The CEO problem with secrecy constraints

A lossy source coding problem with secrecy constraints is considered where a remote information source should be transmitted to a single destination via multiple agents in the presence of an eavesdropper. The agents observe noisy versions of the source and independently encode and transmit their observations to the destination via noiseless rate-limited links. Unbeknownst to the agents, an eavesdropper intercepts one of the links from the agents to the destination to learn as much as possible about the source. The destination should estimate the remote source subject to a mean distortion threshold. This problem can be viewed as the CEO problem with addition of secrecy constraints. We establish inner and outer bounds on the rate-distortion-equivocation region. In addition, we provide the optimal rate-distortion-equivocation region for the quadratic Gaussian case when the eavesdropper has no side information.