Multi-user Information Retrieval via a Multi-antenna Relay: Network Coding and Transceiver Design

We propose a novel coding framework for multi-user information retrieval in a multi-input single-output (MISO) setup, assisted by a multi-antenna base station (BS) that operates as a half-duplex amplify-and-forward relay without direct access to the distributed data. In the uplink (UL) phase, users first transmit missing data fragments to the BS. Using optimized UL combiners and downlink (DL) precoders, the BS forms multicast codewords and forwards tailored combinations of these components in DL so each user can recover its desired data. Our model generalizes the multi-group multi-way relaying, where unlike the existing methods that use analog network coding (ANC) in DL, we employ ANC in both UL and DL. Inspired by coded caching (CC), our model optimizes UL/DL transmissions and transceivers to minimize end-to-end (E2E) delivery time. As the E2E optimization is intractable, we decouple it into two subproblems: first, design the UL combiners and transmit powers to maximize the minimum effective signal-to-interference+noise ratio (SINR) of the data fragments included in the multicast codewords required for DL; second, using optimized UL parameters, design the DL precoders to maximize the achievable SINR of data fragments in DL, resulting in minimizing the E2E delay. At high signal-to-noise ratio (SNR), the degree-of-freedom for UL and DL matches the optimal one achievable in a conventional MISO-CC system with centralized BS access. Comparative finite-SNR evaluations show that, by exploiting spatial multiplexing and CC gains, our model consistently outperforms baseline schemes, achieving minimum E2E delay, or maximum E2E rate. © 1972-2012 IEEE.