A White Paper by Dr David Soldani, CISO, Rakuten Symphony
Download the whole white paper here. Below is a short excerpt:
In general, 6G wireless is projected to be secure by design. to shift from a security-enhanced network to a security-by-design system, 6G needs to integrate security at the heart of the infrastructure and instil the whole network, end to end, with a defence-in-depth strategy. Also, the standardisation process for 6G must provide new mechanisms for security control, security assurance and privacy preservation.
As communications networks evolve, it is expected that there will be an increased reliance on artificial intelligence-enabled smart applications requiring situational, context-aware and customised privacy solutions. Hence, the 5G privacy-preserving approach may not be well suited for future wireless applications due to a diverse and complex set of novel privacy challenges.
One potential solution is the use of pairs of deep neural networks, which can be trained with differential privacy, a formal privacy framework that limits the likelihood that queries of personal identifiable information – sensitive data that can include, for example, the full name of a person, their social security number, driver’s license, financial information, medical records, etc – could identify a real data subject.
The concepts related to federated learning (FL) are also active topics in the research community for ensuring privacy protection. FL is a distributed machine learning technique that allows model training for large amounts of data generated locally; the required modelling is done by each individual learner in the federation. Instead of sending a raw training dataset, each individual learner transmits their local model to an ‘aggregator’ to build a global model.
To read the full article, download the pdf here.
