On Secure Distributed Source Coding

Exploiting correlated observations at multiple remote nodes for generating a secret common randomness to facilitate secure communication of data between the terminals has been addressed by many authors [1][2][3]. But in many applications such as sensor networks, the correlated observations might themselves be the data that need to be communicated. We address this problem of secure distributed source coding where we want to leak the least amount of information about the sources to an eavesdropper who may have access to a correlated observation of its own and who can listen in on the communication between the distributed encoders. The focus here is on minimizing the amount of information revealed to the eavesdropper rather than on minimizing the rate required to effect the communication.