The Internet is becoming the de facto standard for all telecommunications; services like voice-over-IP and Internet Protocol television are overtaking the technologies purpose-built for telephones and televisions. But the protocols underlying the Internet make assumptions about physical connections that do not always fit well with wireless communications: that all links are generally the same and provide close to real-time connectivity from one end of the link to the other. Wireless services are often intermittent, vary widely in the amount of bandwidth they offer, and, especially in the case of satellite services, can lag, making websites sluggish to respond to clicks. As new applications place new demands on Internet-based networks, “maybe it makes more sense to think about some radically different architectures,” says Preston Marshall, director of the wireless-­networking division at the University of Southern California’s Information Sciences Institute and a former manager for many of the advanced wireless technology programs at the Defense Advanced Research Projects Agency (DARPA).

This kind of thinking informs efforts to integrate technologies such as delay-­tolerant networking (DTN) with Internet protocols. Because DTN allows communication even when establishing an end-to-end link is impossible, it would make networks more robust. It’s also well suited for applications such as machine-to-machine communication. For example, a meter on a “smart” electric grid may need to exchange only small amounts of data with the power company within a relatively broad window of time; it doesn’t need a high-bandwidth real-time connection.

Beyond DTN, it may ultimately be possible to shift between different kinds of wireless links from moment to moment, depending on the type of data to be accessed. A user could browse for a movie title using a low-lag but low-bandwidth cellular connection and then download the movie over a high-lag, high-bandwidth satellite connection.