This thesis examines the security issues of Radio Frequency Identification (RFID) technology, one of the most promising technologies in the field of ubiquitous computing. Indeed, RFID technology may well replace barcode technology. Although it offers many advantages over other identification systems, there are also associated security risks that are not easy to address.

RFID systems can be classified according to tag price, with distinction between high-cost and low-cost tags. Our research work focuses mainly on low-cost RFID tags. An initial study and analysis of the state of the art identifies the need for lightweight cryptographic solutions suitable for these very constrained devices. From a purely theoretical point of view, standard cryptographic solutions may be a correct approach. However, standard cryptographic primitives (hash functions, message authentication codes, block/stream ciphers, etc.) are quite demanding in terms of circuit size, power consumption and memory size, so they make costly solutions for low-cost RFID tags. Lightweight cryptography is therefore a pressing need.

First, we analyze the security of the EPC Class-1 Generation-2 standard, which is considered the universal standard for low-cost RFID tags. Secondly, we cryptanalyze two new proposals, showing their unsuccessful attempt to increase the security level of the specification without much further hardware demands. Thirdly, we propose a new protocol resistant to passive attacks and conforming to low-cost RFID tag requirements. In this protocol, costly computations are only performed by the reader, and security related computations in the tag are restricted to very simple operations. The protocol is inspired in the family of Ultralightweight Mutual Authentication Protocols (UMAP: M2AP, EMAP, LMAP) and the recently proposed SASI protocol. The thesis also includes the first published cryptanalysis of SASI under the weakest attacker model, that is, a passive attacker. Fourthly, we propose a new protocol resistant to
both passive and active attacks and suitable for moderate-cost RFID tags. We adapt Shieh et.'s protocol for smart cards, taking into account the unique features of RFID systems. Finally, because this protocol is based on the use of cryptographic primitives and standard cryptographic primitives are not supported, we address the design of lightweight cryptographic primitives. Specifically, we propose a lightweight hash function (Tav-128) and a lightweight Pseudo-Random Number Generator (LAMED and LAMED-EPC). We analyze their security level and performance, as well as their hardware requirements and show that both could be realistically implemented, even in low-cost RFID tags.