Home > Research > Publications & Outputs > DRET

Electronic data

  • 8_11_DRET_1_

    Accepted author manuscript, 2.17 MB, PDF document

View graph of relations

DRET: a system for detecting evil-twin attacks in smart homes

Research output: Contribution to conference - Without ISBN/ISSN Conference paper

Published
  • Zhanyong Tang
  • Yujie Zhao
  • Lei Yang
  • Shengde Qi
  • Dingyi Fang
  • Xiaojiang Chen
  • Xiaoqing Gong
  • Zheng Wang
Close
Publication date29/07/2016
<mark>Original language</mark>English
EventSmart X 2016 : The 2016 International Conference on Smart X - Dalian, China
Duration: 29/07/201631/07/2016

Conference

ConferenceSmart X 2016
Country/TerritoryChina
CityDalian
Period29/07/1631/07/16

Abstract

Evil-twin is one of most commonly attacks in the WIFI environments, with which an attacker can steal sensitive information by cloning a fake AP in Smart Homes. The current approaches of detecting Evil-twin AP uses some identities/fingerprints of legitimated APs to identify rouge APs. Prior work in the area uses information like SSIDs, MAC addresses, and network traffics to detect bogus APs. However, such information can be easily intimated by the attacker, leading to low detection rates. This paper introduces a novel Evil-Twin AP detection method based on received signal strength indicator (RSSI). Our approach exploits the fact that the AP location is relatively stable in Smart Homes, which is to great extent to meet the requirement of the detection factor not easy to imitate as previous refer. We employ two detection strategies: a single position detection and a multi-positioned detection methods. Our approach exploits the multipath effect of WIFI signals to translate the problem of attack detection into AP positioning detection. Compared to classical detection methods, our approach can perform detection without relying on professional devices. Experimental results show that the single position detection approach achieves 20 seconds’ reduction of delay time with an accuracy of 98%, whereas the multi-positioned detection approach achieves 90% correct.