Abstract — In this paper, we describe a method of optimum temperature modulation of metal oxide semiconductor (MOS) based gas sensor, operated in dynamic temperature measurement for identification of gas. The volatile organic compound (VOC) sample space consists of fourteen laboratory chemicals sampled at various concentration. We have used eleven number of gas sensors, manufactured by Figaro sensors, Japan. The heater of the sensors were modulated with sawtooth heating waveform of different frequency. Features were obtained from the response waveforms using wavelet transformation. The best frequency of heating waveform were determined using classification accuracy obtained using support vector machine.