Abstract: When phase spectrum method was applied in TDOA (Time Difference of Arrival) estimation for narrow-band communication signal, the effect of phase unwrapping was bad as signal’s power spectrum was non-flat, Aiming at this phenomenon, this paper proposed a modified weighted least squared fitting method that utilizing these unwrapped phase data more reasonably, to improve the accuracy of TDOA estimation based on phase spectrum method. And, combined with the modified maximum likelihood phase spectrum method, this paper also designed a simple and practical data compression method in frequency domain for TDOA estimation of narrow-band communication signal, to balance he data amount for transmission and the accuracy of TDOA estimation well. Simulations prove that, the modified phase spectrum method can improve the accuracy of TDOA estimation for narrow-band communication signal obviously, and after data compression by proposed method under certain large compression ratio, good enough TDOA estimation results can be still acquired, so the proposed data compression method has good application future in resource-constrained field like sensors network.