Department of Electronic & Computer Engineering - ECE Future Leaders PG Seminar Series
Session 1: Semi-Decentralized Federated Edge Learning with Data and Device Heterogeneity
Speaker: Miss. Yuchang SUN, PhD (ECE), HKUST
Federated edge learning (FEEL) has attracted much attention as a privacy-preserving paradigm for training deep learning (DL) models. Nevertheless, the limited coverage of a single edge server results in an insufficient number of participated client nodes, which may impair the learning performance. We investigate a novel framework of FEEL, namely semi-decentralized federated edge learning (SD-FEEL), where multiple edge servers are employed to collectively coordinate a large number of client nodes. By exploiting the low-latency communication among edge servers for efficient model sharing, SD-FEEL can incorporate more training data, while enjoying much lower latency compared with conventional federated learning.
Session 2: Back Telemetry of 3-Coil WPT and 2-Coil WPT for Biomedical Implants
Speaker: Mr. Sayan SARKAR, PhD (ECE), HKUST
The wireless power transfer (WPT) system is widely adopted as a battery-less solution for biomedical implants. The system consists of a transmitter (TX) primary coil and a receiver (RX) secondary coil. During wireless power transfer, both the loading current of the receiver and the coupling coefficient (k) between the primary and secondary coils may change, which will affect the output voltage of the receiver (Vo). A communication link from the RX side to the TX side, known as back telemetry (BT), could be used to report the RX side status to the TX side, and to maintain a regulated and steady Vo. One popular backscattering scheme is load shift keying (LSK) due to its simple circuit architecture, low power consumption, and small area requirements. In this research, we propose two LSK detection schemes.
In the first scheme, a tertiary detection coil that is co-planar with the primary transmitter coil is used to receive and decode asynchronized BT data from the receiver. This is a 3-coil WPT system. Variation of the tertiary coil along with the effect of tertiary coil parameters on BT, such as the number of turns, compensation capacitance and quality factor, are discussed in detail in maximizing the voltage gain and the end-to-end efficiency through analytical modelling, SPICE simulations and measurement results.
In the second scheme, a 2-coil WPT system is adopted by removing the tertiary coil. A synchronized LSK signal is generated by a linear current sink voltage regulator, and the efficiency is improved by specially designed digital blocks. On the primary side, instead of a large detection coil, an integrated current sensor is designed to sense the current through the power NMOS of the Class-D driver. Compared to previous works, this design allows a longer operating distance, a faster load-transient response, and a higher level of integration with a much lower system and circuitry design complexity.
Yuchang SUN (student member, IEEE) received the B.Eng. degree in electronic and information engineering from Beijing Institute of Technology (BIT) in 2020. She is currently pursuing a Ph.D. degree supervised by Prof. Jun ZHANG in the Department of Electronic and Computer Engineering at the Hong Kong University of Science and Technology (HKUST). Her research interests include federated learning and distributed optimization.
Sayan Sarkar received his undergraduate degree in electrical engineering from the National Institute of Technology, Durgapur, India, in 2016. During his undergraduate studies, he received the ‘General Electric Edison Challenge Award’ and ‘Air India Rank and Bolt’ awards for outstanding academic performance. He is currently working as a PhD candidate in the Department of Electronic and Computer Engineering at HKUST. His research interests include wireless power transfer technology, analog and mixed-signal IC design, biomedical instrumentation and biosignal processing. He published more than 30 peer-reviewed articles in reputed conferences and journals. After his undergraduate education, he worked briefly in Coal India as an assistant manager and later co-founded “WECARE MEDSERVICE LLP”. Wecare worked in the non-critical and maternal healthcare space to improve primary healthcare infrastructure in India and emerged as the winner in top business-model competitions in India.