Convolutional neural network-based signal demodulation method for NOMA-PON

Abstract

Non-orthogonal multiple access (NOMA) is a promising scheme for flexible passive optical networks (PONs), which provides high throughput and overall improved system performance. NOMA with the successive interference cancellation (SIC)-based receiver, which is used to detect the multiplexed signal in a sequential fashion, requires perfect channel state information and suffers from the error propagation problem. In this paper, we propose a convolutional neural network (CNN) based signal demodulation method for NOMA-PON, which performs channel estimation and signal detection in a joint manner. The CNN is first trained offline using the captured data for a given received optical power and then used to recover the data stream directly in the online mode. We show by experimental demonstration that, the proposed CNN-based receiver (Rx) outperforms the conventional SIC-based Rx and is more robust to the nonlinear distortion. We show that for the CNN-based system with 20 km optical fiber, the required received optical power levels at a bit error rate (BER) of 1×10 −3 are lower by 4, 3 and 2.5 dB for power allocation ratios of 0.16, 0.25, 0.36, respectively compared with SIC-based system. In addition, the BER performance of CNN deteriorates considerably less with non-linear distortion compared with SIC.

Investigation of the fifth generation non-orthogonal multiple access technique for defense applications using deep learning

“…For DL, many applications and resources are available. Considering their reliability and usability, matrix laboratory (MATLAB) 2020 41–51 and PYTHON version 3.9 42 are used in our simulations. TensorFlow with graphics processing unit (GPU) acceleration was also used to apply the suggested DL algorithm as a popular open-source DL platform from Google.…”

“…The first UE is allocated to most of the power (85%) and the rest power is allocated to the second UE (15%). The sigmoid function is used at the output layer and the rectified linear unit (ReLU) 40–50 is used at the hidden layers. The total number of training samples is 50,000 and, for increasing the convergence rate, a smaller mini-batch symbol set is considered in the form of n 2 . In the MIMO-NOMA-DL system the input complex modulated data is fed as input to the DNN in the form of a column vector, and eight input layers are considered in a time slot for a 4 × 4 MIMO system.…”

Downlink Video Communications for Enhanced Hybrid NOMA using Deep Learning Model

“…In an uplink, multi-carrier NOMA scenario with UEs requiring varying data rates and latencies, Bansbach, E.M., [23] studied the effect of Buffer state information (BSI) on the performance of a centralized scheduler. The research suggests a new scheduler based on actorcritic reinforcement learning with BSI to deal with the vast combinatorial space of distributing UEs to the resources.…”

Tunable plasmon-induced transparency in graphene-based plasmonic waveguide for terahertz band-stop filters

“…Given that single-band tunable filters cannot largely satisfy the requirements of multiband communications, the research on tunable filters with multiband or even switchable number of frequency bands is in great demand. The tunability can be realized by using phase-change material vanadium dioxide (VO 2 ), liquid crystals, and other materials affected by temperature, electromagnetic field, as well as mechanical regulation combined with microelectromechanical systems [30][31][32][33][34]. In addition, graphene has acted as a promising material for THztunable devices due to its high electron mobility, metallic characteristic, and unique doping capacity [35].…”