Strong adaptive radar, such as cognitive radar (CNR), can perform various missions while ensuring its own security in electronic warfare, via detecting environments and changing the radar parameters in real time. Unfortunately, most of the current military countermeasures, such as jamming-based electronic countermeasures, have rarely been related to jamming for CNR. Since the behaviours of radar in the traditional design of the jammer-radar scenario are always static, it is easy to create a subjective or local optimal jamming effect. In order to dynamically analyse the execution process of a complete jamming radar mission, this work establishes an equivalent attack-defence game in which the radar is regarded as a defence decision agent, and the jammer is an attack decision agent. The attributes of the game's players, the rules of the game, and the conditions for the end of the game are set clearly by setting reasonable parameters. After searching for antagonism strategies by exhaustive method, it can be found that the survivability of the predictive cognitive jamming is much stronger than that of the normal jamming based on real-time sampling data of radars. This conclusion is demonstrated through a 1 ms simulation of the game process.