Minimum sum rate distortion based distributed source compression

In this paper, we consider a topology composed of two correlated source nodes and one aggregation node. Discrete i.i.d samples of each source are assumed. To take advantage of the correlation between two sources, we propose an asymmetric compression scheme, which is motivated by typical set deterministically rather than probabilistically. Block fading and AWGN are assumed for the both source-destination links and global CSI is known at the receiver. Since the most critical concerns of this system rest with source compression and channel bandwidth ratio, which are connected by rate distortion, we optimize power allocation between sources such that the sum rate distortion and decoding error probability can be minimized. Further, optimal bandwidth ratio can be found to maximize instantaneous transmission rate. We give a linear correlation example and the simulation results have demonstrated that our scheme can reach a compression rate of over 85% when correlation is relatively weak.