Currently, there is no imaging method that is able to distinguish the functional activity inside nerves. Such a method would be essential for understanding peripheral nerve physiology and would allow precise neuromodulation of organs these nerves supply. Electrical impedance tomography (EIT) is a method that produces images of electrical impedance change (dZ) of an object by injecting alternating current and recording surface voltages. It has been shown to be able to image fast activity in the brain and large peripheral nerves. To image inside small autonomic nerves, mostly containing unmyelinated fibers, it is necessary to maximize SNR and optimize the EIT parameters. An accurate model of the nerve is required to identify these optimal parameters as well as to validate data obtained in the experiments. In this study, we developed two three-dimensional models of unmyelinated fibers: Hodgkin-Huxley (HH) squid giant axon (single and multiple) and mammalian C-nociceptor. A coupling feedback system was incorporated into the models to simulate direct and alternating current application and simultaneously record external field during action potential propagation. Parameters of the developed models were varied to study their influence on the recorded impedance changes; the optimal parameters were identified. The negative dZ was found to monotonically decrease with frequency for both HH and C fiber models, in accordance with the experimental data. The accurate realistic model of unmyelinated nerve allows the optimization of EIT parameters and matches literature and experimental results.