Multiple-User Cooperative Communications Based on Linear Network Coding

TLDR

The network comprises multiple users with independent data communicating to a common base station over block‑fading channels, where link errors cause a dynamic topology. The authors aim to develop diversity network codes that exploit cooperative diversity and time‑varying fading to improve reliability. They design linear network codes that enable the base station to recover all user messages from the fewest possible coded blocks, with simplified variants reducing design complexity. They prove deterministic DNCs exist, achieve diversity order 2M − 1 surpassing non‑coding or binary coding schemes, and simulations confirm substantial performance gains.

Abstract

We propose a new scheme for cooperative wireless networking based on linear network codes. The network consists of multiple (M ≥ 2) users having independent information to be transmitted to a common basestation (BS), assuming block-fading channels with independent fading for different codewords. The users collaborate in relaying messages. Because of potential transmission errors in links, resulting in erasures, the network topology is dynamic. To efficiently exploit the diversity available by cooperation and time-varying fading, we propose the use of diversity network codes (DNCs) over finite fields. These codes are designed such that the BS is able to rebuild the user information from a minimum possible set of coded blocks conveyed through the dynamic network. We show the existence of deterministic DNCs. We also show that the resulting diversity order using the proposed DNCs is 2 M - 1, which is higher than schemes without network coding or with binary network coding. Numerical results from simulations also show substantial improvement by the proposed DNCs over the benchmark schemes. We also propose simplified versions of the DNCs, which have much lower design complexity and still achieve the diversity order 2 M - 1.