7.8 Methods for Controlling Safety …
711
and/or owners of IP rights. This method is implemented using models, schemes, and
technical documentation that is not transfered to the production plant.
In the last decade, several new and fairly effective methods of hardware metering
have been suggested to ensure control of IP owners over their ICs after production,
including [94, 97, 125, 133–135]. For example, the work [116] contains an example
of classification of all such methods known at the moment.
7.8.2 Models of Monitoring Safety of Produced Microcircuits
As indicated in [116], all methods of hardware metering can be divided into two
main categories: passive and active. Passive methods ensure the possibility of unique
identification of each chip or its functionality in such manner that it can be monitored
passively. Such methods of passive hardware metering had been used before for
many decades, physically identifying the serial number of each device or saving
identifiers in nonvolatile memory. The authors of [116] refer to the above methods
as indented serial numbers, while the second method is known as digitally stored
serial numbers. Digitally stored serial numbers also provide possibilities for passive
tracking of manufactured devices. Due to the potential vulnerability of methods of
serial numbers and methods of digital identification numbers to copying and remote
attacks, the first methods of creating such uncloneable IDs based on the values
of random fluctuations (spreads) of the IC production process parameters [130].
Since any technological process always includes a random factor, which cannot be
controlled or copied, this class of identification methods is known as uncloneable
identification (ID). Uncloneable identifiers (IDs) are a form of physical uncloneable
functions, which are sufficiently classified [136]. A specific PUF type suitable for
creation of such secret keys is known as weak PUF; these PUFs can be used as
uncloned IDs for hardware IC measurements. As opposed to uncloneable IDs, the
authors of [136] refer to all previously known methods of chip identification as
reproducible identification.
Soon after introducing ICID methods, a number of researchers suggested an
entirely new method of passive metering [128, 129]. In this method, identifiers were
linked to internal functional elements of the chip during the logic synthesis stage,
due to which each separate logic function of the microcircuit got its unique signature. This type of passive metering is distinguished into a separate class of functional
metering methods. Both uncloneable and reproducible identifiers can be linked to
specific functions of the microcircuit to create a unique signature for each of them.
In this case, the function will remain unchanged for any microcircuit user in terms
of input/output data, and only the set of internal transactions, which is invisible to
the user, will be unique for each separate chip.
Most passive metering methods described to date are based on introducing an
additional component or using the structure to place identifiers or perform functional
metering. Another method of passive metering, which can uniquely identify each chip
without adding any components or introducing changes in the design is known as
711
and/or owners of IP rights. This method is implemented using models, schemes, and
technical documentation that is not transfered to the production plant.
In the last decade, several new and fairly effective methods of hardware metering
have been suggested to ensure control of IP owners over their ICs after production,
including [94, 97, 125, 133–135]. For example, the work [116] contains an example
of classification of all such methods known at the moment.
7.8.2 Models of Monitoring Safety of Produced Microcircuits
As indicated in [116], all methods of hardware metering can be divided into two
main categories: passive and active. Passive methods ensure the possibility of unique
identification of each chip or its functionality in such manner that it can be monitored
passively. Such methods of passive hardware metering had been used before for
many decades, physically identifying the serial number of each device or saving
identifiers in nonvolatile memory. The authors of [116] refer to the above methods
as indented serial numbers, while the second method is known as digitally stored
serial numbers. Digitally stored serial numbers also provide possibilities for passive
tracking of manufactured devices. Due to the potential vulnerability of methods of
serial numbers and methods of digital identification numbers to copying and remote
attacks, the first methods of creating such uncloneable IDs based on the values
of random fluctuations (spreads) of the IC production process parameters [130].
Since any technological process always includes a random factor, which cannot be
controlled or copied, this class of identification methods is known as uncloneable
identification (ID). Uncloneable identifiers (IDs) are a form of physical uncloneable
functions, which are sufficiently classified [136]. A specific PUF type suitable for
creation of such secret keys is known as weak PUF; these PUFs can be used as
uncloned IDs for hardware IC measurements. As opposed to uncloneable IDs, the
authors of [136] refer to all previously known methods of chip identification as
reproducible identification.
Soon after introducing ICID methods, a number of researchers suggested an
entirely new method of passive metering [128, 129]. In this method, identifiers were
linked to internal functional elements of the chip during the logic synthesis stage,
due to which each separate logic function of the microcircuit got its unique signature. This type of passive metering is distinguished into a separate class of functional
metering methods. Both uncloneable and reproducible identifiers can be linked to
specific functions of the microcircuit to create a unique signature for each of them.
In this case, the function will remain unchanged for any microcircuit user in terms
of input/output data, and only the set of internal transactions, which is invisible to
the user, will be unique for each separate chip.
Most passive metering methods described to date are based on introducing an
additional component or using the structure to place identifiers or perform functional
metering. Another method of passive metering, which can uniquely identify each chip
without adding any components or introducing changes in the design is known as