A Forward Erasure Correction (FEC) scheme improves resilience but can be overwhelmed by losses. Using long-term reference frames (LTRFs), the prediction structures in a video coding may be reset at a special recovery frame. Consequently, there is a need to adequately protect LTRFs. To address such a need, techniques are presented herein that support a FEC scheme for protecting LTRFs without increasing latency. Aspects of the techniques presented herein treat the LTRFs as a base layer for the purposes of layered FEC coding. Thus, it is possible to think of a normal scalable coding scheme with, for example, two temporal layers as in fact consisting of three layers for FEC purposes. Further aspects of the techniques presented herein support a sliding window or convolutional code which offers the opportunity to reduce delay by using a stateful FEC process where the packet memory is incremented by the latest packet before a new packet is produced. Alternatively, a special strong code may instead be used for the LTRFs which depends upon (1) the previous LRTFs and (2) the most recent Layer 0 frames such that if either set is uncorrupted the LTRF can be recovered.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Davies, Thomas, "SUPER FORWARD ERROR CORRECTION (FEC) FOR LONG-TERM REFERENCE FRAME (LTRF) RECOVERY FRAMES", Technical Disclosure Commons, (October 04, 2021)