PrivacyAnalyzer is a commercial product of LamdaNetworks www.lamdanetworks.io.
It implements an AI Spark-based analytics engine which offers the functionality for assessing the privacy strength of applications or services against confidential information disclosure. It can detect issues like finding email addresses, GPS coordinates, names, cities, and many more patterns and allows the user to (i) either deidentify discovered patters or (ii) anonymize them with e-differential privacy algorithms, e.g., adding such ‘noise’ to GPS coordinates so that location services may still be provided with high accuracy.
The SaaS high level architecture shown below is proven in commercial deployments that it scales in respect to the number of incoming messages’ volume and in respect to the number of fully isolated tenants.
Within the context of 5GASP, PrivacyAnalyzer’s codebase has been refactored so that:
- PrivacyAnalyzer is a cross-vertical Network Application, with subscribers being other privacy-sensitive Network Applications;
- PrivacyAnalyzer is integrated with NEF emulators across 5GASP testbeds.
Adhering to the interim review comments, the code enhancements have resulted that PrivacyAnalyzer acts like a ‘Intrusion Detection System (IDS)’ within the 5G Core of the testbeds of the project. As an example, the PrivacyAnalyzer deployment shown below is currently used in the UoP testbed to assess the privacy strength of the network messages stemming from the NEF emulator which operates within the testbed.
The angular screenshot depicts that the UE with supi 202010000000001 transmits -from a certain endpoint – sensitive information that may need to be anonymized or suppressed. Specifically, each message from this endpoint contains unencrypted data: (a) an email address and (b) two GPS coordinates, the longitude and the latitude of the UE.