Alternative Title物理複製不能関数における安全性の評価と向上に関する研究
Note (General)In this thesis, we focus on Physically Unclonable Functions (PUFs), which are expected as one of the most promising cryptographic primitives for secure chip authentication. Generally, PUFbased authentication is achieved by two approaches: (A) using a PUF itself, which has multiple challenge (input) and response (output) pairs, or (B) using a cryptographic function, the secret key of which is generated from a PUF with a single challenge-response pair (CRP). We contribute to:(1) evaluate the security of Approach (A), and (2) improve the security of Approach (B). (1) Arbiter-based PUFs were the most feasible type of PUFs, which was used to construct Approach (A). However, Arbiter-based PUFs have a vulnerability; if an attacker knows some CRPs, she/he can predict the remaining unknown CRPs with high probability. Bistable Ring PUF (BR-PUF) was proposed as an alternative, but has not been evaluated by third parties. In this thesis, in order to construct Approach (A) securely, we evaluate the difficulty of predicting responses of a BR-PUF experimentally. As a result, the same responses are frequently generated for two challenges with small Hamming distance. Also, particular bits of challenges have a great impact on the responses. In conclusion, BR-PUFs are not suitable for achieving Approach (A)securely. In future work, we should discuss an alternative PUF suitable for secure Approach (A).(2) In order to achieve Approach (B) securely, a secret key ? generated from a PUF response?should have high entropy. We propose a novel method of extracting high entropy from PUF responses. The core idea is to effectively utilize the information on the proportion of ‘1’s including in repeatedly-measured PUF responses. We evaluate its effectiveness by fabricated test chips. As a result, the extracted entropy is about 1.72 times as large as that without the proposed method.Finally, we organize newly gained knowledge in this thesis, and discuss a new application of PUF-based technologies.
2014
Collection (particular)国立国会図書館デジタルコレクション > デジタル化資料 > 博士論文
Date Accepted (W3CDTF)2018-01-02T17:18:43+09:00
Data Provider (Database)国立国会図書館 : 国立国会図書館デジタルコレクション