UC Riverside

Wireless network security is an increasingly important problem for current and future generations of wireless networks as the computing powers available for attackers to break the traditional encryption methods increase rapidly. This problem is compounded by the need for low-latency required by real-time artificial intelligence and virtual reality applications. This thesis focuses on the issue of information security against eavesdropping and examines the performances of two types of novel methods.

The first type is called continuous encryption which encrypts and decrypts transmitted messages directly using continuous numbers for which good estimates are only available at legitimate nodes. The examples of such continuous numbers include the reciprocal channel parameters between two legitimate wireless nodes in typical scattering rich environment. Continuous encryption does not need the traditional step for two nodes to first agree upon a secret key from their estimates of a reciprocal channel response, and hence reduces the encryption latency. This thesis also examines an application of continuous encryption for UAV communications, the advantages of our proposed continuous encryption function over prior continuous one-way functions, and a useful role of continuous encryption for secret-key generation.

The second type is based on channel probing for secret-key generation or secret-message transmission. The channel probing methods do not require any reciprocal channel response between two legitimate nodes, and are able to yield a positive secret-key rate and/or secrecy rate in bits per channel use even if the channel coherence time is infinite. This is again useful for low-latency security. This thesis will present an insightful expression of the secret-key capacity for Gaussian probing signals over Gaussian MIMO channels, and also a number of power scheduling policies for a multiple carrier version of Secret-message Transmission by Echoing Encrypted Probes (STEEP).