Network Coding with Multi-Generation Mixing: A Generalized Framework for Practical Network Coding

Due to the broadcast nature of wireless networks they have been a natural platform for applying Network Coding (NC). Wireless networks can benefit significantly from NC due to their broadcast nature and the opportunity of enhancing bandwidth utilization. In this paper, we develop Multi-Generation Mixing (MGM), which is a generalized approach for generation based network coding. With traditional generation based NC sender packets are grouped in generations where encoding and decoding are performed on packets that belong to the same generation. In scenarios where losses cause insufficient reception of encoded packets, NC losses occur. NC losses are expensive; the minimum unit of loss is the loss of one generation. The proposed MGM framework allows the encoding among generations for the purpose of enhancing NC decodability. With MGM in scenarios where insufficient number of encodings received of a generation, it is still possible to recover the generation using data encoded in other generations. We develop MGM encoding and decoding approaches, and demonstrate the improvements in performance achieved by MGM. Further, a canonical analytical model for MGM network coding is developed, and, extensive simulations over random wireless networks experiencing random packet losses are presented.