We investigate energy-efficient federated learning (FL) in computation and communication resource-constrained edge intelligence networks using model compression. An edge device selection strategy is designed to select appropriate edge devices for participating in FL at the beginning of each training iteration. An optimization problem is then formulated to jointly optimize the pruning ratio, CPU frequency, uplink power, and bandwidth allocation for the selected edge devices. Due to the non-convexity of the optimization problem, it is decomposed into three subproblems, and closed-form solutions or efficient algorithms are developed for each subproblem. Based on these solutions, an alternating optimization algorithm is constructed to solve the original problem. Simulation results show that the proposed scheme outperforms baseline schemes in improving the energy efficiency of FL.

Efficient Federated Learning in Resource-Constrained Edge Intelligence Networks using Model Compression

This article presents the design and implementation of a digital pre-distortion (DPD) scheme based on machine learning (ML) algorithm, which has been envisioned for the sixth generation of mobile communications (6G) analog radio over fiber (A-RoF) system. The DPD scheme employs an augmented real-valued time delay neural network (ARVTDNN) to compensate for the memory and memory-less non-linear effects introduced by the A-RoF communication chain. The ARVTDNN receives the base-band samples of the transmit waveform and stores them in a time-delay line (TDL), which means that the information necessary to compensate for time-correlated effects is available for the ML algorithm. Therefore, the ARVTDNN can compensate for both memory and memory-less effects, including the chromatic dispersion introduced by the optical fiber. The novel DPD can bring a significant positive impact in enhanced remote areas (eRAC) applications since the proposed DPD allows A-RoF to be used to connect the base-band unit with a low-cost radio frequency (RF) radio head installed in remote areas. In this case, the severe non-linearities introduced by the Mach-Zehnder optical modulator (MZM) and power amplifier (PA) and the time dispersive distortions introduced by the fiber can be mitigated, improving the quality of the signal. Furthermore, the novel ML-based linearization scheme can be flexibly tailored to cover variations in the operating scenario by adjusting its hyperparameters and training data. The root mean square error vector magnitude (EVMRMS), normalized mean square error (NMSE) and adjacent channel leakage ratio (ACLR) metrics have been used to evaluate the performance of the proposed DPD. Numerical results demonstrate promising linearization performance, reducing the signal in- and out-band distortions when the fronthaul link is extended to cover remote and rural areas.

Novel Machine Learning Linearization Scheme for 6G A-RoF Systems

In this article, a novel block-oriented recurrent neural network (RNN) model is proposed for behavioral modeling and digital predistortion (DPD) of radio frequency (RF) power amplifiers (PAs). This article provides an insightful discussion on the importance of input-end parallel finite impulse response (FIR) filters for performance enhancement and finds, for the first time, the unique linearization correction effect of each FIR filter in input-end parallel FIR filters at different frequencies, which is also the reason why block-oriented time-delay NN (BOTDNN) outperforms vector decomposition-based time-delay NN (VDTDNN) in terms of linearization performance. In order to retain the interaction information between nonlinearity and memory effects, the proposed model preserves the feedback path in feedback PA behavioral model. With a view to addressing the challenge of parameter extraction caused by the feedback structure, this article first demonstrates the potential relationship between RNN cells and feedback structures. Subsequently, considering the trade-off between complexity and performance, the Just Another NETwork (JANET) cell is chosen to construct the feedback structure to form the proposed block-oriented JANET (BO-JANET) model. The BO-JANET model is validated using two PAs with the center frequencies of 2.4 and 3.55 GHz, respectively. Experimental results demonstrate that the proposed model achieves further linearization performance improvements compared with other advanced models. 

Block-Oriented Recurrent Neural Network for Digital Predistortion of RF Power Amplifiers

In this paper, an efficient neural-network-based adaptive DPD design which performs well under power varying conditions is presented. The DPD design is derived on the basis of the envelop time-delay neural network (ETDNN). The redefined ETDNN-DPD requires the part of parameter updates, which enables to adapt it to the rapid change of power amplifier (PA) distortion. Additionally, the redefined ETDNN-DPD also maintains the stability of the compensation performances under varying power condition while its structure is pruned by the structured pruning. Furthermore, to verify its practical use, we also propose the weight scaling technique, which reduces multiplications of the redefined ETDNN-DPD, and applied it to the implementation of the redefined ETDNN-DPD on FPGA. Compared FPGA-implemented ETDNN-DPD with FPGA-implemented conventional memory polynomial DPD, we verified that our proposed DPD achieved 3.2 dB better error vector magnitude (EVM) while lower hardware resource utilization at the fixed power level. Moreover, our proposed DPD kept better performance under the varying power condition only by the partial update of its parameters than memory polynomial DPD. 

A Sparse Neural Network-Based Power Adaptive DPD Design and Its Hardware Implementation

Digital predistortion (DPD) using neural network(NN) has attracted attention as a promising technique for compensating complex nonlinear distortions caused by wideband radio frequency power amplifiers. However, NN-DPD is difficult to implement in hardware due to the large number of multiplications. In this paper, we propose an NN-DPD using DeepShift that replaces neural network multiplications with bitwise shift and sign flipping during both training and inference. First, when DeepShift was applied, we showed that applying residual learning to DPD can reduce performance degradation. Next, we examined pre-trained models using floating-point baseline models and scratch-trained models and found that the pre-trained models perform better in exchange for requiring pretraining. The scratch model, on the other hand, has the advantage of improving the training computational complexity. In an actual experiment of passing a signal with DPD applied through a power amplifier at 2.14 GHz, the error vector magnitude was slightly degraded from 1.31 % trained by the floating-point model to 1.57 % with the pre-trained model and 1.90 % with the scratch model. 

