With the development of electronic warfare technology and the complex electromagnetic environments of battlefield, the confrontation between the guidance radar and the jammer has become increasingly fierce. Aimed at the problem that performance degrades in the process of confrontation when the guidance radar and the jammer are out of condition to obtain the complete information of the other party during the game, a guidance radar and jamming Bayesian game model is established based on the SINR criterion, and the unknown information is expressed in the form of target probability set through the Harsanyi conversion method. Optimized waveforms of radar and jammer are designed based on game model solution, signal frequency domain energy is distributed based on secondary waterfilling and iterative waterfilling methods, and the existence of Bayesian game Nash equilibrium is studied. Finally, different strategies of radar and jammer are analyzed through simulation. The specific form of the signal verifies the waveform detection performance. The simulation results prove that the detection probability of the iterative waterfilling signal and the second waterfilling signal is up to 15.41% and 12.79% ( higher than that of the chirp signal) respectively, providing a solution for the optimization of the guidance radar signal in the incomplete information game.