The memristive technology is predicted to bring a qualitative change not only to the high capacity memory, but also to the embedded application world. One of the application domains that can benefit from introduction of memristors is embedded security. The addition of memristor-based circuits will enrich the toolset for designing hardware security primitives, but may also introduce new security vulnerabilities.
For example, resistive memory may solve the long standing problem of embedded security devices of finding an appropriate secure and cost effective embedded non-volatile memory.
However, how secure is this memory, for example in light of the known side channel or physical attacks, is currently an open question. On the other hand, the non-linearity and non-volatility of the memristor may allow for building new or enhancing the existing hardware security primitive, for example the Physically Unclonable Functions (PUF), True Random Number Generators (TRNG) or encrypted embedded
We are looking at two directions related to security with memristors:
1. Investigation of immunity of the resistive memory to various threats, such as power analysis, fault injection or invasive attacks.
2. Building secure primitives that exploit the unique characteristics of memristors; for example a keyed secure hash function.
This research is partially funded by Cisco.