Antenna arraying is an effective method to improve the signal-to-noise ratio (SNR) of received signal in deep space communications. Although a variety of signal combining algorithms for antenna arraying have been studied, few of them can be applied to low SNR environment with small response time or short received sequence. In this paper, a new signal combining algorithm based on iterative optimization and data-aided SNR estimation is proposed. This algorithm transforms the problem of phase estimation among different antennas into a multi-function optimization, with the phase offsets as the variables and the SNR gain of combined signal as the objective function. By iteration, all variables could be well estimated. Simulation results show that the new algorithm could obtain excellent combining performance with a relatively short received sequence.