Deep-Learning-Assisted IoT-Based RIS for Cooperative Communications
Bulent Sagir, Erdoğan Aydın, Hacı İlhan
Abstract
Reconfigurable intelligent surfaces (RISs) are software-controlled passive devices that can be used as relay <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(R)$ </tex-math></inline-formula> systems to reflect incoming signals from a source <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(S)$ </tex-math></inline-formula> to a destination <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(D)$ </tex-math></inline-formula> in a cooperative manner with optimum signal strength to improve the performance of wireless communication networks. The configurability and flexibility of an RIS deployed in an Internet of Things (IoT)-based network can enable network designers to devise stand-alone or cooperative configurations that have considerable advantages over conventional networks. In this article, two new deep neural network (DNN)-assisted cooperative RIS (CRIS) models, namely, DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R} -$ </tex-math></inline-formula> CRIS and DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R, D} -$ </tex-math></inline-formula> CRIS, are proposed for cooperative communications. In DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R} -$ </tex-math></inline-formula> CRIS model, the potential of RIS deployment as an IoT-based relay element in a next-generation cooperative network is investigated using deep-learning (DL) techniques for RIS phase optimization. In addition, to reduce the maximum-likelihood (ML) complexity at <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$D$ </tex-math></inline-formula> , a new DNN-based symbol detection method is presented with the DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R, D} -$ </tex-math></inline-formula> CRIS model combined with DNN-assisted phase optimization. For a different number of relays and receiver configurations, the bit error rate (BER) performance results of the proposed DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R} -$ </tex-math></inline-formula> CRIS and DNN <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$_{R, D} -$ </tex-math></inline-formula> CRIS models and traditional CRIS scheme (without a DNN) are presented for a multirelay cooperative communication scenario with path loss effects. It is revealed that the proposed DNN-based models show promising results in terms of BER, even in high-noise environments with low system complexity.