A novel method to predict epilepsy seizures using a single iEEG (Invasive Electro Encephalogram) channel is presented in this paper. This method can potentially facilitate surgeries performed on epileptic patients by avoiding the insertion of grid or stripe electrodes. Both surgery complications and cost would become more acceptable. Also, using a single electrode would significantly reduce the computational effort and; consequently, the power consumption of the Bluetooth implant that will be designed in a later phase. Since epileptic seizures are characterized by a change in the EEG entropy, the signal is processed through a combination of linear and non-linear techniques to determine the seizure signature and to perform its prediction process. Nonlinear techniques allow therefore describing, in a suitable way, the output dynamics of the IEEG channel and then the different evolution stages of oscillatory states. Stability could be characterized by appropriate tools like Lyapunov exponent and phase diagramMoreover, we describe the proposed Bluetooth implant design to give more insight about how we process the seizure detection.