ANC: Adaptive unsegmented network coding for applicability

Unsegmented network coding (UNC) is a promising technology to overcome poor source information scheduling of segmented network coding (SNC) in large scale networks, where unresponsive feedback slacks the scheduling. However, three unsolved problems limit its applicability. First, UNC employs ACK on witness as the feedback. The frequently triggered witness-ACK will introduce considerable overhead and depress throughput. Second, a new constraint from practical decoding requirement on the slide window has recently been proved. The additional constraint will make UNC much different, which has not been studied. Third, although UNC may outperform SNC in large scale networks, it does not work well in small and moderate-sized networks, exhibiting poor universality. In this paper, we address these problems and propose the Adaptive unsegmented Network Coding (ANC). ANC applies technologies for improvements, which is derived through reinvestigating UNC (solve the second problem), to improve achievable throughput, and save a majority of control overhead (solve the first problem). In addition, ANC incorporates a novel hybrid source packets admission scheme and can well adapt to various network conditions (solve the third problem). Simulation results show that ANC outperforms both SNC and UNC in universal network conditions, and the throughput gain over both can be up to 22%.