This project develops a new approach to protecting information from interception by adversaries in wireless communication networks, with a particular focus on the scenario where an eavesdropper obtains a higher-quality version of the transmitted signal than the desired recipient. The standard method of providing protection in such a situation is to encrypt the information so that it is beyond the eavesdropper's computational capabilities to decrypt the message; this approach has met with longstanding critiques, including the lack of a fundamental proof establishing the difficulty of the problem presented to the adversary. An approach to everlasting security initiated in this project exploits the non-commutativity of certain nonlinear operators: different orderings of the same two nonlinear systems applied to the same input signal can produce different outputs. This property is exploited for security in wireless communication links by employing a short-term cryptographic key to force the eavesdropper's signal to be subjected to nonlinear operations in the reverse order of that of the signal at the desired recipient. The appropriate design of the nonlinear systems yields differences in the signals at the desired recipient and eavesdropper that can be exploited for everlasting secrecy - even when the ephemeral cryptographic key is revealed to or broken by the eavesdropper immediately after transmission. The development of this approach under applicable system constraints and a comparison of the obtained security performance to current approaches form the technical core of this project.
By establishing a new approach that expands the system conditions under which everlasting security is obtained, a new avenue of research opens in this critical area of wireless security. In particular, a security approach that exploits characteristics of the eavesdropper's receiver rather than assumptions of noise or loss on the propagation channel will provide an extra layer of security that has not been exploited in current systems. Also, integral to the project is the involvement of undergraduate and graduate students, and a research experience for high school teachers will draw technical ideas from the project. These activities provide workforce development in a critical area by providing a compelling application and further broaden the impact of the project.
There are some messages in wireless communication systems which need to be kept secure forever from eavesdroppers. The standard security approach of cryptography allows the eavesdropper to record the ciphertext, but then relies on the difficulty of solving a hard mathematical problem to keep the eavesdropper from extracting the message content from the ciphertext. However, if the eavesdropper later obtains the key, breaks the cryptographic implementation, or achieves unforeseen advances in computation, the message can be compromised. Information-theoretic secrecy focuses on a keyless approach that keeps the eavesdropper from recording a signal from which the message contents can ever be extracted - even with unbounded advances in computation. Information-theoretic secrecy must therefore rely on an advantage of the communicating parties over the eavesdropper; for example, the intended recipient might be closer to the transmitter than the eavesdropper and thus receive a higher fidelity version of the transmitted signal. However, in a wireless communications environment, the eavesdropper might be closer than anticipated and hence the message contents immediately compromised. This risk has been a major impediment to the implementation of information-theoretic security in practice, and, more generally, a major impediment to the development of an approach in wireless communications to obtain everlasting secrecy. In this project, we focus on inherent weaknesses in the front-end of wireless communication receivers that can be exploited to keep the eavesdropper from recording a signal of sufficient fidelity to extract the message contents in the future. The received wireless communications signal must be stored digitally by the eavesdropper in order to be processed at a later time, and the conversion process from the received signal to its digital representation (analog-to-digital, or A/D, conversion) is a challenging operation for any receiver. In particular, A/D converter technology progresses very slowly, and the aperture jitter which limits such has not been reduced significantly since 2005. Hence, we consider an approach which exploits weaknesses in the A/D converter of the eavesdropper to obtain everlasting security. The proposed technique works as follows. First, the transmitter and intended recipient establish an ephemeral cryptographic key, which need only be kept secret for the (short) duration of the wireless transmission. Next, the transmitter sends a signal with significant distortion based on the key. Because the intended recipient has knowledge of the key, he/she can remove the distortion before recording the signal. However, the eavesdropper, without current knowledge of the key, must store the distorted signal. Even if we assume that the eavesdropper is then handed the key immediately, and hence finds out what distortion the transmitter has added, the loss in the recording operation is not recoverable, as the A/D conversion operation is nonlinear, and nonlinear operators cannot necessarily be commuted. Information-theoretic secrecy is then obtained. In this project, we have developed three instantiations of the general technique, with each instantiation employing a different form of intentional transmitter distortion that is removed at the receiver of the intended recipient. The most promising technique is also the simplest: the transmitter uses the shared cryptographic key as the basis for the value of a large jamming signal to be added to the transmitted signal. The intended recipient, with knowledge of the key, then easily subtracts off the jamming before A/D conversion, while the eavesdropper is forced to process a very large signal through his/her A/D converter. The loss of fidelity in the latter operation achieves information-theoretic (and, hence, everlasting) secrecy, even if the eavesdropper obtains the key and attempts to subtract the jamming signal from the stored digital representation of the received signal. Extensive numerical results support our conclusions. Even if the eavesdropper receives a higher fidelity signal and has a better A/D converter than the intended recipient, everlasting secrecy can be obtained through our jamming approach. This is true even in the case when the eavesdropper might have access to an exact version of the transmitted signal at the input to his/her receiver, because loss in the eavesdropper's A/D conversion process can still be exploited. Finally, we have extended our results to wideband communication channels, where the jamming approach is employed in conjunction with traditional frequency-hopping, and simulation results show that significant rates of perfectly secure message bits can be conveyed even against state of the art A/D converter technologies.
