simple checks. Most defenses watch only the updates behavior or only the model performance; this framework scores both.
A behavioral layer grades each client on gradient alignment, consistency with its own history, and deviation from the crowd. A semantic layer separately tests what the client model actually does, using stress validation on server data. The two scores are multiplied, not averaged, into one trust value, so a client cannot hide a rotten dimension behind a good one. Strong factories keep their data private while the network expels saboteurs, which is the bargain the design protects.
Catalog record
| CREATOR | Ammar Alazab, Abin Kumbalapalliyil Tom, Tony Jan, Md Whaiduzzaman, Thien Nguyen, Ansam Khraisat, Ali Almazrouei |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51764 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |