Study Of Attack,Detection And Defense For Stealthiness-Enhanced Hardware Trojan In Network-on-Chip

Information security is very important to systems like data centers,server,etc.In such systems,many-core systems are the building blocks.Network-on-chip(No C)connects each processor and storage unit in the many-core system.No C is vulnerable to hardware Trojan(HTs).HTs are inserted into No C by the malicious producers or designers to disrupt the normal function of No C or leak users' sensitive information,which pose a huge threat to many core chip.We need to explore how to ensure the security of No C from two perspectives: attacks and defense.In terms of attacks,it is necessary to study hardware Trojan with high steahthiness and strong aggressiveness.In terms of defense,it is necessary to design the effective detection and defense mechanism to protect many-core systems from HTs with high steahthiness and strong aggressiveness.In terms of attacks,the paper proposes an stealthiness-enhanced hardware Trojan,to launch sinkhole/blackhole Do S attack in No C.In blackhole attack,the HTs tamper with the packets in No C,causing the packets are dropped or forward to the malicious node.In sinkhole attack,the HTs tamper with the number of free buffers in the malicious node to solicits packets from neighbor nodes.A blackhole attack can follow a sinkhole attack.The attack result in the loss of packet or leak of sensitive information.For further improving the attack effects,we modeling the attack effects and their influence factors,and then an optimization problem is formulated to optimize the attack effects.Experimental results show the proposed HT cause a serious attack effect in No C and the attack effects increase more than 30% after optimization.In terms of defense,the paper proposes the adaptive defense method against the stealthiness-enhanced hardware Trojan by detecting and defending it.The detection scheme of adaptive defense method generates the ECC-based detection packet to detect the tampering attack in No C,and then calculate the probability of the malicious node by Bayesian equation.A bypass routing algorithm is used to defend the HTs.For further improving the defense effects and balancing the performance and overhead of the defense method,we modeling the defense effects,system performance and their influence factors,and then an optimization problem is formulated to optimize the defense effects and system performance.Experimental results show the defense method can detect all the HT quickly,which reduces the attack effect by 92% with defense.The defense effect increases by more than 56% after optimization.