Aggregate Preamble Sequence Detection with Deep Learning


Speaker:    Prof. Vincent Wong

Time:        11:00-12:00, May 20

Location:    SIST 1C-201

Host:         Prof. Yong Zhou


Massive Internet of Things (mIoT) is a major use case of the fifth generation (5G) wireless systems. To enable mIoT, it is required to support a large number of simultaneous connection requests from the IoT devices. The conventional Long Term Evolution (LTE) random access procedure hinders supporting mIoT due to the limited number of available random access preambles. In this talk, we propose to aggregate two Zadoff-Chu preamble sequences to obtain a larger set of random access preambles by considering all the combinations of pairing two different sequences. Decoding the aggregate preambles is challenging because we need to decode two Zadoff-Chu preamble sequences that are allocated half of the transmit power each. Hence, we present a preamble decoding receiver by training a deep neural network (DNN) to decode the aggregate preambles successfully. The proposed design outperforms other collision avoidance techniques such as access class barring (ACB) in terms of a lower average total service time. Furthermore, the proposed receiver design decodes the aggregate preambles with lower probability of misdetection and false alarms compared to decoding single Zadoff-Chu preambles with a conventional LTE preamble decoding receiver.


Vincent Wong is a Professor in the Department of Electrical and Computer Engineering at the University of British Columbia, Vancouver, Canada. His research areas include protocol design, optimization, and resource management of communication networks, with applications to the Internet, wireless networks, smart grid, fog computing, and Internet of Things. Dr. Wong is an executive editorial committee member of the IEEE Transactions on Wireless Communications, an Area Editor of the IEEE Transactions on Communications, and an Associate Editor of the IEEE Transactions on Mobile Computing. Dr. Wong is a Fellow of the IEEE.

SIST-Seminar 18155