Adaptive Network Coding With Noisy Packets

In this paper we study the problem of employing network coding in presence of bit corruptions in the packets. We consider a network model within which a minimal amount of processing is employed at intermediate nodes. Under such a paradigm the efficiency of a network coding function, is dependent on the bit-error rates (BER) observed in the packets. Hence under time-varying channel conditions the network coding functions need to adapt in accordance to the channel state associated with the packets. We propose a systematic optimization approach to design an adaptive network coding (ANC) scheme which facilitates efficient exchange of mutually independent information. Under the proposed scheme the intermediate nodes relay the packets with or without piggybacking, depending on the corruption in the received packets and in accordance to pre-designed exchange rules. We illustrate the performance benefits of the proposed optimizations on the basis of channel models derived from actual 802.11 b and 802.15.4 measurements.