https://imt-atlantique.hal.science/hal-03675543Song, ShulinShulinSongHokkaido University [Sapporo, Japan]He, JiguangJiguangHeCWC - Centre for Wireless Communications [University of Oulu] - University of OuluMatsumoto, TadTadMatsumotoIMT Atlantique - MEE - Département Mathematical and Electrical Engineering - IMT Atlantique - IMT Atlantique - IMT - Institut Mines-Télécom [Paris]JAIST - Japan Advanced Institute of Science and TechnologyRate-Distortion and Outage Probability Analyses of Wyner-Ziv Systems over Multiple Access ChannelsHAL CCSD2021Wyner-Ziv problemmultiple access channeloutage probabilityhelper-selectionrate distortion analysis[MATH.MATH-IT] Mathematics [math]/Information Theory [math.IT]Dupraz, Elsa2022-05-23 11:16:342022-08-05 14:54:522022-05-30 08:20:22enJournal articleshttps://imt-atlantique.hal.science/hal-03675543/document10.1109/TCOMM.2021.3087128application/pdf1In this paper, we conduct rate-distortion and outage probability analyses for Wyner-Ziv (WZ) systems, where the information sequences from a source and a helper are transmitted to the destination over block Rayleigh fading multiple access channels (MACs). This work has been motivated by decisionmaking systems, where the aim is to make right decisions based on the observations. Hence, the most critical performance metric is the accuracy of decisions, even allowing distortion in wireless transmission. A sufficient condition for the successful transmissions is that the WZ and MAC regions intersect. For the ease of calculating the outage probability, we propose an approximated, yet accurate, WZ rate region. The outage probabilities of the WZ systems are then evaluated both in orthogonal and MAC transmissions. It is shown that the transmission efficiency of the system with MAC is significantly improved compared to orthogonal transmission. Furthermore, a helper-selection scheme is introduced to further reduce the outage probability of the WZ-MAC systems. The results show that the outage probability decreases, of which decay corresponds to the (L + 1)-order diversity with L helpers in small value range of the average signal-to-noise ratio (SNR), while it converges into L as the average SNR becomes large.