On Thermal Covert Channel Attacks And Its Countermeasure In Many-core Systems

The security of on-chip many-core systems is the cornerstone of modern information systems security.Among various security threats of on-chip many-core systems,thermal covert channel(TCC)attack uses the thermal correlation among the processor cores of on-chip many-core systems to transmit data with heat as its medium,which can be exploited by hackers to cause serious information leakage of the on-chip many-core systems.With high stealthiness and low bit error rate(BER),TCC attack poses a severe threat to the information security of on-chip many-core systems.To evaluate TCC attack and its countermeasures,a large number of simulation or real machine experiments are needed,which is time-consuming and costly.In view of the above challenge,this paper provide an analytical model to determine the key TCC performance metrics,which can quickly evaluate the TCC attack and its countermeasures.The proposed model shows that the noise of TCC attack has high amplitude at low frequency and moderate amplitude at middle frequency.This paper uses the above characteristic to propose a stealthy TCC attack with reduced signal amplitude,which cannot be detected by the existing countermeasures.To enhance the robustness of TCC countermeasures,this paper proposes a countermeasure for the proposed stealthy TCC,which can effectively block the TCC’s data transmission.With the challenge that evaluating TCC attack and its countermeasures is time-consuming and costly,this paper provides an analytical model of TCC attack.The proposed model can quickly determine the key TCC performance metrics,in terms of BER,signal to noise ratio(SNR),and channel capacity,without going through lengthy computer simulation and/or physical experiments that are normally needed in current TCC performance studies.In addition,this paper uses the proposed model to analyze TCC countermeasures based on jamming noise and Dynamic Voltage Frequency Scaling(DVFS),a feature that can be explored to design/optimize the countermeasures against the TCC attacks.Experiments show that the TCC performance predicted by the proposed theoretical model is found in a good agreement with that obtained from computer simulations,with an average error lower than 7%.The above model shows that the noise of TCC attack has a moderate amplitude at medium and low frequencies,a feature that can be used to enhance the stealthiness of TCC attacks.This paper proposes a stealthy TCC attack based on direct sequence spread spectrum(DSSS).With the thermal data modulated by a pseudorandom bit sequence,the signal strength of the stealthy TCC drops to such a low level that the signal is nearly indistinguishable from the noise.Existing TCC countermeasures cannot detect this stealthy TCC attack.To combat this stealthy TCC,this paper proposes a countermeasure that lets the received signal pass through a differential filter where irrelevant frequency components occupied mainly by the noise gets eliminated and extract the key information;the filtered signal is next compared against a threshold to recover the original data transmitted.According to the above model,when a TCC attack is detected,the SNR is reduced by injecting interference noise,so as to block the TCC’s data transmission.Experiments show that the stealthy TCC attack can effectively survive being detected by the existing detection methods with its BER as low as 4%.In contrast,with the proposed countermeasure applied,the detection accuracy jumps to 89%,and the BER of the TCC changes from less than 4% to 50%.Therefore,the proposed countermeasure can effectively block the TCC’s data transmission.