Forward-Aware Information Bottleneck-Based Vector Quantization: Multiterminal Extensions for Parallel and Successive Retrieval

Consider the following setup: Through a joint design, multiple observations of remote data source shall be xmlns:xlink="http://www.w3.org/1999/xlink">locally compressed before getting transmitted via several xmlns:xlink="http://www.w3.org/1999/xlink">error-prone , rate-limited forward links to (distant) processing unit. For addressing this specific instance multiterminal xmlns:xlink="http://www.w3.org/1999/xlink">Joint Source-Channel Coding problem, in article, foundational principle xmlns:xlink="http://www.w3.org/1999/xlink">Information Bottleneck method is fully extended obtain purely statistical design approaches, enjoying xmlns:xlink="http://www.w3.org/1999/xlink">Mutual Information as their fidelity criterion. Specifically, forms stationary points for two types distributed compression schemes are characterized here. Subsequently, those acquired solutions utilized centerpiece proposed generic, iterative algorithm, termed xmlns:xlink="http://www.w3.org/1999/xlink">Multiterminal Forward-Aware Vector Information Bottleneck (M-FAVIB) optimizations. Leveraging an unfolding trick, it will proven that both fall into category xmlns:xlink="http://www.w3.org/1999/xlink">Successive Upper-Bound Minimization ensuring convergence point. Eventually, effectiveness substantiated well by means numerical investigations over some typical transmission scenarios.