4.11 Analytical Review of Basic Techniques …
429
paper count does not grow and even drops a little bit. This can indicate as follows:
(1) the structure of Trojans or malicious circuits has been explored so much that
developing novel Trojan attacks has become too challenging or (2) there are many
types of functional Trojans with various triggers or payloads that are still hard to
detect, so researchers realize that more effort should be spent on the development of
countermeasures. Figure 4.71 b shows that countermeasure techniques have grown
faster in the last 4 years and the paper count has almost doubled. More research
resources have been devoted to overcoming the hardware Trojans, so different types
of countermeasures have been developed in the last decade. Figure 4.70 illustrates a
timeline about the appearance of different Trojan countermeasures. In this plot, we
can also see the general trend shifts from Trojan detection to design-for-trust and
split manufacturing approaches in recent years.
It should be emphasized that the methods of Trojan detection and Trojan prevention attempt to address the hardware Trojan issue from different perspectives. Trojan
detection techniques contain all typical approaches shown in the first column in
Fig. 4.70 and a portion of design methodologies in the second column that facilitate Trojan detection. The rest of the countermeasures are trying to either prevent
hardware Trojan insertion or prevent a Trojan’s malicious behavior, including splitmanufacturing-for-trust approaches. Figure 4.71b shows the publication trends for
Trojan detection and Trojan prevention. Few papers about Trojan prevention were
published before 2013, but prevention methodology got sufficient attention only
during 2 years from 2013 to 2014.
Researchers have developed various detection methods but detection of “tiny”
concealed Trojans is still very challenging. Additionally, most of the hardware Trojan
detection techniques are still developed based on the available golden models or
golden ICs.
Obtaining a golden model or reference (golden) IC in the semiconductor supply
chain is an extremely difficult (practically impossible) task. Therefore, hardware
Trojan prevention might be a more effective and practical way to overcome the
hardware Trojan threat. However, as shown in Fig. 4.71b, newly published detection
techniques are still two times more than those prevention techniques in the categories
of design-for-trust and split manufacturing-for-trust. It is obvious that techniques for
prevention of hardware Trojan insertion in the chips deserve more attention in the
near future.
In the previous sections, the authors [259] discussed the past research, current
and future Trojan attack models, as well as the trends of current hardware Trojan
research. A major conclusion from the previous discussion is just how much is still
left partially or completely unsolved. Below we look more closely at the current
situation.
429
paper count does not grow and even drops a little bit. This can indicate as follows:
(1) the structure of Trojans or malicious circuits has been explored so much that
developing novel Trojan attacks has become too challenging or (2) there are many
types of functional Trojans with various triggers or payloads that are still hard to
detect, so researchers realize that more effort should be spent on the development of
countermeasures. Figure 4.71 b shows that countermeasure techniques have grown
faster in the last 4 years and the paper count has almost doubled. More research
resources have been devoted to overcoming the hardware Trojans, so different types
of countermeasures have been developed in the last decade. Figure 4.70 illustrates a
timeline about the appearance of different Trojan countermeasures. In this plot, we
can also see the general trend shifts from Trojan detection to design-for-trust and
split manufacturing approaches in recent years.
It should be emphasized that the methods of Trojan detection and Trojan prevention attempt to address the hardware Trojan issue from different perspectives. Trojan
detection techniques contain all typical approaches shown in the first column in
Fig. 4.70 and a portion of design methodologies in the second column that facilitate Trojan detection. The rest of the countermeasures are trying to either prevent
hardware Trojan insertion or prevent a Trojan’s malicious behavior, including splitmanufacturing-for-trust approaches. Figure 4.71b shows the publication trends for
Trojan detection and Trojan prevention. Few papers about Trojan prevention were
published before 2013, but prevention methodology got sufficient attention only
during 2 years from 2013 to 2014.
Researchers have developed various detection methods but detection of “tiny”
concealed Trojans is still very challenging. Additionally, most of the hardware Trojan
detection techniques are still developed based on the available golden models or
golden ICs.
Obtaining a golden model or reference (golden) IC in the semiconductor supply
chain is an extremely difficult (practically impossible) task. Therefore, hardware
Trojan prevention might be a more effective and practical way to overcome the
hardware Trojan threat. However, as shown in Fig. 4.71b, newly published detection
techniques are still two times more than those prevention techniques in the categories
of design-for-trust and split manufacturing-for-trust. It is obvious that techniques for
prevention of hardware Trojan insertion in the chips deserve more attention in the
near future.
In the previous sections, the authors [259] discussed the past research, current
and future Trojan attack models, as well as the trends of current hardware Trojan
research. A major conclusion from the previous discussion is just how much is still
left partially or completely unsolved. Below we look more closely at the current
situation.