I'm pretty certain that RF jammers work by overpowering the target signal with their own higher powered signal at the same frequency. So the question is, how does anti-jammer technology negate the effects of a jammer?
One method is by actively steering the antenna (mechanically or electronically) to place a "null" in the direction of the jammer, reducing its signal strength significantly, while affecting the desired signal minimally, if at all.
Also, assuming the jammer signal strength isn't so strong that it saturates the receiver front end, advanced DSP techniques can be used to estimate and cancel the effects of the jamming signal. The communications protocol itself can be designed to optimize the ability to do this. The problem for the jammer is to mimic the desired signal closely enough to confuse the anti-jam algorithm.
When directional antennas are not practical, spread-spectrum techniques can be used. This causes the bandwidth of the signal to be very large, with very little energy at any particular frequency, making it much more difficult to jam. A similar approach is frequency hopping, where the carrier frequency is changed frequently according to a predetermined schedule. Of course, this must be done at both the transmitter and receiver.
For a signal to be received, the transmitted power at the frequency being monitored must be large relative to the amount of power the jammer is transmitting at that frequency at that moment. Even if a jammer has more power available than the entity that's trying to transmit useful information, total power will still be limited; that power must be divided among all the frequencies to be jammed. Additionally, a receiver which is expecting to receive data at a slow speed may be more frequency-selective than one which is trying to receive data at a faster speed.
Suppose a device were trying to transmit 1,000 bits/second using frequencies from 2,414.012 Mhz to 2,414.013Mhz. A jammer which could identify that frequency could overpower that transmission by concentrating all its power at that frequency.
Now suppose the device sent 100-bit bursts of data, with each burst being sent using one of 5,000 different 2kHz-wide frequency bands somewhere in the range of 2,410Mhz-2,420Mhz, selected via some method that the sender and receiver both know, but the jammer does not. For the jammer to obstruct even 10% of the transmissions, it would have to send as much power at each of 500 bands as would have been required to completely jam the single-frequency transmission. In other words, the use of frequency-hopping would have increased the amount of power required to obtain even 10% jamming to 500 times the level required to jam a non-hopping signal.
If the party trying to transmit data weren't using any form of forward error correction, successfully jamming 10% of the transmissions might make them all useless. On the other hand, if 90% of packets can get through, the transmitter can include some redundant information so as to allow reconstruction of the original message. The jammer's ability to jam 10% of the packets may increase the cost of transmitting the data by 20% or 25% (depending upon desired reliability), but the fact that a 500x increase in jammer power force only forces a 20% increase in transmit power isn't exactly a win for the jammer.
A sufficiently powerful jammer will be able to prevent a sender who is confined to using a certain frequency band from transmitting more than a certain amount of data reliably. On the other hand, the required ratio of jammer power to transmit power will be roughly proportional to the ratio of the available spectrum to the amount that would be needed for "simple" transmission. When transmitting low data rates in a wide area of spectrum, that ratio can be made quite large.