Interleaved coding and sequential transmission scheme for packet‐level forward error correction over burst loss channels

Abstract

Burst packet loss is a common problem over wired and wireless networks and leads to a significant reduction in the performance of packet-level forward error correction (FEC) schemes used to recover packet losses during transmission. Traditional FEC interleaving methods adopt the sequential coding-interleaved transmission (SCIT) process to encode the FEC packets sequentially and reorder the packet transmission sequence. Consequently, the burst loss effect can be mitigated at the expense of an increased end-to-end delay. Alternatively, the reversed interleaving scheme, namely, interleaved coding-sequential transmission (ICST), performs FEC coding in an interleaved manner and transmits the packets sequentially based on their generation order in the application. In this study, the analytical FEC model is constructed to evaluate the performance of the SCIT and ICST schemes. From the analysis results, it can be observed that the interleaving delay of ICST FEC is reduced by transmitting the source packets immediately as they arrive from the application. Accordingly, an Enhanced ICST scheme is further proposed to trade the saved interleaving time for a greater interleaving capacity, and the corresponding packet loss rate can be minimized under a given delay constraint. The simulation results show that the Enhanced ICST scheme achieves a lower packet loss rate and a higher peak signal-to-noise-ratio than the traditional SCIT and ICST schemes for video streaming applications.KEYWORDS burst loss channel, forward error correction, interleaved coding, interleaving delay, sequential transmission | INTRODUCTIONThe data delivery quality over the Internet is readily affected by transmission channel dynamics, such as delays and packet loss. Thus, data recovery schemes such as forward error correction (FEC) are essential in achieving a robust data transmission performance. 1,2 The basic principle of FEC is to group the source data items into blocks of a predetermined size at the sender end, and then add a certain number of redundant data items to each block such that if losses occur during transmission, the source data items can still be reconstructed at the receiving end. Typically, the number of redundant data items added to the source block determines the data recovery capacity (ie, the number of lost source data items that can be recovered). In general, FEC schemes operate at either the byte (ie, symbol) level or the packet level. In byte-level FEC, the source bits and FEC redundant bits are packetized into transport packets. 3-5 By contrast, in packet-level FEC, the source packets and FEC packets are constructed separately so as to provide strong protection against packet losses. 6-8 In other words, byte-level FEC and packet-level FEC are designed mainly to recover transmission errors within the packet and occasional packet losses, respectively. In the event of burst packet losses, however,