- 13 June 2018Time: - 12:00pm
Venue: Room: Eng 3.24, Mile End Campus, QMUL Address: 10, Godward Square, Queen Mary University of London, Mile End Rd, London E1 4FZ
Speaker: Yun Wen from Fujitsu Laboratories, Japan.
Title : Machine Learning based Attack Method against Artificial Noise-aided Secure Communication
Physical layer security (PLS) technologies have attracted much attention in recent years for its potential of providing information-theoretical secure communication. Artificial Noise-aided (AN) transmission is considered as one of the most practicable PLS technologies, as its secure transmission doesn’t involve the channel information of the eavesdropper. However, the performance of AN transmission under the condition of that the eavesdropper has more antenna than the transmitter, hasn’t been investigated sufficiently and may be one of the serious obstacles of AN’s realization. In this talk, an effective attack method based on machine learning approach, to attack AN transmission and evaluate its performance will be presented. In our method, we assume that the received signals in the eavesdropper are distributed in different hyper-planes corresponding to the constellation points of modulation, therefore adopt a clustering algorithm to estimate the channel between transmitter and receiver to decode the AN-aided packet. Numerous simulation results show that our algorithm performs much better than the conventional method and can decode AN-aided transmission with varies modulation method. We also evaluated in which condition can AN transmission achieve secure communication, even when eavesdropper equipped with our effective attack method.
Yun Wen received the B.E. degree in electrical and electronic engineering from ZheJiang University, HangZhou, China, and M.S. degrees in informatics from Kyoto University, Kyoto, Japan, in 2004 and 2009 respectively. He joined Fujitsu Laboratories Ltd., Kawasaki, Japan in 2009. He has research experiences on the area of wireless sensor networks, routing protocol in multi-hop networks and MAC protocol design for WLAN networks. His current research interests include ultra-reliable wireless communication and physical layer security.