Secure protocols for contactless credit cards and electronic wallets
The contactless credit card protocol in use today is insecure. The credit card industry has chosen to use the NFC channel for contactless transactions. However, reliance on NFC's short range has led to poor assumptions in the contactless credit card protocol. For example, the card assumes (sometimes incorrectly) that its ability to receive a solicitation implies the cardholder's intent to purchase. In this dissertation, we examine the protocol currently in use, and present a family of three replacement protocols to defend against its deficiencies.
First, we consider "outsider" attacks (e.g. eavesdropping, skimming attacks, relay attacks, and attacks facilitated by compromised points of sale) and design our first protocol to defend against these attacks. We call this protocol the Externally Secure CC Protocol, and design it using stepwise refinement. This protocol makes use of single-use "charge tokens" verifiable by the bank, while minimizing computation that needs to occur on the card.
Second, we identify two attacks which may be carried out by malicious retailers: Over-charge attacks and Transparent Bridge attacks. Both attacks are predicated on the customer's lack of participation in the protocol, and involve modifying or replacing a charge after it has been confirmed by the customer. We look to Electronic Wallet applications (such as Android Pay and Apple Wallet), which provide a channel between customer and card. We augment the Externally Secure CC Protocol using this channel to construct the Secure CC Protocol, binding charge tokens to a given price, and thus stymieing both outsider and malicious retailer attacks.
The Secure CC Protocol supports a property known as linkability: while only the bank can verify charge tokens, tokens from the same card can be recognized as such by the retailer. This property is also supported by the (insecure) protocol in use today, and is commonly used by retailers to construct marketing profiles on their customers. However, linkability has serious consumer privacy consequences, so we consider the converse property of unlinkability, where a retailer cannot identify different purchases as having been made by the same card. We require that our unlinkable protocol make use of existing infrastructure, so as not to require retailer cooperation. In response, we design the Unlinkable Wallet Protocol, leveraging techniques from the Secure CC Protocol to guard against malicious outsiders and retailers, while tunneling secure and unlinkable charge tokens through the protocol in use today.