Abstract
This thesis proposes a jamming-resistant frequency hopping (FH) system that utilizes local channel observations for secret key generation (SKG). FH is a spread spectrum technique used in both military and consumer wireless applications to avoid jamming attacks, but requires pre-shared secret keys among communicating terminals, say Alice and Bob, to ensure that the same FH sequence is used at both sides. In our scheme, Alice and Bob utilize local observations of the channel between them as the source of common randomness to generate the shared secret key. By gathering multiple time slots into a frame, the sequence of channels observed in each frame can be used to determine the FH sequence in the next frame. In this case, the key generation rate must be high enough to identify the FH sequence in the next frame and, thus, to sustain the operation over time. However, by further considering the data transmission, an interesting tradeoff exists between the power allocated for SKG and channel estimation in the training phase and that for communication in the data transmission phase. Given the number of FH channels and the number of channels that the adversary can jam at once, we derive the minimum pilot signal power required for sustainability and also determine the optimal power allocation between pilot and data signals that maximizes the ergodic rate between the two users. We also propose a practical algorithm for this FH system to generate secret keys via vector quantizations. Simulations are provided to demonstrate the effectiveness of the proposed scheme.