Digital Predistortion of RF Power Amplifiers using DeepShift

The paper presents several novel residual recurrent neural network (RNN) models for digital predistortion of radio-frequency power amplifiers. Then, unstructured and structured neural network (NN) pruning algorithms are proposed to reduce the per-sample computational complexity and the required memory, without compromising their modeling performance. The mathematical equations of the complexity and the memory are derived to analyze the hardware cost and design low cost models. An experimental platform is developed to measure the modeling performance in terms of adjacent channel leakage ratio (ACLR), normalized mean square error, error vector magnitude (EVM) and bit error rate. The platform uses an envelope tracking PA excited by 80 MHz IEEE 802.11ac signals. Compared with the state-of-the-art RNN model in Sun2019, the proposed models reduce the per-sample complexity and required memory by $94-99$%, they also improve the ACLR and EVM by at least 3 dB and 1 dB at 0.45 dB and 3.15 dB gain compression, respectively. Furthermore, an observation path with time alignment is used to remove the transmit latency in the state-of-the-art approach. At 3.15 dB gain compression, the proposed models outperform the state-of-the-art feedforward and convolutional NN models by at least 1.3 dB in EVM, outperform the general memory polynomial model by at least 3.3 dB in EVM. The above evidences indicate that the proposed RNN models can be implemented with relatively low hardware cost and achieve the best EVM in high power efficiency scenario.

Residual RNN Models With Pruning for Digital Predistortion of RF Power Amplifiers

pdf/exploiting-non-uniform-quantization-for-enhanced-ilc-in-du531kernc59.pdf

Exploiting Non-Uniform Quantization for Enhanced Ilc in Wideband Digital Pre-Distortion

The objective of this work is to address one of the recent ITU AI/machine learning challenges, specifically on the RF power amplifier (PA) behavioral modelling. This paper presents a novel model namely real-valued time-delay temporal convolutional network (RVTDTCN) to learn the long-term memory effect from PA in 5G base-station transmitters. The proposed model employs dilated convolutions to incorporate information from various time stamps and capture intricate temporal dependencies. This is achieved without the requirement of additional coefficients. In our experiment, the ITU ML5G-PS-007 datasets collected from commercial 5G base-stations are employed for both training and testing purposes. The performance of the learned PA behavior is evaluated using both normalized mean square error (NMSE) and adjacent channel error power ratio (ACEPR). The results show that the RVTDTCN model achieves competitive performance with the state-of-the-art feedforward neural network (FNN) models but reducing the computational complexity by 60%. Moreover, the proposed model outperforms the state-of-the-art convolutional neural network (CNN) model by 2-3dB in NMSE and ACEPR, with a similar complexity.

Learning the Long-term Memory Effect of Power Amplifiers Using Temporal Convolutional Network

Handle everyday research tasks with reliable, citation-backed results

Your personal Research Agent to handle research tasks with citation-backed results

Popular Tasks used by Researchers

How can I help with your research?

Meet SciSpace

Get more enhanced response by uploading the PDFs you want me to reference.

No relevant PDFs in your library

SciSpace is the AI research assistant for academics. Run systematic literature reviews on 280M+ papers, and write papers with cited sources. Trusted by 1M+ students, PhDs & researchers.

SciSpace | AI for Research

Analyze PDFs

Code & Manuscripts

Funding & Grants

Literature & Patents

Medical & Clinical Data

Systematic Review

Visualize & Present

Web & Data

Build a Google Scholar-like website for your research.

Build a website

Create charts and images for your research

Create a Chart

Write a paper for submission to a journal

Draft a manuscript

Patent Search

Design eye-catching scientific posters in minutes.

Scientific Poster Generation

Systematic Literature Review

One task is running at the moment. Your messages will be shown right after.

Drag and drop or click here to browse

Loved by <highlight>1 million+</highlight> researchers

Extract a list of specific topics and their sources from unstructured text

Topics

Compare and analyze relevant papers that matches with your search

Papers

Get insights from PDFs and bookmarked papers from your library

My library

Recent searches

Try searching for:

Catch AI-generated content in scholarly and non-scholarly content

{ai} Detector

Ai Writer

Get PDF Summaries, highlighted text explanations 

Chat with PDF

Effortlessly create in-text citations and bibliographies in APA and 2,500 other formats

Citation generator

Get explanations, summaries, and answers on academic papers

Ease up your research workflow with {scispace}'s cohort of exciting AI tools

Elevate your academic writing skills and convey your ideas the way you want

Paraphraser

Explore our range of reading and writing tools

Your file is being prepared and should be ready in a few minutes. If it's a large file, it might take a bit longer. You can close this window, and we'll email you the file when it's done.

You have reached a maximum limit of <strong>{limit}</strong> columns in the table. Remove at least <strong>1</strong> column to add or create another one.

Fei Tong

Author Tools

Chat about Author

Papers

Residual RNN Models With Pruning for Digital Predistortion of RF Power Amplifiers

Exploiting Non-Uniform Quantization for Enhanced Ilc in Wideband Digital Pre-Distortion

Learning the Long-term Memory Effect of Power Amplifiers Using Temporal Convolutional Network