Optimum Physical-Layer Frame Size for Maximising the Application-Layer Rateless Code´s Effective Throughput

The tolerable packet-loss ratio of an Internet Protocol (IP) based wireless networks varies according to the specific services considered. File transfer for example must be error free but tolerates higher delays, whereas maintaining a low delay is typically more important in interactive Voice Over IP (VOIP) or video services. Classic Forward Error Correction (FEC) may be applied to the data to provide resilience against bit errors. A wireless IP network provides the opportunity for the inclusion of FEC at the physical, transport and application layers. The demarcation between the analogue and digital domain imposed at the Physical layer (PHY) predetermines the nature of the FEC scheme implemented at the various layers. At the PHY individual packets may be offered FEC protection, which increases the likelihood of their error-free insertion into the protocol stack. Higher layers receive packets that are error free and the purpose of a FEC scheme implemented here is to regenerate any missing packets obliterated for example by the Binary Erasure Channel (BEC) of the IP network´s routers. A rateless code may be beneficially employed at a higher Open Systems Interconnection (OSI) layer for replenishing the obliterated packets, but unless the characteristics of the channel are considered, the ultimate rate achieved by such a code may be compromised, as shown in this contribution.