Multi-fusion Based Distributed Spectrum Sensing against Data Falsification Attacks and Byzantine Failures in CR-MANET

Abstract

In mobile ad-hoc cognitive radio networks (CRMANET), accurately identifying primary user spectrum occupancy is an important requirement in successful utilization of the spectrum by the secondary user. This is made difficult by malicious mobile secondary users launching spectrum sensing data falsification (SSDF) attacks, byzantine failures of devices, primary user signal fading, and hidden terminal problems. Another ability of these malicious users is to use their mobility to hide under the changing neighborhood. Existing state of the art techniques either consider a centralized approach in decision making or when considering an ad-hoc network, do not take into account the mobility of the devices which adds unique limitations. We present a light weight multi-fusion based distributed spectrum sensing scheme (MFDSS) in a mobile ad-hoc secondary user network to overcome the aforementioned problems. MFDSS allows each secondary user to collect semi-global information to make decisions in a distributed manner. It uses outlier detection and data fusion to remove incorrect sensing data generated by byzantine failures. Reputation information of the devices is used to suppress a SSDF attack. MFDSS incorporates a reputation propagation and fusion scheme to prevent malicious devices from hiding behind changing topology and to maintain the freshness of the reputation information. Additionally, MFDSS includes an incubation period to discourage devices from changing identity to perform a Sybil attack or to mask a bad reputation. Detailed analysis is presented and the results show significant improvement in correct primary user spectrum occupancy identification. We also show that without the reputation propagation of MFDSS, a malicious device is able to move its position and hide its malicious intentions in the new neighborhood.