JOINT POWER COST AND LATENCY MINIMIZATION FOR SECURE COLLABORATIVE LEARNING SYSTEM

This work investigates the model update security in a collaborative learning or federated network by using covert communication. The CC uses jamming signal and multiple friendly jammers (FJs) are deployed that can offer services to owner, i.e., base station (BS). To enable BS select best FJ, lowest cost truthful auction is adopted. Then, problem formulated optimize power, transmission local accuracy. objective minimize training latency, subject performance requirement budget of BS. solve non-convex problem, we adopt Successive Convex Approximation algorithm. simulation results reveals some interesting things. For example, trustful reduces as number FJs increases.