The outstanding potential of multicast ABR
One of the most interesting subject over the last 10 years has been hybridation of services. This hybridation is supported by several factors. The most important for us is the transport.
Hybrid transport is often a combination of unicast and multicast (not always but let's stick with this idea for this blog post). This combination raises many challenges such as synchronization at a frame-accuracy level (which owned us a NAB Innovation Award). For OTT application it raises the question of using multicast ABR (mABR) with a unicast fallback signal.
Building an open-source prototype showing a seamless switch between sources is something that we should complete within the 2021S1 semester. However this wouldn't solve two of the key issues: how to set up mABR in a production environment and how to provide a software stack that goes beyond IP and towards DVB-I grade services?
Set up mABR in a production environment is hard. Our CEO Romain Bouqueau gave training in some of the most important Cable operators in the US: most of them couldn't operate the multicast signals with OTT as they previously did for IPTV. Broadcasting a file system is supported by two protocols: FLUTE and its better-suited ROUTE child. Existing professional implementation rely on FLUTE which comes with its set of drawbacks, needing some none-standard extensions which make the solution practically vendor-locked.
As for the extension of our open-source stack to hardware layers, we hoped some people could improve some existing code. However it seems like the solution could come from dedicated entities such as 5G-MAG. 5G-MAG could foster cooperation toward a baseline open-source stack that goes beyond a simple demonstrator. With GPAC Licensing we support this idea of open-source stacks that can be re-used from standardization to actual deployment.
Last, DVB-I grade applications should come for free for two reasons. First our implementation can act as a gateway (we hope to make some improved demo in 2021S2). And DVB seems to push for extending their original work toward a reference DVB-I client.
