Greetings! My name is Bo Yang, a CS PhD student at Delft University of Technology. My research focuses on Embedded AI for 6G Networks, optimizing AI algorithms to adapt to diverse hardware and computing capabilities across advanced communication infrastructures. With an emphasis on designing hardware-aware, adaptable AI models, I aim to enhance performance and scalability within 6G base stations that manage densely packed frequency bands. My work supports seamless connectivity across space, air, and ground base stations, driving innovation in the co-design of AI and hardware for efficient, next-generation network solutions.

🔥 News

• 2024.11:  🎉🎉 I started my PhD studies in Computer Science at TU Delft. Supervised by Prof. Qing Wang(TuDelft) and Prof. Fernando A. Kuipers(TuDelft)
• 2024.09:  🎉🎉 I graduate from TU/e with Cum Laude, and finish my internship at NXP system group.
• 2023.12:  🎉🎉 I will be an intern at NXP for 9 months, supervised by Prof. Ronald Aarts(TU/e) and Prof. Frans Widdershoven(Tud)
• 2023.07:  🎉🎉 I will be an intern at IMEC for 5 months, supervised by Dr. Alireza Sheikh(IMEC) and Prof. Hamdi Joudeh(TU/e)
• 2022.09:  🎉🎉 I started my master’s study at TU/e in Electrical Engineering (Signal Processing System).
• 2022.07:  🎉🎉 I am graduated from Shandong University with a bachelor’s degree, supervised by Jifang Tao.

📝 Publications

OJSP 2025

Hybrid Real- and Complex-valued Neural Network Architecture (under submission)

Alex Young(Corresponding), Luan Vinıcius Fiorio(Corresponding), Bo Yang, Boris Karanov, Wim van Houtum, Ronald Aarts

JC&S 2024

Successive Threshold-Based Multipath Mitigation Aided by Neural Network for UWB Ranging

Alireza Sheikh, Bo Yang, Mohieddine El Soussi, Amirashkan Farsaei, and Peng Zhang

• NN-aid-High-accuracy-Range-Estimation, JC&S 2024

• 2021, A wireless intelligent sensor and its application. patent number: CN202110094598.5

📖 Educations

• 2022.09 - 2024.08, Master in Eletrical Engneering(track:Signal Processing System),(CUM LAUDE), Eindhoven University of Technology, Netherlands

  GPA: 8.5/10, THESIS: 9/10

  Relevant Course: Statistical signal processing, Non-linear optimization, Bayesian machine learning and information processing , Adaptive array signal processing, Techniques for video compression & analysis, Convolutional neural networks for computer vision, Advanced sensing using deep learning.

• 2018.09 - 2022.07, Bachelor in Electronic Engineering, Shandong University, Qingdao, China

  GPA: 84.76/100

💬 Invited as TPC reviewer

• 2024.03, Invited as TPC reviewer for IEEE JC&S 2024.

💻 Internships

• 2023.12 - 2024.09, Research Intern, NXP, Eindhoven NL

  • I am a part of the ML&AI department at NXP, where my role centers around the optimization of the innovative Complex-Real Value Neural Network (CRVNN) developed by NXP.
  • My primary focus lies within the realm of autoML, particularly in areas such as Hyperparameter Optimization (HPO) and Neural Architecture Search (NAS).
  • I leverage autoML frameworks like Optuna and NNI to enhance both the architecture and other hyperparameters of the CRVNN. Additionally, I am actively involved in crafting automated procedures for the design of this novel network.
  • My responsibilities extend to the development of an architecture optimization framework, followed by efforts to enhance the interpretability and explainability of the network. Once the CRVNN architecture is refined, I apply it to specific tasks, such as optimizing its performance for applications like hearing aids.

• 2023.07 - 2023.12, Research Intern, IMEC, Eindhoven NL

  • In my role as a research intern with the UWB4z group at IMEC, my primary responsibility is to enhance the accuracy of range estimation and localization for UWB devices in diverse environments.
  • I delve into the development of innovative algorithms and construct efficient neural networks to elevate the precision of UWB devices. Through a comprehensive analysis of UWB signal characteristics, I introduced a cutting-edge neural network architecture named ‘STMnet’.
  • This architecture, when integrated with IMEC’s internal algorithm, resulted in a significant enhancement in range estimation performance compared to established methods. The successful outcomes of this work have been documented and published in the International Symposium on Joint Communications & Sensing 2024.

• 2021.09 - 2022.01, Research Intern, DiDi Global, China

  Assisting with the development and execution of test plans and test cases for software applications or websites. Participating in the testing of software releases to identify bugs, defects, and usability issues. Documenting and tracking defects and issues found during testing, and working with developers to resolve them. Conducting regression testing to ensure that previously identified defects have been fixed and do not reoccur.

🔥 Project

• 2023.07 - 2023.12, UWB signal Range Estimation by STMnet

  • Use a successive threshold-based multipath mitigation algorithm (STM) that improves the ranging performance in multipath conditions.
  • Then further improve the ranging performance of STM in multipath conditions by aiding the STM with STMnet as a Neural network that estimates the ranging error of the STM.

• 2023.03 - 2023.06, DOA Estimation By Deep Learning

  • Propose a solution for the estimation of number of sources together with their corresponding DOAs.
  • Combine neural network with maximum likelihood estimation (MLE) to obtain a good results and save computation resources. Test the solution by different scenarios and the performance is great for signals with low SNR.

• 2023.01 - 2023.03, Anomaly Detection with Autoencoder

  • Implement, run and analyze an autoencoder architecture from scratch for anomaly detection purpose on MNIST dataset.

• 2022.11 - 2023.01, Cityscapes Pixel-Level Segmentation Benchmark and Robust Segmentation

  • create a network that performs semantic segmentation on the Cityscapes dataset. - implementation of a trainable network which successfully evaluates on the cityscapes dataset.
  • Neural networks degrade in performance when unexpected things occur. We Finetune the network to become a robust solution for various conditions, to overvcome the degrade in performance when unexpected things occur.
  • Use two robustness benchmarks(Degrading of image quality and Generalization) to evaluate the NN.