The bio-edge: a survey and research agenda for the Internet of Bio-Nano Things, 2026-2035

The Internet of Bio-Nano Things (IoBNT) extends the Internet of Things into the biochemical domain of living systems through nanoscale bio-engineered devices that sense, actuate, and communicate primarily via molecular signalling. Eleven years after the founding vision of Akyildiz et al. [4], the field has accumulated working architectures, microfluidic testbeds, and mature channel models, but its system-integration challenges - latency, privacy, and energy - are substantially edge-computing challenges: in-body decision loops cannot tolerate cloud round-trip latency, biomolecular data cannot safely stream to remote servers, and harvested-power devices cannot continuously transmit raw high-rate signals. This survey reframes the IoBNT layer stack as a five-layer bio-edge reference architecture in which the bio-cyber interface (BCI, distinct from brain-computer interface) is upgraded from a transduction gateway to a first-class compute layer with its own latency, energy, and trust accounting. We construct a DOI-deduplicated bibliometric snapshot of 311 entries, identify three under-occupied subtopics that constitute the field's strategic white space - TinyML on harvested power, federated learning across edge gateways, and Bio-SDN orchestration - and survey the technical state of each. The centrepiece is a ten-prediction research agenda for 2026-2035 with each prediction stated as a dated metric, a causal mechanism, and a falsifier, designed to give the IoBNT community a structured object that subsequent work can measure itself against.