Error-Control Coding for Physical-Layer Secrecy

The renewed interest for physical-layer security techniques has put forward a new role for error-control codes. In addition to ensuring reliability, carefully designed codes have been shown to provide a level of information-theoretic secrecy, by which the amount of information leaked to an adversary may be controlled. The ability to achieve information-theoretic secrecy relies on the study of alternative coding mechanisms, such as channel resolvability and privacy amplification, in which error-control codes are exploited as a means to shape the distribution of stochastic processes. This use of error-control codes, which goes much beyond that of correcting errors, creates numerous new design challenges. The objective of this paper is threefold. First, the paper aims at providing system engineers with explicit tools to build simple secrecy codes in order to stimulate interest and foster their integration in communication system prototypes. Second, it aims at providing coding and information theorists with a synthetic overview of the theoretical concepts and techniques for secrecy. Finally, it aims at highlighting the open challenges and opportunities faced for the integration of these codes in practical systems.

I. INTRODUCTION

The emergence of wireless communication networks has profoundly affected every aspect of our modern societies. By allowing access to information from virtually anywhere at anytime, wireless devices have not only increased the efficiency of existing infrastructures but also fostered the emergence of new services and systems, ranging from electronic payment to implanted medical devices. However, this seamless connectivity also opens numerous vulnerabilities that are causing growing public concern about secrecy and privacy. As an increasing amount of sensitive information is now conveyed through wireless networks, from financial and medical records to travel patterns and consumption habits, the intrinsically open nature of the wireless transmission medium makes all communications particularly susceptible to eavesdropping with nothing but the simplest off-the-shelf laptop. While the upper layers of the protocol stack traditionally ensure secrecy by means of public-key- and private-key-based encryption algorithms, new technologies and devices are emerging in which the power and complexity constraints make it difficult to justify a direct deployment of standard security solutions. For instance, implanted medical devices, wireless sensors, or RFID tags, deployed in a home as part of the “Internet of Things” may not have the capability to host a full-blown public key infrastructure, hence rendering key generation and distribution more complex than the system can afford, either in terms of communication or computation resources. Motivated by these challenges, there has been a renewed interest in keyless techniques ensuring an intrinsic level of secrecy at the physical layer of communication systems. The rationale for this “physical-layer security” approach is that secret keys are a means to introduce randomness in the system, and that one could potentially reduce the burden of randomness generation by harnessing the randomness stemming from the communication channel itself. Numerous works studying secure communication over the wiretap channel [1] and secret-key generation from correlated observations [2], [3] also support the benefits of physical-layer security techniques. All these studies suggest that physical-layer security schemes could achieve some level of information-theoretic secrecy. Although the information-theoretic models admittedly suffer from several weaknesses, such as idealized assumptions regarding the knowledge of the channel, one of the major hurdles toward broader acceptance has been the absence of explicit low-complexity algorithms ensuring a provable level of information-theoretic secrecy. However, there has recently been much progress toward this goal, thanks in part to new conceptual approaches to secrecy and advances in error-control coding.