Learning-Based Signal Detection for Wireless OAM-MIMO Systems With Uniform Circular Array Antennas

Norifumi Kamiya

doi:10.1109/access.2020.3043004

Abstract

This paper presents a neural-like network-based signal detection method for orbital angular momentum multiplexing systems with uniform circular array antennas. The signal detection network is derived by unfolding the alternating direction method of multipliers (ADMM), and in addition, a parallel interference cancellation (PIC) function is integrated, which enhances the tolerance to inter-mode interference while keeping the complexity feasible. The number of parameters to be learned in each layer of the network is a linear order of the number of antenna elements. Simulation results show that the ADMM-PIC detector exhibits excellent error performance, which cannot be achieved by a conventional minimum mean square error-based detector.

Highlights

I MPROVING data transmission capacity with multi-input and multi-output (MIMO) multiplexing technologies [1] has become increasingly important for wireless communication systems to meet continuous demand for higher data rates over limited frequency resources
In this paper, we investigated the channel capacity for UCAbased orbital angular momentum (OAM)-MIMO systems and demonstrated using examples that the mode-wise minimum mean square error (MMSE) capacity and the MMSE detector performance deteriorate seriously due to inter-mode interference that occurs when Tx and Rx uniform circular array antenna (UCA) are not ideally aligned
We presented a learning, neural-like network-based signal detection method, which is derived by integrating unfolded alternating direction method of multipliers (ADMM) with parallel interference cancellation (PIC)

Summary

INTRODUCTION

I MPROVING data transmission capacity with multi-input and multi-output (MIMO) multiplexing technologies [1] has become increasingly important for wireless communication systems to meet continuous demand for higher data rates over limited frequency resources. Inspired by the work in [19]–[21], we present a learning, neural-like network-based signal detection method for UCAbased OAM-MIMO transmission systems. The main contributions of this paper can be summarized as follows: (i) the channel capacity is investigated for multi-ring UCA-based OAMMIMO multiplexing with mode-wise MMSE detectors. (ii) In order to improve the detector performance, a neural-like network-based signal detector is presented for the OAM-MIMO systems, which is derived by unfolding the ADMM [22]. The rest of this paper is organized as follows: In Section II, we briefly review the UCA-based OAM-MIMO transmission model and investigate its Shannon capacity and mode-wise MMSE capacity. W (N−1)), and the computational complexity of the mode-wise MMSE detector becomes O(NM 3) In this case, the OAM-MIMO channel can be thought of as a collection of N parallel M×M

CHANNEL CAPACITY

NETWORK TRAINING

SIMULATION RESULTS

CONCLUSION

MMSE-PIC