-Hardware piracy is a threat that is becoming more and more serious these last years. The different types of threats include mask theft, illegal overproduction, as well as the insertion of malicious alterations to a circuit, referred to as Hardware Trojans. To protect circuits from overproduction, circuits can be encrypted so that only authorized users can use the circuits. In this paper, we propose an encryption technique that also helps thwarting Hardware Trojan insertion. Assuming that an attacker will attach a Hardware Trojan to signals with low controllability in order to make it stealthy, the principle of the encryption is to minimize the number of signals with low controllability.

Testing a secure system is often considered as a severe bottleneck. While testability requires to an increase in both observability and controllability, secure chips are designed with the reverse in mind, limiting access to chip content and on-chip controllability functions. As a result, using usual design for testability techniques when designing secure ICs may seriously decrease the level of security provided by the chip. This dilemma is even more severe as secure applications need well-tested hardware to ensure that the programmed operations are correctly executed. In this paper, a security analysis of the scan technique is performed. This analysis aims at pointing out the security vulnerability induced by using such a DfT technique. A solution securing the scan is finally proposed.

International audienceThe conflict between security and testability is still a concern of hardware designers. While secure devices must protect confidential information from unauthorized users, quality testing of these devices requires the controllability and observability of a substantial quantity of embedded information, and thus may jeopardize the data confidentiality. Several attacks using the test infrastructures (and in particular scan chains) have been described. More recently it has been shown how test response compaction structures provide a natural counter-measure against this type of attack. However, in this paper, we show that even in the presence of response compactors the scan-based attack is still possible and it requires low complexity computation. We then give some perspectives concerning the techniques that can be used to increase the scan-based attack complexity without affecting the testability of the device