Title: “From Open RAN to Open Oceans: Toward a Programmable Infrastructure for Underwater Autonomy“

Abstract: Over the past decade, cellular infrastructure has undergone a significant architectural transformation. Functions once implemented in specialized hardware have moved increasingly to software on general-purpose computing platforms; interfaces have been opened and standardized; and network behavior has become accessible to adaptive control and optimization. Open RAN reflects this broader shift toward programmable, observable, and increasingly intelligent communication systems.

Underwater communication systems are beginning to undergo a similar transition, but under substantially different physical constraints. The underwater acoustic channel is highly variable in time and space, with propagation affected by multipath, Doppler, environmental conditions, platform motion, and interference. These characteristics make fixed-function architectures particularly limiting and increase the value of programmability and adaptation.

We argue that softwarization does more than make underwater modems more flexible: it changes what an underwater communications network can be. This talk examines that transition along five dimensions. Software-defined modems decouple signal processing from specialized hardware and support multiple waveforms and physical modalities. Programmable protocol stacks expose physical-, link-, and network-layer functions to runtime control. Onboard computing enables nodes to perform sensing and inference at the edge, allowing networks deployed for connectivity to become distributed and multistatic sensing systems. Digital twins provide environments in which communication and control policies can be developed and evaluated before deployment. Finally, these capabilities provide a foundation for distributed multi-agent autonomy, in which networked underwater systems jointly adapt communication, sensing, and vehicle behavior under intermittent connectivity. 

The result is a shift from the underwater modem as a fixed communication device to the network as a programmable infrastructure for autonomy: simultaneously a communication system, sensing system, inference platform, and control substrate. We discuss recent experimental results, open technical questions, and a research roadmap in this direction. We argue that this architectural transition is a prerequisite for scalable underwater autonomy: communication, sensing, inference, and control can no longer be treated as independent subsystems, but must increasingly be exposed to a common programmable and adaptive infrastructure.