Design of Optimized Sliding-Window BATS Codes

Batched sparse (BATS) codes are a low-complexity random linear network coding scheme that can achieve asymptotic bandwidth optimality. The destination node of BATS codes starts to decode the information packets only after receiving a sufficient number of coded packets which induces delay. Furthermore, the larger the size of source block is, the longer the delay is. In this letter, we propose a scheme by using sliding windows applied over BATS codes, named sliding-window BATS codes, which divides the information data into smaller blocks so as to reduce delay. Moreover, for improving the performance, we optimize the degree distributions with linear programming, and its performance is verified by computer simulations.