Sum-rate maximization for cognitive radios in multi-user MIMO-OFDM context

• Published in: 2018 International Conference on Electrical Sciences and Technologies in Maghreb (CISTEM) pp. 1 - 5
• Main Authors: Labdaoui, Rym, Ghanem, Khalida, Ettoumi, Fatiha Youcef, Belaoura, Widad
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.10.2018
• Subjects: Cognitive radio; MIMO communication; MIMO-OFDM; OFDM; Optimization; resource allocation; Resource management; Signal to noise ratio; stochastic algorithms; Stochastic processes; UWB
• DOI: 10.1109/CISTEM.2018.8613482

An efficient allocation of the radio resources is an important issue in multi-user orthogonal frequency division multiplexing (MU-OFDM) systems. In this paper, we investigate the downlink resource allocation problem (subcarrier and power allocation) in MU-OFDM-based underlay cognitive radios networks (CRNs) using multipleinput-multiple-output (MIMO) configurations, referred to as MU-MIMO-OFDM in both underlay and overlay setups. For this, stochastic algorithms are adopted to maximize the weighted sum-rate capacity of the secondary users (SUs) over the eigen-channels obtained through the singular value decomposition (SVD) of the MIMO channel matrix, subject to the average power and the probability of interference constraints. The optimal resource allocation policy is achieved with the maximization of the user-quality indicator, which is a function of the rate of the SUs, the activity of the primary users (PUs), and the Lagrange multipliers associated with the considered constraints. The three diversities (spatial, frequency and users) are exploited by these stochastic algorithms in MUMIMO-OFDM configuration, and the performance in terms of the weighted sum-rate capacity is analysed and compared to the MU-OFDM alternative using a single transmit and receive antennas (SISO). Adopting MIMO configuration in MUOFDM system, with adaptive resource allocation over the resulting eigen-channels, is shown to significantly outperform the SISO-based counterpart.