CIF: Small: Exploiting Interference via Data-Dependent Precoding

点击翻译按钮获取中文摘要

In wireless communications, managing interference is a key problem that has been studied for many years. When a cell tower or a WiFi access point sends signals simultaneously to several users, it has to ensure that these signals do not get so mixed up with each other that they can't be separated by the individual receivers. The most common approach is to encode the signals in a way that attempts to completely eliminate this multi-user interference, usually by assigning each signal to an "orthogonal" channel, e.g., sending the signals at different times, on different frequency bands, or using different transmission beams. The goal of this project is to exploit the observation that not all interference is "bad." When the signals are confined to a limited alphabet, it is unnecessary that the signal be received in exactly the form it was transmitted; it is only necessary that the receiver properly decode which symbol from the alphabet of signals was transmitted. While multi-user interference inevitably distorts the waveform, the transmitter can encode the waveform in such a way that the interference does not prevent (and in fact can enhance) the ability of the receivers to correctly decode their respective symbols. The key advantage of this approach is that it enables the transmitter to use much less power to get the same performance, since the interference essentially serves the purpose of adding extra power to the desired signals for each receiver. The significant energy savings that can result from this approach could have a revolutionary effect on the performance of wireless systems, and enable a much wider deployment of wireless network infrastructure and IoT devices at a fraction of the currently required energy consumption. Multi-antenna implementations have become standard in today’s WiFi and cellular communication networks, and are one of the key technologies for achieving the large throughputs and high reliability required by next-generation systems. This proposal is focused on a new symbol-level precoding paradigm that has recently emerged in which not only the channel state information is exploited, but also knowledge of the symbols to be transmitted. This approach provides a powerful extra dimension for optimization that can yield dramatic improvements in performance. While most precoding methods try to eliminate interference, symbol-level precoding (SLP) exploits useful or "constructive" interference and repurposes it as energy for the desired signals. This increases the robustness of the signal detection and enables wireless systems to operate with significantly less power, and thus makes constructive interference SLP a promising candidate for low-cost and high-reliability applications. This project seeks to study methods for reducing the complexity of SLP, operation of the algorithms in scenarios with constrained radio-frequency front ends (e.g., low-resolution quantization, per-antenna power constraints), physical layer security, user selection in large networks, theoretical analyses of performance, network settings other than broadcast, etc.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.