Extraction of action potential waveform from summed neural signals

We consider the problem of interpreting the signal recorded from an electrode, which sums currents from a large number of neurons responding asynchronously to an applied stimulus. We use a markov chain numerical simulation to establish conditions under which the action potential waveform can be extracted from the autocorrelation of the summed signal, by means of the Levinson algorithm. We tested the efficacy of the algorithm in recovering the signal for two different action potential (AP) waveforms under different test conditions. We conclude that our system is both capable of recovering the average AP and detecting changes in its morphology.