Signal jamming seems like an annoyingly simple way to sabotage a lot of interesting ideas. It has historically been used to block reception of propaganda broadcasts, and also to ruin military communication. My interests are more in the latter, but wider and looking far into the future.
The fundamental problem of interest is that of making good decisions after losing contact with superiors. This can be encountered in a variety of situations without dramatic consequences, but is amplified by two factors that will only become more relevant in the future: One of the factors is how much is at stake. An example of high-stake autonomous decision making is that of having to decide whether to launch nuclear weapons, as in this story from the Cold war. The other factor is the importance of unmanned systems. Even a remotely controlled system must have some fall-back mode when communication is lost, and for certain systems, typically expensive and/or dangerous ones, especially those exercising (military) power, neither inaction, deterministic behavior nor self-destruction are acceptable options. An example would be a space-deployed nuclear attack surveillance and defense system.
If communication can always be jammed, the only solution I see is increasing local autonomy. For sufficiently complex missions, this entails solving the problem of aligning the goals of agents possessing artificial general intelligence with the goals of their owner. That is in itself an immensely difficult problem. Since communication latency inevitably grows with distance, and power inevitably shrinks, sufficiently remote operations alone will require solving the same problem. In a question at the World Building StackExchange, I previously proposed that the difficulties of goal alignment could motivate a super-intelligent ruler agent to limit its own autonomous presence, and thus its power, far away from its base. Such considerations are not exclusive to an artificial super-intelligence, though possibly both compatible with one and a (nearly) convergent instrumental goal. The principle outlines/reiterates a solution to Fermi's paradox, in the sense that very advanced civilizations will choose to stay at home and perhaps even hide their existence for safety.
Here, however, we may restrict the interest to cases where signal jamming is the foremost concern, i.e. where (for any reason) all acceptable paths to success crucially depend on working communication. Therefore, this is a physics question. It also has (relatively) down-to-Earth relevance in the context of safety against nuclear war.
The Wikipedia entry on Radio jamming mentions several methods for jamming radio signals, including (among others) noise and pulses. The countermeasures mentioned are increasing transmission power and using other frequencies than those jammed.
Intuitively, as a non-physicist, I presume that a way to counteract noise-based jamming is to portion out the transmissions in short, high-power pulses that stand out from the jamming noise. Furthermore, assuming that pulse-based jamming likewise depends on prioritizing its input for high-power pulses, it should leave vacant signaling time between the pulses. If the signal to be jammed has a predictable pattern, the jammer can target that pattern. If the signal is not made robust through redundancy, the jammer can break the communication line merely by spewing randomly spaced pulses.
But what if the signal of powerful pulses follow an externally unpredictable pattern (defined by something like secret key encryption) and has built-in error correction based on redundancy? Would that still be jammable? And if not: Would it be practically usable? And would it have side effects of jamming other communication or destroying electronic equipment? (In the latter case, the technology could still find uses in crucial applications like nuclear defense.)
Wikipedia apparently does not even have an article named pulse-based communication. I did, however, find an article on Ultra-Wideband communication, which is sometimes referred to as "pulse radio". It does not look particularly promising, being described as having short range and low Signal-to-Noise Ratio. I am interested in hearing some expert views on this.