With simulcast, a sender encodes and uploads the same video in several quality layers at once — for example a high-resolution layer, a medium one, and a low one — and the SFU (the forwarding media server) then delivers to each receiver the single layer that receiver's bandwidth can actually carry. Instead of forcing one shared quality on everyone, the call adapts per viewer.
For a telemedicine product this is the standard answer to the reality that participants in one consult almost never share the same network. A clinician on hospital fiber, a patient on rural LTE, and an interpreter on home Wi-Fi can join the same call and each receive a stream appropriate to their connection, rather than the whole consult dropping to the quality of the weakest link or stuttering for the strongest. That directly supports the "clinically good enough" quality bar a consult needs to be diagnostically useful.
The cost of simulcast falls entirely on the sender: encoding several layers simultaneously consumes more upload bandwidth and more encoder CPU than sending one stream. On older phones, low-end devices, or constrained home connections, that overhead can itself degrade the experience, which is the common pitfall — turning on aggressive simulcast layers for a population of patients on weak devices and inadvertently making their outbound video worse. The practical implication is to tune the number and resolution of layers to the actual device and network mix of your patients, not to a lab-grade ideal.
