New method keeps data private in large-scale distributed computing systems

Researchers have introduced a new privacy-preserving method for multi-access distributed computing, a model that improves on frameworks like MapReduce. The work develops private coded schemes for two specific connectivity models, using new families of extended placement delivery arrays. These schemes guarantee the privacy of each reducer's assigned function, protecting sensitive information during large-scale data processing tasks where dedicated mapper and reducer nodes work separately to reduce communication bottlenecks.