An Efficient Construction Method for Ultra-Reliable Finite Length Analog Fountain Codes

Abstract

Ultra-reliable low latency communication (URLLC) is defined by the the third generation partnership project (3GPP) for the fifth generation (5G) mobile systems. In order to realize the stringent requirements of ultra-high reliability and low latency, the data block length needs to be small. However, most of the state-of-art channel codes are designed and optimized for long block length. Analog fountain code (AFC) is a capacity-approaching self-adaptive code on wireless channels with linear encoding/decoding complexity. In this paper, we study the optimization of AFC in the finite length regime. We first analyze the effect of average degree on decoding performance of AFC in the finite length regime. Then, based on the characteristic of weight coefficient in AFC, we proposed a joint design of weight optimization (WO) progressive edge growth (PEG)-AFC algorithm to improve the decoding performance in the finite length regime, named PEG-AFC algorithm. Simulation results show that the proposed PEG-AFC algorithm can achieve lower bit error rate (BER) in a wide range of signal to noise ratio (SNR).