Deep Learning Based Successive Interference Cancellation for the Non-Orthogonal Downlink

Abstract

Non-orthogonal communications are expected to play a key role in future wireless systems. In downlink transmissions, the data symbols are broadcast from a base station to different users, which are superimposed with different power to facilitate high-integrity detection using successive interference cancellation (SIC). However, SIC requires accurate knowledge of both the channel model and channel state information (CSI), which may be difficult to acquire. We propose a deep learningaided SIC detector termed SICNet, which replaces the interference cancellation blocks of SIC by deep neural networks (DNNs). Explicitly, SICNet jointly trains its internal DNNaided blocks for inferring the soft information representing the interfering symbols in a data-driven fashion, rather than using hard-decision decoders as in classical SIC. As a result, SICNet reliably detects the superimposed symbols in the downlink of non-orthogonal systems without requiring any prior knowledge of the channel model, while being less sensitive to CSI uncertainty than its model-based counterpart. SICNet is also robust to changes in the number of users and to their power allocation. Furthermore, SICNet learns to produce accurate soft outputs, which facilitates improved soft-input error correction decoding compared to model-based SIC. Finally, we propose an online training method for SICNet under block fading, which exploits the channel decoding for accurately recovering online data labels for retraining, hence, allowing it to smoothly track the fading envelope without requiring dedicated pilots. Our numerical results show that SICNet approaches the performance of classical SIC under perfect CSI, while outperforming it under realistic CSI uncertainty.

Enhanced deep soft interference cancellation for multiuser symbol detection

“…Using this technique, the receiver first detects a symbol with the maximum channel gain, subtracts the detected symbol from the received signal, and then sequentially detects the remaining symbols. Following the basic concept of HIC, [23,24] proposed DL approaches for downlink signal detection in MU-MIMO-NOMA. In both [23] and [24], only the downlink environment was considered, particularly in [23], where only two users were assumed.…”

“…Following the basic concept of HIC, [23, 24] proposed DL approaches for downlink signal detection in MU‐MIMO‐NOMA. In both [23] and [24], only the downlink environment was considered, particularly in [23], where only two users were assumed. In [24], it was extended to users.…”

“…Many researchers in the field of wireless communication have attempted to apply DL to communication technologies [7][8][9][10]. These attempts have included the use of neural networks to design channel estimators [11], channel denoisers [12], channel decoders [13,14], and end-to-end transceivers [15][16][17][18] to manage multiuser interference [19][20][21][22] and design multiuser symbol detectors [23][24][25]. In this study, we focused on the use of DL for joint-symbol detection in multiuser MIMO systems.…”

Outage Probability and Ergodic Capacity Analysis of MIMO-NOMA Heterogeneous Network for 5G System

“…These signals' power allocation coefficients are different. A Successive Interference Cancellation (SIC) approach is used on the receiver side to separate overlapping signals [25]. The user channel conditions are used to determine the power allocation coefficients.…”

Adaptive Data Augmentation for Deep Receivers

“…An alternative approach designs receivers with compact task-specific DNNs such that they can be trained with relatively small data sets, by integrating them into classic receiver processing via model-based deep learning techniques [7], [8], see, e.g., [21]- [25]. These hybrid model-based/data-driven receivers build on existing communication models and incorporate deep learning into their design.…”