You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Provide the MATSim version you are using. You can find the version in the pom.xml. E.g. 2025.0.
MATSim version 2025.0.
Did you build a minimal example to reproduce the issue?
Yes!
Describe the problem you are facing.
Hello! I'm researching how privacy-enhancing technologies — such as differential privacy, k-anonymity, federated learning, secure computation, synthetic data generation, and others — can be applied in transport simulation contexts, particularly involving MaaS. I'm interested in using MATSim to model and benchmark the effects of these mechanisms on simulation quality and agent behaviour.
A few questions:
Has anyone experimented with applying any form of PET to MATSim — not only on the demand side (e.g., anonymising population plans, travel demand matrices, activity chains) but also on the agent behaviour side? For instance, if agents dynamically replan or make mode choices during execution, could privacy constraints be applied to the information available to them or to the data they share with other agents or operators?
In scenarios where MaaS operators or service providers are modelled as agents (e.g., in DVRP or MaaS extensions), has anyone explored what happens when those actors operate under privacy restrictions — such as limited access to traveller data, or receiving only aggregated or noisy demand signals rather than individual trip requests?
What would be the most natural extension points in MATSim's architecture for introducing privacy mechanisms — whether at the input stage, within agent scoring/replanning, or at the operator/service level?
What output metrics or analysis tools would be most appropriate for benchmarking the impact? For example, link volumes, mode shares, travel times, score convergence, service efficiency, demand fulfilment rates — what do people typically use to evaluate simulation fidelity and agent performance when inputs or information flows have been modified?
Any pointers to related work, extensions, or suggestions would be greatly appreciated. Thank you!
If possible, provide a link to your minimal example. This might be a GitHub repository.
No response
If you want, provide some background information about you and your project. This might help us understand your issue better.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Provide the MATSim version you are using. You can find the version in the
pom.xml. E.g.2025.0.MATSim version 2025.0.
Did you build a minimal example to reproduce the issue?
Describe the problem you are facing.
Hello! I'm researching how privacy-enhancing technologies — such as differential privacy, k-anonymity, federated learning, secure computation, synthetic data generation, and others — can be applied in transport simulation contexts, particularly involving MaaS. I'm interested in using MATSim to model and benchmark the effects of these mechanisms on simulation quality and agent behaviour.
A few questions:
Has anyone experimented with applying any form of PET to MATSim — not only on the demand side (e.g., anonymising population plans, travel demand matrices, activity chains) but also on the agent behaviour side? For instance, if agents dynamically replan or make mode choices during execution, could privacy constraints be applied to the information available to them or to the data they share with other agents or operators?
In scenarios where MaaS operators or service providers are modelled as agents (e.g., in DVRP or MaaS extensions), has anyone explored what happens when those actors operate under privacy restrictions — such as limited access to traveller data, or receiving only aggregated or noisy demand signals rather than individual trip requests?
What would be the most natural extension points in MATSim's architecture for introducing privacy mechanisms — whether at the input stage, within agent scoring/replanning, or at the operator/service level?
What output metrics or analysis tools would be most appropriate for benchmarking the impact? For example, link volumes, mode shares, travel times, score convergence, service efficiency, demand fulfilment rates — what do people typically use to evaluate simulation fidelity and agent performance when inputs or information flows have been modified?
Any pointers to related work, extensions, or suggestions would be greatly appreciated. Thank you!
If possible, provide a link to your minimal example. This might be a GitHub repository.
No response
If you want, provide some background information about you and your project. This might help us understand your issue better.
No response
Beta Was this translation helpful? Give feedback.
All reactions