This Plenary Talk will introduce a paradigm for networked control systems operating in an uncertain environment, where the number of instants when a controller communicates with the plant (for example sends control signals to the plant) and the number of instants sensors communicate plant information to the controller are severely restricted, and both the sensors and the controllers have to make most of the few opportunities they have. Derivation of both optimal controller and optimal sensor policies under such nontraditional restrictions will be discussed, and the framework will be extended to scenarios where the communication at both links could be disrupted by an adversary, which also acts intermittently. In this latter case, what is sought for is not an optimal solution, but rather a saddle-point solution between two opposing parties. In both cases, the optimal or saddle-point policies are of the threshold type.