Cell loss recovery using adaptive LDPC-based FEC method for real time services over best effort networks

Real-time services over best-effort networks, such as teleconferencing over unspecified bit rate (UBR) class asynchronous transfer mode (ATM) networks or the Internet, have a very high demand as they are cost effective. However, cell loss caused by buffer overflow at switches is a serious problem in this service class, and is responsible for degradation of the quality of service (QoS). Forward error correction (FEC) is a promising technology for recovering lost cells and improving the QoS of best-effort networks. Conventional FEC methods utilize a FEC matrix with fixed dimensions and are usually designed to cope with the worst case of cell losses. A new method that provides dynamic recovery capacity to deal with the current cell loss ratio through the use of a variable-dimension FEC matrix is proposed. The scheme also takes into account the coding method of the FEC matrix. Numerical results show that the cell loss ratio and transmission efficiency of the proposed method are superior to those of conventional methods.