Security models for delegated keyword searching within encrypted contents

Abstract

Within modern internet infrastructure including networks that are ubiquitous, there is often a need for delegatable communication between nodes without compromising the confidentiality of information. In practice, this should be enforced while allowing some basic functionality for intermediate delegated nodes such as searching through encrypted content. This can be achieved using a Public key encryption with keyword search (PEKS) scheme, first proposed by Boneh et al., which enables to search publicly encrypted messages for keywords without revealing any information about the message. The issue of PEKS schemes being vulnerable to keyword guessing attacks (KGAs) was first shown by Byun et al., and two of the most recent PEKS schemes, i. e., due to Rhee et al. in (ASIACCS, pp 376-379, 2009; IEICE Electron Express 6(5):237-243, 2009) and (J Syst Softw 83(5):763-771, 2010), respectively, are designed with this security in mind. In this paper, we treat this KGA problem in detail and define new security models to capture KGAs against PEKS and designated PEKS schemes. These models are more security sufficient than the model considered by Rhee et al. (J Syst Softw 83(5):763-771, 2010); indeed the latter model does not afford sufficient adversarial capability in the sense that it is much weaker than the adversarial capability considered in the original IND-CKA model of Boneh et al. Our new models allow to capture KGAs on three recent designated PEKS schemes that cannot be captured in the weaker model of Rhee et al. © 2012 The Brazilian Computer Society.