Excited to Share Our IEEE Transactions on Information Theory Publication!

About

🎺 Excited to share that our paper “Performance Analysis of Joint Antenna Selection and Precoding Methods in Multi-User Massive MISO”, co-authored with Xiuxiu Ma, Dr Abla Kammoun, Prof. Tareq Al-Naffouri, and Prof. Mohamed-Slim Alouini, has been accepted for publication in IEEE Transactions on Information Theory! This is a truly meaningful gift for my graduation.

📚 It is well known that applying an ℓ₁-norm penalty on precoded vectors encourages sparsity, which reduces the number of active transmit antennas. However, this penalty also introduces distortion and degrades performance, especially when the solution becomes highly sparse. To address this, we reduce the ℓ₁-norm penalty and introduce a thresholding step to transmit only the dominant elements. While the idea is intuitive, the asymptotic analysis becomes challenging due to the non-linear thresholding operation.

💡 In this work, we systematically analyze this problem and provide a precise asymptotic characterization.
🌟 We establish a mapping between a desired sparsity level and the threshold for a given ℓ₁-norm penalty.
🌟 We show how performance depends on system parameters, and demonstrate that combining thresholding with the ℓ₁-norm penalty outperforms using the ℓ₁-norm alone. The optimal “threshold + ℓ₁-norm” pair in terms of performance can be obtained, and we further reveal how this optimal performance varies with the transmit budget.
🌟 To derive these results, we revisit Gordon’s Comparison Theorem (1985) and extend it into a novel Gaussian Min-Max Theorem. In our 51-page, double-column journal article, we detail the complete methodology leading to our conclusions. We hope this approach will inspire future work on characterizing large-dimensional systems with non-linear post-processing.

👉 Read the full paper here: Performance Analysis of Joint Antenna Selection and Precoding Methods in Multi-user Massive MISO | IEEE Journals & Magazine | IEEE Xplore