As communication systems embrace ever wider bandwidths, analog-to-digital converters (ADCs) struggle to meet rate, resolution, and power requirements. The problem is exacerbated by the massive antenna arrays under consideration for next-generation wireless, which imply tens or even hundreds of receive channels. In response, this project develops new communication systems that operate with very low-resolution (e.g., 1-3 bit) ADCs. In particular, the investigators derive fundamental limits on quantized many-antenna communication and design advanced communications strategies to approach those limits.
The use of low-resolution ADCs radically changes both the theory and practice of communication, motivating a thorough re-examination of capacity bounds, modulation and coding designs, receiver processing algorithms, and limited-feedback strategies. The investigators address these topics in the context of multiple-input multiple-output systems with massive arrays. In particular, they derive theoretical capacity (or achievable rate) bounds that account for coarse ADC quantization, imperfect channel state information (CSI) at the receiver, partial CSI at the transmitter, and multi-user interference. They also characterize the mean-squared error achievable for sparse-channel estimation and the symbol-error rate achievable via equalization, both in the large-system limit. In addition, they develop limited-feedback strategies to provide transmitter CSI to the transmitter, transmitter-precoding designs that exploit that CSI, and optimized training sequences. Furthermore, they develop channel-estimation algorithms that learn and exploit channel sparsity; equalization (and/or multi-user detection) algorithms that take channel-estimation error into account; and efficient strategies for joint channel-estimation, equalization, and decoding. Finally, they develop optimization and adaptation strategies for the ADC thresholds, and blind calibration strategies that account for ADC imperfections. The research methodology leverages recent results from wireless communication theory, one-bit compressed sensing, and approximate message passing.
