Detecting Attacks For Relay Systems With Unreliable Channel State Information

With the development of relay technology,the coverage and the reliability of the system has been increased.The combination of cooperative communi-cation and relay technology has promoted the communication system to high spectral efficiency,high throughput and high robustness.In view of the radio characteristics of wireless communications,security has become a major bottle-neck restricting its development.In relay systems,we consider the case where a potentially malicious relay node may corrupt the network by forwarding gar-bled information.In this situation,misleading feedback results from the fact that reliable CSI is not available at the source and/or the destination.The main contents of this dissertation are as folllows:To begin with,the dissertation propose an attack-detection technique with unreliable CSI for the relay system with direct link.This tenique extracts statis-tic desicion metric from physical layer observations.we employ mathmatical expression to model the relay behaviors.distribution function is chosen to char-acterize the relay forwarding behavior.The feedback behavior is characterize by comparing reliable and unreliable channel information.For the source node,sending information is a priori known;for the destination node,direct link is reliable,and these side information can be used as a valid reference.Judging the distribution computed according to unreliabel channel information with the distribution extracted from physical layer observation information,we decom-pose the relay behavior into a detectable and undetectable class.The physical layer observations can be converted to empirical distribution by the law of large numbers and it will be used to construct statistic decsion metric.By classify-ing the action of the relay as detectable or undetectable,we propose a novel approach which can deal with any detectable malicious attacks and which con-tinues to work effectively in the presence of unreliable CSI.The approach is executed simultaneously at both source node and destination node.For the de-tectable class,it is shown that there is a high probability that malicious attacks can be successfully detected.For the undetectable class,it is demonstrated that,even though an attack may fool the proposed detection approach,these unde-tectable attacks do not affect the performance improvements that are achievable by cooperative diversity.Second,the dissertation extended the attack-detection technique to the relay system without directlink.Lacking of direct link,the destination node hardly acquires side information.When the source distribution is known,it can be converted into existing detecting scheme.Otherwise,all the received infor-mation is not trustworthy.The destination node can access side information by support artificial noise.We employ the source node and the destination node to emulate multi-antenna environment.Artificial noise can be changed with time by changing the coeffcient matrix with time to defend i.i.d.relay attack.In addition,when the unreliable channel information is linearly correlation with reliable channel information,artifical noise failures.We modify it to a none-zero value to against this situation.