Active galaxies hosting radio jets can exhibit distinct active phases marked by two sets of radio lobes. Typically, these episodic radio sources have been identified through morphological observations. In addition, spectral characteristics-based methods are also employed wherever multi-frequency deep radio observations are available. However, these methods are inefficient in detecting restarted radio sources that do not exhibit a clear morphology. To address this, a method of using the spectral curvature ( \({{\textrm{SPC}}} = \alpha ^{1400~{\textrm{MHz}}} _{150~{\textrm{MHz}}}-\alpha ^{150~{\textrm{MHz}}} _{74~{\textrm{MHz}}}\) ) to identify restarted radio sources is presented. This is based on the fact that restarted radio sources with significant remnant emission are expected to have concave spectra in contrast to the convex or straight spectra observed in most radio sources. We use available wide area radio surveys in the range of frequencies from 74 MHz to 1.4 GHz to search for episodic radio sources and to shortlist 9,405 sources based on the criteria of \({\textrm{SPC}} \ge 0.5 \) . The candidates thus identified can be followed up for detailed morphological and spectral index studies. This method will find application in the automated identification of episodic radio sources in large radio sky surveys from telescopes like LOFAR and SKA.
