edfas.org ELECTRONIC DEV ICE FA I LURE ANALYSIS | VOLUME 24 NO . 4 26 whenever tampering is detected.[11] The objectives and consequences of successful attacks are considered to depend on the strengths and capabilities of the adversaries. Table 1 summarizes the threat model where different attack approaches are presented, to compromise specific asset vectors. VULNERABILITY ANALYSIS Parameters: The extent of security vulnerabilities of nonvolatilememories to physical attacks can be assessed based on different quantifiable parameters. These parameters can be classified into three categories: dimensional, physical, and operational (refer to Fig. 2). Spot size is a dimensional parameter that is significant for every FA tool. The closer the cell size of a memory device is to the spot size of an imagingor probingmodality, the more vulnerable the device becomes to revealing the stored data. As emerging NVM cell size becomes smaller, spot size also needs to shrink down in tandem. With the advancement ofmodern FA tools, spot size is approaching nano-scale range, posingmore threat on emerging NVMs. Physical parameters include optical, thermal, or magnetic properties of the material that makes up the NVMdevice. An adversary can utilize the temperature and magnetic field present tomodulate the system, in order to get unauthorized access. The attacker can tamper with the functionality of the devices by manipulating and sensing the thermal activity of the devices. Themagnetic property of the device can easily be sensed by imaging modalities, such as magneto-optical current imaging (MOCI) and quantum diamond imaging (QDM). Operational parameters are those that change dur- ing different stages of operation of an NVM device. This includes electrical properties such as resistance or capacitance, kinetic energy of particles inside the device, or physical transformation of a specific structure inside the device. During operation, if an adversary can monitor these parameters of the target using specialized FA tools, they can infer what data is being written or read in the target. Security metric development: Quantifiable metrics for assessing the vulnerability of a chip design against physical attacks have not been established so far. Security metrics[12] can also demonstrate the effectiveness of a countermeasure against a specific attack, by comparing the value of that metric for a DUA with, or without, implementing the countermeasure. For developing quantifiable metrics against physical attacking modalities, it is a requirement to understand the scientific mechanism being used by themodality equipment, aswell as theNVM device. A basic framework for developing a securitymetric against a modality is shown in Fig. 3. The first step is to gain a comprehensive understanding of the physics of storing data in a particular type of NVM cell. This leads to an understanding of what are the potential parameters, or physical properties that can be probed or measured, such as, resistance, inductance, magnetic field, and optical activity. The ability of a probing or imagingmodality tomeasure a specific property or parameter in an NVM depends on certain technical parameters of the equipment to be used. These parameters could include resolution of imaging and spot size of a probing device. The most critical step is to find a relationship or dependency between the parameters tobemeasured Fig. 2 Parameters used to assess vulnerability of NVM devices to physical modalities. Fig. 3 A basic framework for developing security metrics.
RkJQdWJsaXNoZXIy MTMyMzg5NA==