Shannon-Kotel'nikov mappings in joint source-channel coding
IEEE Transactions on Communications
Cascade multiterminal source coding
ISIT'09 Proceedings of the 2009 IEEE international conference on Symposium on Information Theory - Volume 2
Uncoded Transmission Is Exactly Optimal for a Simple Gaussian “Sensor” Network
IEEE Transactions on Information Theory
Hi-index | 0.00 |
We investigate lossy transmission of two intercorrelated sources over a two-hop cascade network. A low delay analogue based joint source-channel coding scheme is proposed for the case of jointly Gaussian sources over channels with additive white Gaussian noise (AWGN), subject to a quadratic distortion constraint. For the first hop, uncoded transmission is used, with an optimal minimum mean-squared error (mmse) decoder at the relay node that utilizes the other source as side information. For the second hop, we first perform decorrelation using Karhunen-Loève Transform, followed by dimension reduction mappings. Simulation results show that relative to the optimal performance theoretically attainable (OPTA), uncoded transmission with the optimal mmse joint decoder for the first hop is able perform close to OPTA when the intercorrelation is relatively weak. For the second hop, result of the proposed scheme is able to come within 1.5dB from OPTA.