Abstract: In order to sort frequency-hopping (FH) networks efficiently using frequency domain feature parameters, a two dimensional direction of arrival (2D-DOA) estimation algorithm based on the multiple signal classification (MUSIC) symmetrical compressed spectrum is proposed by using the space-time frequency analyzing method. Firstly, the spatial time-frequency matrix of each hop is constructed by using the time-frequency domain features of FH signals, and the covariance matrix of time-frequency domain is obtained consequently. Then the idea of conjugate subspace is introduced into the traditional MUSIC algorithm, and the dimension of the noise subspace is descended by the singular value decomposition on the intersection of noise subspace and its conjugate one. Finally efficient DOA estimation is realized through half-spectrum searching. At the same time, the time-frequency map is amended via the morphological filtering method in order to enhance the performance of the low SNR algorithm. Efficient extraction of hop will be implemented on amended maps. Theoretical analysis and simulation results show that the proposed algorithm can reduce the computational complexity of the traditional MUSIC algorithm by 50% while the RMSE is equivalent to it and the estimated success rate is higher than it.