Xinzhuo Jiang

Xinzhuo Jiang

Auto-generated profile, based on publicly available information

Xinzhuo Jiang

School of Electronics and Information Engineering, Harbin Institute of Technology, Harbin, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
27 January 2025 in IEEE Transactions on Cognitive Communications and Networking
Verified
Journal Article
Closed access
XLXiaodong Li
Xiaodong Li
YGYulong Gao
Yulong Gao
Yansha DengYansha Deng
Yansha Deng
XJXinzhuo Jiang
Xinzhuo Jiang
Federated learning (FL) enables edge nodes to collaboratively train a global model under the coordination of a server without sharing local private data. However, data heterogeneity across nodes leads to serious performance degradation or even non-convergence of the learned model. To tackle this challenge, most existing methods typically involve either local model regularization or global model adjustments. Nevertheless, these methods primarily perform model aggregation based on the dataset size proportion, while the exploration of the impact of non-independent and identically distributed (non-IID) data on aggregation weights remains insufficient. To this end, we theoretically derive an analytical expression for the aggregation weights by minimizing the convergence upper bound of standard FL on non-IID data across nodes, achieving a tighter bound and superior convergence performance. Accordingly, we propose an adaptive aggregation weight strategy, called FedAAW. It can be easily incorporated into other FL methods to improve their convergence performance with negligible additional communication overhead. Extensive experiments on four common datasets show that FedAAW can effectively mitigate the performance degradation caused by data heterogeneity in various cases. Simply applying FedAAW to other methods can significantly improve their performance, achieving a maximum improvement of 37.32% in test accuracy and outperforming other state-of-the-art aggregation weight strategies.

4 June 2024 in IEEE Transactions on Vehicular Technology
Verified
Journal Article
Closed access
XJXinzhuo Jiang
Xinzhuo Jiang
YGYulong Gao
Yulong Gao
WZWei Zhang
Wei Zhang
Yansha DengYansha Deng
Yansha Deng
An effective precoding method stands as an imperative task, for the dynamic fluctuations inherent in fading channels that have a huge impact on communication systems. This paper introduces a novel transformation method for constructing a domain based on rotation matrices, termed Iterative Rotation Orthogonal Transformation (IROT). Our method centers on investigating the key factor in determining symbol decision accuracy, which is the distribution of noise energy superimposed on each symbol within a symbol block. The optimal distribution is influenced not only by channel fading but also by the average signal-to-noise ratio (SNR). By employing linear equalization, we quantitatively evaluate the impact of the domain transformation method on noise energy distribution. When the SNR surpasses a specific threshold, the symbol error rate (SER) reaches its minimum through the average allocation of noise energy. To achieve controllable adjustments of the noise energy, IROT constructs a unitary matrix by a double-layer iteration. Then we calculate the SNR threshold and analyze the complexity and convergence of IROT. Furthermore, we proposed the inverse balanced IROT to improve the SER performance at low SNR. The simulation results show the superior SER performance of IROT compared to other domain transformation and pre-equalization methods.