No abstract

With the fast growth of the automatic authentication and asset tracking usage in a wide variety of applications in different fields, government, logistics, transportation and retail are among the most supportive of the radio‐frequency identification (RFID) market. Security is a crucial issue and must be addressed seriously. RFID security must meet the public demand of data protection. Recently, several lightweight RFID authentication protocols conforming to the EPCglobal Class 1 Generation 2 (EPC C1‐G2) standard have been proposed. In this paper, we present efficient attacks against the authenticated RFID security mechanism of Chang et al. based on Chebyshev chaotic maps, which is the first solution that adopted the chaos in the RFID authentication process. It turns out that this protocol has fundamental weaknesses that can be used by an adversary to break the system. We will show that this protocol is vulnerable to tracking attack, secret disclosure attack, impersonation attack and desynchronization attack. The proposed attack techniques are in light of two flaws related to the message generation and the shared‐secret update process, which are not neatly scrutinized. Then, we propose an improved RFID authentication protocol based on the Chebyshev chaotic map hard problem, conforming to the EPC C1‐G2 standard with more flexibility, security and mobility for the RFID application. Copyright © 2014 John Wiley & Sons, Ltd.

In radar applications, the received echo signals reach the array elements via a multiplicity of paths despite the fact that there exists only one target. We address the problem of joint direction of arrival and Doppler frequency estimation using a sensor array in partially known additive noise. We consider a specular reflection model with a radar cross section fluctuating from one pulse repetition interval to another. The proposed model allows the estimation of more paths than sensors. Two approximate maximum likelihood algorithms are proposed. The first approach uses a linear expansion of the noise covariance matrix, whereas the second employs a combination of oblique projections and a zero-forcing solution to alleviate the effect of noise. In contrast to other classical methods, the two approaches are more robust to spatially correlated noise, and they employ more compact cost functions that reduce the dimension of the optimization search. Numerical simulations are provided to assess the basic performance of the two approaches, which are compared to the Cramér-Rao bound.

No abstract

No abstract