Jamming optimization in fully-connected, spread-spectrum networks

A novel framework for the analysis of the throughput-delay behavior of fully connected monohop code-division random access (CDRA) networks in the presence of additive white Gaussian noise and temporally selective jamming with fixed average power, and no side information, is presented. Three different types of probabilistic jamming models are described, namely the long term, the two-state-Markovian, and the block jammer. Using these models, it is shown that there exists an optimal value of the jammer´s temporal duty cycle ρ_t which minimizes the system´s normalized throughput and/or maximizes the average delay. This value depends on the coding rate, the access parameters, and the jamming waveform. Contrary to that, the optimal value of ρ_t in the unspread ALOHA case does not depend on the random-access (link-level) aspects of the network. For the synchronous case the dominant parameter for the analysis of the system´s performance is typically ρ_t. In the asynchronous case performance depends not only on ρ_t but also on the frequency of the on-off jamming transitions, as well as the relative size of the offset. Specific examples of direct-sequence (DS) and frequency-hopping (FH) spreading formats show how the general model applies