Jump to content
Main menu
Main menu
move to sidebar
hide
Navigation
Main page
Recent changes
Random page
Help about MediaWiki
Edge Computing Wiki
Search
Search
Appearance
Create account
Log in
Personal tools
Create account
Log in
Pages for logged out editors
learn more
Contributions
Talk
Editing
Federated Learning
(section)
Page
Discussion
British English
Read
Edit
View history
Tools
Tools
move to sidebar
hide
Actions
Read
Edit
View history
General
What links here
Related changes
Upload file
Special pages
Page information
Appearance
move to sidebar
hide
Warning:
You are not logged in. Your IP address will be publicly visible if you make any edits. If you
log in
or
create an account
, your edits will be attributed to your username, along with other benefits.
Anti-spam check. Do
not
fill this in!
== 5.5.4 Sybil and Free-Rider Attacks == Sybil attacks occur when a single adversary creates multiple fake clients (Sybil nodes) to manipulate the FL process. These clients may collude to skew aggregation results or overwhelm honest participants. This is especially dangerous in decentralized or large-scale FL environments, where authentication is weak or absent.<sup>[1]</sup> Without strong identity verification, an attacker can inject numerous poisoned updates, degrading model accuracy or blocking convergence entirely. Traditional defenses like IP throttling or user registration may violate privacy principles or be infeasible at the edge. Cryptographic registration, proof-of-work, or client reputation scoring have been explored, but each has trade-offs. Clustering and anomaly detection can identify Sybil patterns, though adversaries may adapt their behavior to avoid detection.<sup>[4]</sup> Free-rider attacks involve clients that participate in training but contribute little or nothing—e.g., sending stale models or dummy gradients—while still downloading and benefiting from the global model. This undermines fairness and wastes resources, especially on networks where honest clients use real bandwidth and energy.<sup>[3]</sup> Mitigation strategies include contribution-aware aggregation and client auditing.
Summary:
Please note that all contributions to Edge Computing Wiki may be edited, altered, or removed by other contributors. If you do not want your writing to be edited mercilessly, then do not submit it here.
You are also promising us that you wrote this yourself, or copied it from a public domain or similar free resource (see
Edge Computing Wiki:Copyrights
for details).
Do not submit copyrighted work without permission!
Cancel
Editing help
(opens in new window)