Oceans cover roughly 71% of the Earth’s surface and underpin much of life on it, yet about 95% of them remain unexplored, largely because the underwater environment is so unforgiving to the tools we rely on above the surface. An Internet of Underwater Things (IoUT) could open a new era of ocean applications, but its communication, networking, and localization differ from terrestrial IoT in almost every respect. [1], [6]
The trade-off is fundamental. Underwater acoustic networks (UANs) reach long distances of 1 to 10 km but are throttled to tens of kbps and suffer highly variable delay, because acoustic signals have limited bandwidth and travel at only ~1500 m/s. Underwater optical wireless networks (UONs) flip those constraints: at the cost of short range (under ~200 m), light propagates at ~2.55×10⁸ m/s and delivers low-latency, gigabit links. My IoUT research hybridizes the two, optical and acoustic, so the network can complementarily reap the benefits of both. [4]
Network localization
Localization is paramount for optical underwater networks, for three reasons: the optical link budget depends heavily on pointing and alignment; multi-hop and geographical routing need node positions; and sensory data is only useful if it is tied to a location. On top of harsh channel impairments, limited battery is a second hard constraint, since recharging or replacing a submerged node is a formidable task.
We first studied how the duty cycle of energy-harvesting IoUT nodes affects localization. Rather than the usual shortest-path approach, our method reduces the estimation error of each block kernel matrix and yields a closed-form estimator for every optical node, reaching within a few centimeters of the Cramér–Rao lower bound depending on range and anchor count. [8], [10] A higher energy-arrival rate keeps more nodes active, producing more received-signal-strength (RSS) measurements and, in turn, sharper localization.
A follow-up fused noisy RSS from both optical and acoustic nodes under a weighted multiple-observations scheme that trusts accurate observations more. [10] Because anchor placement matters so much, we then analyzed how anchor-position uncertainty degrades 3D localization (exploiting time- and angle-of-arrival instead of RSS), and found a degradation that simply adding more anchors cannot fix. [3], [13]
Connectivity and routing
Connectivity and localization are interwoven: denser connectivity gives more RSS measurements and better localization, while better location knowledge enables the precise pointing that sustains reliable multi-hop links. Inspired by the conical shape of light beams, I modeled UONs as randomly scaled sector graphs and quantified the probability of connectivity as a function of node density, range, and the optical transmitters’ divergence angle. [11], [12] End-to-end performance of multi-hop decode- and amplify-and-forward links was analyzed under location uncertainty. [5], [7], [14]
To extend range, we proposed a distributed Light Path Routing protocol that leverages the range–beamwidth trade-off and works even without a pointing mechanism or a global view of the network. Building on the lessons of pointing errors, the broadcast nature of light beams motivated a Sector-based Opportunistic Routing (SectOR) protocol: instead of unicasting to one node, it targets a set of candidates to raise the packet-delivery ratio, selecting and prioritizing them by manipulating the rate–error and range–beamwidth trade-offs. SectOR also handles multimodal nodes that split control and data traffic across acoustic and optical channels, and in well-connected networks can outperform even optimal unicast routing. [2], [9]
A software-defined vision
These efforts grew out of a widely cited survey that laid out underwater optical wireless communication, networking, and localization layer by layer. [1], [6] My broader vision uses software-defined networking (SDN) and network function virtualization (NFV) to realize opto-acoustic IoUT: once the interwoven roles of the SDN layers are made explicit, NFV provides application-specific cross-layer protocol suites through a management-and-orchestration framework, turning a heterogeneous tangle of optical and acoustic links into a programmable whole. [4]
