Taking the Trash Back In: Practical Joint Channel and Network Coding for Improving IEEE 802.11 Networks

Although Forward Error Correction (FEC) is key to the efficient operation of wireless communications networks, it can result in the recovery of an erroneous message. Practical systems detect it by using a CRC and discarding the packet. However, we are discarding potentially useful information, while forcing the system to re-transmit the data. Network coded flows can correct these errors by exploiting the network code structure, albeit by incurring in added computational costs, e.g., using PRAC. This paper has two objectives. First, quantify the fraction of packets that the FEC recovers with errors and the rate of bits flipped in each of these packets for different data rates in 802.11g. Our measurements show that for 54 Mbps, 46% of received packets have errors after FEC, but that the amount of bit errors per packet is small. Thus, there is a potential to double the effective data rate when using network coding for error correction. Second, advocate for a new technique for error correction in coded flows that reduces the computational effort by an order of magnitude or more compared to PRAC and more than doubles the number of errors that can be corrected while maintaining an acceptable computation time.